Analyzing and Predicting Ventilation Coefficient over India using Long-term Reanalysis Datasets and Hybrid Machine Learning Approach 

Analyzing and Predicting Ventilation Coefficient over India using Long-term Reanalysis Datasets and Hybrid Machine Learning Approach | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Analyzing and Predicting Ventilation Coefficient over India using Long-term Reanalysis Datasets and Hybrid Machine Learning Approach Amitabha Govande, Raju Attada, Krishna Kumar Shukla, Soumya Muralidharan, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4551619/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 29 Oct, 2025 Read the published version in Theoretical and Applied Climatology → Version 1 posted 18 You are reading this latest preprint version Abstract The concentrations of atmospheric pollutants are a serious concern due to their adverse impacts on human health. The ventilation coefficient (VC) is an indicator that measures the dispersion capacity of air pollutants (air pollution potential) in the atmosphere, providing insights into air quality. In this study, we aim to investigate the spatio-temporal variation and trends of VC over the Indian subcontinent using India’s first high-resolution regional reanalysis (IMDAA) and global reanalysis datasets (ERA5) for the period 1980-2019. The spatial pattern of the seasonal climatological mean ERA5 and IMDAA derived VC shows a lower magnitude during winter and post-monsoon seasons, indicating poor air quality over the Indian region, especially in the northern parts of India. We noticed a gradual declination of VC during different seasons, implying increasing surface-level air pollutants and worsening air quality over India. The study further investigates the changes of VC during strong phases of El Niño and La Niña events. The results reveal that El Niño significantly impacts air quality over northern and western parts of India during pre-monsoon and monsoon seasons. At the diurnal scale, the VC exhibits the highest magnitude and variability during daytime due to increased dispersion of pollutants and higher human activities, while remaining low and stable during night due to stagnant atmospheric conditions. These essential characteristics of VC are well represented in IMDAA, albeit with some discrepancies. Furthermore, we have examined the fidelity of a machine learning model-Convolutional Neural Network and Long Short-Term Memory (CNN-LSTM), in predicting the VC for the year 2019 over Delhi city. Various statistical metrics are computed to evaluate the performance of the CNN-LSTM model. The results confirm that the model successfully predicts the VC compared to observations from ERA5. Air quality Ventilation Coefficient Variability Trends CNN-LSTM model Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 Figure 13 Highlights The spatial pattern of seasonal climatological mean VC during winter and post-monsoon seasons indicates poor air quality in India. VC exhibits substantial diurnal variability. ENSO has been found to have an effect on air pollution dispersion capacity. CNN-LSTM machine learning model successfully predicts air pollution potential over India. 1. Introduction Air pollution is caused due to various human activities such as waste generation, transportation, construction, coal burning, and resource exploitation, which are intensified by population growth and economic development (e.g. Saha et al., 2019 ; Hou et al., 2018 ; Karuna et al., 2017 ; Iyer and Raj, 2012 ). The accumulation and dispersion of pollutants depend on regional meteorological conditions. Additionally, the faster pace of urbanization of metropolitan cities of India affects the deterioration of Air Quality (AQ) rapidly, which has a detrimental effect on the environment and human health (Chowdhary et al., 2018 ; Apte et al., 2015 ). Altogether, the high-level exposure to air pollutants causes morbidity and mortality in any region. Air pollutant sources are generally classified into natural and anthropogenic. Natural sources release air pollutants directly into the atmosphere, including volcanoes, biogenic emissions, dust storms, and forest fires, emitting substances such as Particulate Matter (PM), SO 2 , CO, NOx, etc. On the other hand, anthropogenic sources release harmful particles through activities such as vehicular emissions, combustion processes, burning fossil fuels, waste management, and construction. Major air pollutants from these sources include CO, VOCs, PM, CO, and NOx. Primary pollutants in the atmosphere are directly emitted from the sources as discussed above, while secondary pollutants are formed through photochemical reactions (when primary pollutants react with other compounds) in the presence of solar energy (Castelli et al., 2020 ; Al-Salem and Khan, 2010 ). Distinguishing between natural and anthropogenic sources and their contributions is vital for effective air quality management and pollution control efforts. The air pollutants loading shows the regional and seasonal heterogeneity due to changing background meteorology and topography over the Indian region (Babu et al., 2013 ; Ratnam et al., 2021 ). The dominance of different pollutant sources varies during different seasons in India which has been reported previously (Dey et al., 2020 ; Shukla et al., 2022 ). The Indo-Gangetic Plain (IGP) is highly populated and is also considered as a hotspot for elevated aerosol loading. The air pollutant loading across IGP is higher throughout the year than in the rest of the country. The pollutant loading is maximum during the winter season due to bio-fuel burning for heating and cooking, crop-residue burning and calm wind conditions over IGP and north India favours the accumulation of these aerosols (Chimurkar et al., 2020 ). The west India region is dominated by dust aerosols during pre-monsoon season due to the frequent occurrence of dust storms and long-range transport of dust from arid regions (Shukla et al., 2022 ), while it also showed high PM loading during the post-monsoon season due to stubble burning (Jethva et al., 2019 ). Central India receives more pollutants during winter and post-monsoon seasons due to biomass and crop-residue burning, whereas low concentrations during monsoon due to wet removal of aerosols (Maheshwarkar and Sunder Raman, 2021 ). In addition to the above, the coal burning in thermal power plants and industries is the main contributor of pollutants in eastern and peninsular India (Li et al., 2017 ; Venkatraman et al., 2018). Anthropogenic aerosols (such as carbonaceous aerosols and sulphate) are dominant over eastern parts of India which is mainly due to biomass burning and emission from the thermal power plants located in neighbouring states (such as Bihar, Uttar Pradesh and Odisha) (Kalita et al., 2020 ; Rawat et al., 2019). The source variability of pollutants and seasonal heterogeneity highlight the complexity of air quality management in India, necessitating region-specific strategies to effectively address air pollution challenges. The Ventilation Coefficient (VC) is an essential and well-recognized factor that determines the air pollution dispersion potential and AQ over a region. VC depends on Planetary Boundary Layer Height (PBLH) and average wind speed over any region. A higher VC indicates good AQ with high dispersive capacity, while a lower VC indicates poor ventilation (or AQ). Several studies have discussed the variation of VC across the Indian sub-continent (Kannemadugu et al., 2021 ; Kumar, 2019 ; Saha et al., 2019 ; Karuna et al., 2017 ; Sujatha et al., 2016 ) as well as global scale (Abiye et al., 2016 ; Chan et al., 2012 ; Holzworth, 1967 ). For example, Holzworth ( 1967 ) reported that Los Angeles had low VC (poor air quality) among the seven locations studied in the United States. Abiye et al. ( 2016 ) reported the highest (lowest) VC during day (night) time over an industrial site in Nigeria. Previous studies also demonstrate the influence of VC on particulate matter (PM 2.5 and PM 10 ) mass concentration at various point locations in China across the seasons (Hou et al., 2018 ; Chan et al., 2012 ). As far as the Indian context, the AQ deteriorates over time due to urbanization and other economic developments that impact various sectors such as human health, aviation, soil characteristics, and water quality. This underpins the importance of studying the air quality dispersion potential using VC over different sub-regions of India. Hence, some researchers have attempted to investigate the VC and its linkages with AQ over the Indian region which is well-summarized in Table 1 . Recently, Kannemadugu et al. ( 2021 ) demonstrated spatial variability (2014–2019) of VC across the Indian subcontinent, with high VC observed on the east coast of Andhra Pradesh and Tamil Nadu during winter and in western India and New Delhi during the summer monsoon. Iyer and Raj, (2013) showed a decreasing VC trend over some Indian metropolitan cities, leading to poor AQ during 1971–2000. Saha et al. ( 2019 ) reported a rising trend in VC (70 m 2 /s/year) over the capital of India but for a short period from 2006 to 2014. The changes/variability in VC is a result of various regional factors. For example, the local conditions, including topography, and other meteorological parameters, have a significant impact on both the Planetary Boundary Layer Height (PBLH) and wind speed. As a result, the VC is a multifaceted phenomenon that is shaped by various factors. Budakoti and Singh, ( 2021 ) reported a negative correlation between PM 2.5 and Planetary Boundary Layer Height (PBLH) over the Indian region. On the other hand, Sujatha et al. ( 2016 ) found that VC has a strong negative correlation with Black Carbon (BC) in the city of Hyderabad. Table 1 Literature review on ventilation coefficient studies over Indian subcontinent. Region, point location over India Time period Month, seasons, annual Method of calculation VC using observation and reanalysis Main conclusions Reference Mumbai, Delhi, Kolkata, Chennai 1971–2000, Winter months Daily radiosonde data, obtained from IMD Pune Decreasing trend in all the four regions. During Dec and Feb, VC decreased by 49 and 32 m 2 /s/year over Delhi, 15 m 2 /s/year over Mumbai and 14 and 17 m 2 /s/year Iyer and Raj, ( 2012 ) Delhi 2005–2014 PBLH -SODAR data over Delhi region - obtained by CPCB Increasing trend over Delhi with 70 m 2 /s/year VC: monsoon > pre-monsoon > winter > post-monsoon. ARIMA model fit reasonably well with the data series with some discrepancies in VC during monsoon and pre-monsoon season. Saha et al., ( 2019 ) New Delhi, Thar, Jaisalmer, Jodhpur, Patna, Kolkata, Nagpur, Rourkela, Mumbai, Hyderabad, Bangalore, Chennai 2015–2019 PBLH - CrIS SOUMINPP satellite, wind speed - ERA5 Low pollution potential (< 6000 m 2 /s) at east coast of Andhra Pradesh and TN during winter and western India and New Delhi during summer monsoon. During pre-monsoon, western Gujarat, south-west Rajasthan, and parts of IGP shows low pollution potential. Low pollution potential is observed at east coast of Tamil Naidu during post-monsoon. Kannemadugu et al., ( 2021 ) IVRI institute, Bareilly city, UP 2013–2017 Hourly meteorological data from wunderground.com High VC during day and Low VC at night and early morning. During winter max VC was found at 6000 m 2 /s and during summer, maximum VC was 15000 m 2 /s. Karuna et al., ( 2017 ) VBIT, Hyderabad (17.4° N − 78.5° E) April 2012 - Feb 2013, Pre-monsoon (April-May), monsoon (July-August), post-monsoon (Oct-Nov), winter (Dec-Jan) High resolution GPS radiosonde flights carried out with iMet (USA) radiosonde VC: monsoon > post-monsoon > pre-monsoon > winter High wind speed during monsoon is causing high VC. Diurnal pattern of VC shows high VC during the day or afternoon and low VC during night and early morning Kumar, ( 2019 ) CSIR - National Physics Laboratory - Delhi Mar 2019 - Feb 2020 SODAR VC high and variable during day/afternoon and low and stable during night and early morning. High VC of 30800 m 2 /s in March over Delhi Priyanka et al., ( 2022 ) Additionally, global climate drivers like El Niño and La Niña also play an important role in modulating regional circulation patterns and associated meteorological conditions. In a recent study, Wang et al. ( 2022 ) examined the influence of El Niño–Southern Oscillation (ENSO) on air quality and reported that in China good air quality is associated with El Niño events because of increased precipitation, while poor air quality is linked to La Niña because of reduced precipitation. Therefore, the interannual variability of the VC often links to the ENSO phases. In India, El Niño is usually associated with decreased precipitation and La Niña, on the other hand, induces more precipitation during summer monsoon months and this can contribute a lot to the AQ (Gao et al., 2019 ). These large-scale weather phenomena affect not only precipitation but also other factors like relative humidity and particulate matter and are likely to impact VC as well. The impact of ENSO on AQ and PM 2.5 concentrations has been the subject of a few studies (Xie et al., 2022 ; Wang et al., 2022 ). However, understanding the role of El Niño and La Niña events on VC over India remains unexplored. Prediction of VC is also extremely important for mitigating the adverse effects of air pollution in highly populated countries like India. Towards this end, it is imperative to examine the predictive models in capturing the observed VC in any given region. Artificial Intelligence and Machine Learning (AI/ML) are emerging as powerful tools to predict future weather/climate variables. It is important to emphasize that AQ prediction is quite complicated and challenging due to its quick variability at a shorter time scale over any given region (e.g., Castelli et al., 2020 ). AQ is a nonlinear phenomenon that is associated with numerous complexities and various meteorological factors (e.g. Castelli et al., 2020 ). Some recent studies have highlighted the importance of AI/ML models in predicting regional weather and climate (Zhang and Li, 2022 ; Castelli et al., 2020 ; Kumar et al., 2015 ). Traditional time series forecasting models, for example, Auto-Regresive (AR), Moving-Average (MA), Auto-Regressive Integrated Moving Average (ARIMA), etc, and machine learning-based (e.g. Multi-Layer Perceptron, Convolutional Neural Network, etc) predictions are among the currently used models for time series prediction (Zhang and Li, 2022 ). The former one has been used for the prediction of VC over Delhi by Saha et al. ( 2019 ), while the latter has been less explored for VC. Statistical models like time series forecasting are mostly based on linear correlations and therefore may not be able to comprehend the nonlinear characteristics effectively (Gibson et al., 2021 ; Kumar et al., 2015 ). However, Machine learning models can be used to solve this problem. Kumar et al., ( 2015 ) has shown the successful application of Wavelet-Neuro-Fuzzy model in predicting VC derived from Sonic Detection and Ranging (SODAR) over Delhi. Masood and Ahmad., (2020) tried to predict in situ PM2.5 levels from Central Pollution Control Board (CPCB) over Delhi using Support Vector Machine (SVM) and Artificial Neural Network (ANN). Algorithms involving neural network like ANN, Recurrent Neural Network (RNN), Convolutional Neural Network (CNN), have been successful in forecasting air quality (Masood and Ahmad, 2020 ; Huang and Kuo, 2018 ; Xiao et al., 2020 ). Although, whenever dealing with extensive datasets sometimes features of only one algorithm are not enough to get the best possible predictions. Using a combination of two different models is a good option. Previous studies have reported that better results were observed when such hybrid models were used (Ayutran et al., 2020; Huang and Kuo, 2018 ; Wang et al., 2022 ; Zhang and Li, 2022 ). One such model is CNN coupled with Long Short-Term Memory (CNN-LSTM), which has previously shown higher predictive ability over a few cities in China (Huang and Kuo, 2018 ; Zhang and Li, 2022 ). CNN is well-known for extracting patterns or features even with the most complex nonlinear data (Zhang and Li, 2022 ; Huang and Kuo, 2018 ). On the other hand, LSTM is known for solving long-term time dependency. Each node in the LSTM network acts as a memory cell, so it remembers every piece of information throughout time (Zhang and Li, 2022 ; Huang and Kuo, 2018 ). Importantly, every node in the LSTM network propagates the output backward, learns the error and continues moving forward for the correct/better predictions (e.g. Huang and Kuo, 2018 ). LSTM, when used with CNN, allows effective extraction of features and improves the accuracy of predictions. The studies on VC over India are primarily focused on a point location, short period or different seasons based on the availability of datasets. However, the long-term variations including the impact of large-scale climate modes on VC are still limited over the Indian subcontinent (Table S1 presents a comparison of our study with some previous works). Moreover, the point location observation of PBLH over India is available whereas the spatial homogeneous in-situ PBLH observations are still lacking. Given the above, the global and regional reanalysis datasets fill the gap in spatial and temporal coverage to understand the VC patterns. It provides high spatial and temporal resolution long-term data of PBLH and meteorological parameters. Thus, in this work, we first aim to investigate the long-term variability (seasonal, interannual and diurnal time scales), trends, and impact of El Niño events on VC over India using reanalysis datasets covering from 1980 to 2019. We have examined the VC trends for four different seasons defined by the India Meteorological Department (winter - DJF, pre-monsoon - MAM, monsoon - JJAS, and post-monsoon - ON) using ERA5 and IMDAA data. We further employed a CNN-LSTM model to construct a univariate time series model for predicting VC at seasonal and annual scales. This manuscript is organized as follows: section 2 describes the data and methods, section 3 presents results and discussions, section 4 elucidates the machine learning predictions and Section 5 summarizes the key findings of the study. 2. Data and Methodology 2.1 Data used We employed two high-resolution reanalysis datasets (global and regional) to understand the VC variability. The fifth-generation European Center for Medium-Range Weather Forecasts (ECMWF) reanalysis (ERA5, global) is the latest climate reanalysis with high spatial and temporal resolution available hourly at a spatial grid of 0.25 o × 0.25 o on 37 pressure levels from 1979 to the present ( https://cds.climate.copernicus.eu ; Hersbach et al., 2020 ). Additionally, Indian Monsoon Assimilation and Analysis (IMDAA, regional) provides long-term Indian regional reanalysis data for a wide range of atmospheric variables, available from 1979 to the present at a spatial grid of 12 km × 12 km (Rani et al., 2020). This is India’s first high-resolution dataset that enhances the benefits of regional representation by accounting for better surface forcing, including orography, small-scale processes and features. It further improves the representation of the interaction between small- and large-scale processes. Therefore, the VC from the high-resolution IMDAA database provides an opportunity to study the long-term Air Pollution Climatology over India at sub-regional scales as it exhibits a regional heterogeneity over the Indian landmass. We obtained hourly PBLH, zonal (u) and meridional (v) wind datasets from IMDAA and ERA5 for 1980–2019. ERA5 have previously been validated for PBLH and wind speed at a global scale (Guo et al., 2021 ), and used in studies related to VC (Kannemadugu et al., 2023 ; Kannemadugu et al., 2021 ). And since it is a global reanalysis, our assumption is to consider ERA5 as the standard value and compare the performance of IMDAA based on it. This validation is important for the development of new generation reanalysis datasets for air quality studies. Further, we also checked the long-term seasonal PM2.5 trends, for this we used Modern-Era Retrospective analysis for Research and Applications – Version 2 (MERRA-2) reanalysis (e.g. Gelaro et al., 2017 ) available at 0.5° × 0.625° spatial resolution from 1980 to 2019. 2.2 Methodology 2.2.1 Estimation of ventilation coefficient VC is a product of PBLH and average wind speed through the mixing layer. PBLH represents the vertical mixing of pollutants (Budakoti and Singh, 2021 ; Allabakash and Lim, 2020; Sujatha et al., 2016 ) while wind speed represents the horizontal mixing of pollutants (Chan et al., 2012 ). The Ventilation coefficient is given as: VC = [PBLH (m) ⋅ average wind speed (m/s)] --------- (1) A Low (high) ventilation coefficient means less (more) dispersion potential of pollutants. The IMDAA data was re-gridded to 0.25° ⋅ 0.25° using bilinear interpolation to compare the ERA5 results. The linear regression model was used to determine the best-fit line, slope, and p-value (to check statistically significant values at the confidence interval of 95% or p-value of 0.05). As shown in Fig. 1 , the Indian region was further divided into five sub-regions to examine temporal variability of VC across the seasons as suggested by (Ramachandran et al., 2012 ; Nizar and Dodamani, 2019 ). Furthermore, we also extracted metropolitan cities lie in respective sub-regions of India to understand the diurnal patterns of VC. 2.2.2 Estimation of PM2.5 derived from MERRA-2 We estimated the PM 2.5 concentration (Eq. 2) using five major air pollutants (dust, black carbon (BC), organic carbon (OC), sulfate and sea salt in µg/m 3 ) which are contributing to the PM 2.5 concentration. All of these components were derived from MERRA-2 at monthly averaged scales. The data was re-gridded to ERA5 resolution (0.25° ⋅ 0.25°) using bilinear interpolation. More details regarding PM 2.5 mass concentration calculation can be found in Bali et al. ( 2021 ) and Chimurkar et al. ( 2020 ). PM 2.5 = [DUST 2.5 ] + [SS 2.5 ] + [BC] + 1.8 ⋅ [OC] + 1.375 ⋅ [SO 4 ] --------- (2) In Eq. 2, DUST 2.5 and Sea Salt (SS 2.5 ) are particulate matter with a size less than 2.5 micrometers and BC is Black Carbon. To estimate the organic matter, OC is multiplied by a factor ranging from 1.2–2.6 depending on space and time. This factor tells the contributions from other elements that are associated with organic matter (Bali et al., 2021 ). For the Indian region, the factor used is 1.8 (Chimurkar et al., 2020 ). The sulfate ion is mainly present in the form of ammonium sulfate in the atmosphere, to calculate the sulfate concentration, it is multiplied by the mass of the sulfate ion (1.375) (Bali et al., 2021 ). The present study estimated PM2.5 concentrations using the methodology outlined by Chimurkar et al. ( 2020 ). This study also highlighted the limitations associated with this method in calculating PM2.5 concentrations from MERRA-2 data. Daily mean PM2.5 concentrations derived from MERRA-2 typically underestimate actual PM2.5 levels and exhibit a lower correlation, and unable to accurately capture PM2.5 concentrations during days of elevated pollution levels. Conversely, monthly mean PM2.5 (MERRA-2) concentrations demonstrate a high correlation with in-situ observations of PM2.5. Therefore, we used monthly data in our analysis. Furthermore, Bali et al ( 2021 ) also validated hourly MERRA-2 PM2.5 across multiple sites over the country and found high correlation with the in-situ PM2.5 data from CPCB. We recommend utilizing the MERRA-2 data for analyzing a wide spatial coverage, but the data can give some discrepancies due to coarser resolution for city-scale analysis. 2.2.3 CNN-LSTM framework In this study, we employed CNN-LSTM deep learning model for the prediction of VC at different lead times. For this, we used VC daily data obtained from ERA5 from 1980 to 2019 on a grid of the Delhi (extracted over 77° E to 77.25° E and 28.5° N to 28.75° N grids). It is worth mentioning that when it comes to air pollution, Delhi is one of the metropolitan Indian cities which usually suffers from poor air quality for most of the months in a year (e.g. Jena et al., 2021 ; Kumar and Goyal, 2011 ). Total, 14610 data points have been retrieved over the span of 1980 to 2019. Then, our model was trained using 97.5% of the data from 1980 to 2018 period, and our testing data were 365 days (2.5% of the total data) for the year 2019. Due to the right-skewed nature of the data, log transformation was first performed (on both training and testing set) and the data appeared to be fairly symmetrical after this procedure. The transformed data was then processed in such a way that each value was predicted based on the 15 data points that came before it. Therefore, to predict the final value, we use the previous 15 values as our variables. After running the model, the final predictions were subjected to exponential transformation to retract the values of VC over Delhi. A similar kind of chronology was adopted for seasonal forecasts as well, for the same region. Here, for every season (winter, pre-monsoon, monsoon and post-monsoon), daily data from 1980 to 2018 (97.5% of the data) were taken for training and data for the year 2019 (2.5% of the data) was used for testing. For winter and post-monsoon, the previous three values were considered for prediction and in the case of pre-monsoon and monsoon previous four values were taken to predict the final values, just because the model performed best in these conditions. Detailed Schematic diagram (Fig. 2 ) depicts the data processing part and final structure of the CNN-LSTM model: the Conv1D layer, the Maxpooling1D layer, the LSTM layer, and a dense layer before the output. Conv1D layer identifies and extracts the feature of the time series, Maxpooling1D obtains the hidden information and reduces the dimensionality of the data and LSTM extracts the time dependence in the data. So, CNN encodes the features of the time series data, and LSTM decodes these features to determine the connection between data timing and nonlinearity (Zhang and Li, 2022 ; Huang and Kuo, 2018 ). Since it is a univariate time series model, the total number of features was set to 1. The parameters settings and output shapes in the model can be found in Table S2 and Table S3 from supplementary material. Percentage Mean Absolute Error (MAE), percentage root mean squared error (RMSE) with respect to the observed mean, mean absolute percentage error (MAPE) and coefficient of determination (R 2 ) were calculated to evaluate the model performance. It is to be noted that the lag was set to one timestep (i.e. one day) in our prediction experiments, as the CNN-LSTM model typically exhibits some degree of lag as suggested by Zhang and Li, ( 2022 ). 3. Results and Discussion 3.1 Climatology and Annual Cycles of VC The spatial climatological (1980–2019) patterns of seasonal mean VC are presented in Fig. 3 (a-h). VC has been classified into six categories based on the air pollution potential index over India (e.g. Kannemadugu et al., 2021 ; Kannemadugu et al., 2023 ). Very high pollution potential occurs when VC is between 0–2000 m 2 /s, high pollution potential occurs when VC is between 2000–4000 m 2 /s and medium pollution potential occurs when VC is between 4000–6000 m 2 /s. Low pollution might be considered when VC exceeds 6000 m 2 /s, which is not usually found over the Indian subcontinent, except few locations. During winter (Fig. 3 a, 3 e), extremely low VC (< 1000 m 2 /s) values are noticeable in most of the Indian regions, indicating very high pollution potential occurs during this season. This might be because of low (decreasing) PBLH and weak winds throughout this season, which are responsible for the air stagnation conditions resulting in poor air quality (Kannemadugu et al., 2021 ; Iyer and Raj, 2012 ). Factors such as low PBLH, less solar insolation and cold temperature might also cause the trapping of air pollutants (e.g. Ramachandran et al., 2012 ). Further, low temperature during winter leads to denser air, because of which the vertical mixing is less, giving rise to lower PBLH and therefore poor dispersion of pollutants. Some regions of south India especially the coast of Tamil Nadu show a slightly high VC. These patterns are well depicted by IMDAA as compared to ERA5. During pre-monsoon (Fig. 3 b, 3 f), better dispersive capacity is observed over western parts of the country, indicating the suitability of setting up of new industries in west India during this season. These changes from winter to pre-monsoon could be attributed to elevated temperature that helps in deepening the PBLH (e.g. Ramachandran et al., 2012 ). The pre-monsoon conditions like high solar insolation give rise to high temperatures causing vertical mixing that leads to an increase in the PBLH resulting in the dispersion of the pollutants thereby good air quality. Furthermore, land surface response such as dry soil conditions and high surface sensible heat flux could cause an increase in the PBLH (e.g. Budakoti and Singh, 2021 ) and this can result in increased VC. During monsoon (Fig. 3 c, 3 g), western and southern India exhibits high to medium pollution potential ranges, these regions act as ventilation corridors during this season whereas other parts of the country show poor VC values, indicating the possibility of poor air quality. The PBLH is low during this season for most parts of the country because of high cloud cover, precipitation and strong wind speed (Budakoti and Singh, 2021 ; Kannemadugu et al., 2021 ; Nizar and Dodamani, 2019 ). However, the PBLH is high in west India and it increases gradually from north to south India, this may be because of the transfer of heat from the Bay of Bengal to northwest India, travelling over the monsoon winds that cause an increase in PBLH (e.g. Ramachandran et al., 2012 ). Strong convection during the season may also give rise to a better ventilation of air. IMDAA (Fig. 3 c) slightly overestimated these VC patterns, especially over the southern parts of India. During post-monsoon (Fig. 3 d, 3 h), most of the country exhibits lower values of VC (below 2000 m 2 /s) suggesting lower ventilation prevails, as in winter. This may be attributed to low wind speed during this season over the Indian region. Less insolation of the earth’s surface may also be responsible for low PBLH during post-monsoon, and this forms an inversion layer which opposes the vertical mixing of air pollutants (e.g. Ramachandran et al., 2012 ). It is to be noted that VC over the Andhra Pradesh and Tamil Nadu appear to be slightly affected by northeast monsoon and increased wind speed due to storm and cyclonic activities. IMDAA (Fig. 3 h) realistically represents the spatial patterns of the VC during the post monsoon. The spatial patterns of VC during the post-monsoon season are almost consistent with Kannemadugu et al. ( 2021 ) and Kannemadugu et al. ( 2023 ). The spatial climatological distribution of the daily standard deviation of VC can be seen in Figure. 4 (a-h), a high variability in VC is observed during pre-monsoon and monsoon, especially over the Indo-Gangetic Plains (IGP), west and south Indian region. During winter, all parts of the country report relatively lower variability (Fig. 4 a, 4 e). Some high variability pockets are noticeable in the eastern parts of Ladakh (Fig. 4 a, 4 e). During pre-monsoon, the higher standard deviation was noticed over western India and central India with IGP, northwest Rajasthan and eastern Ladakh showing the highest variability (Fig. 4 b, 4 f) with slightly higher VC observed by IMDAA (Fig. 4 b) than ERA5 (Fig. 4 f). During monsoon, central, western and southern India shows very high variability, with IGP, Rajasthan and some parts of southeast India showing the highest variability (Fig. 4 c, 4 g). On the other hand, low variability was observed at all parts of the country (Fig. 4 d, 4 h), except for eastern Ladakh for IMDAA during the post-monsoon season (Fig. 4 d). Among IMDAA and ERA5, much higher values of VC and standard deviation of VC were observed in IMDAA than in ERA5 but the pattern looks almost similar in both, based on Figs. 3 and 4 . So, from the analysis, it is clear that IMDAA can resolve some of the features of ERA5. The monthly mean VC climatology for five sub-regions of India during 1980–2019 is illustrated in Fig. 5 (a-e). The maximum VC values, with higher variability, are noticeable during the pre-monsoon and monsoon seasons whereas low VC (relatively lower variability) during the post-monsoon and winter seasons. The highest VC was observed in west India (Fig. 5 c) followed by south (Fig. 5 e), central (Fig. 5 b), east (Fig. 5 d) and north India (Fig. 5 a). In north India, the highest VC was observed in May (ERA5) and June (IMDAA) and the lowest VC was observed in January (ERA5) and November (IMDAA) (Fig. 5 a). In central India, the highest VC was observed in May and the lowest VC was observed in October (ERA5) and December (IMDAA) (Fig. 5 b). In west India, the highest VC was observed in June and the lowest VC was observed in October (Fig. 5 c). In east India, the highest VC was observed in April (ERA5) and May (IMDAA) and the lowest VC was observed in October (ERA5) and September (IMDAA) (Fig. 5 d). In south India, the highest VC was observed in July and the lowest VC was observed in October (Fig. 5 e). The exact VC values for each case have been given in Table 2 . The same analysis was done over city scale, where a metropolitan city was picked from each of the five sub-regions of India (North India – Chandigarh, Central India – Nagpur, West India – Jodhpur, East India – Kolkata and South India - Bangalore). Figure 5 (f-j) depicts the annual cycle at city scale, it can be seen that the trend in each city follows the same trend as their respective regions, only the magnitude of VC is higher. The maximum and minimum VC values for each case have been given in Table 3 . Overall, IMDAA follows the same annual cycles as ERA5, only the magnitude of VC is higher in IMDAA and, in some instances, the peak is slightly shifted compared to ERA5. Table 2 Maximum and Minimum values of VC (m 2 /s) for the annual cycle of each Indian sub-region for the period 1980–2019. Data source Vs Sub-regions ERA5 Maximum VC in m 2 /s (month) ERA5 Minimum VC in m 2 /s (month) IMDAA Maximum VC in m 2 /s (month) IMDAA Minimum VC in m 2 /s (month) NORTH 336.08 ± 70.89 (May) 158.40 ± 26.92 (Jan) 553.52 ± 284.52 (June) 118.56 ± 50.32 (Nov) CENTRAL 1729.30 ± 388.72 (May) 347.27 ± 84.76 (Oct) 2379.63 ± 763.37 (May) 276.49 ± 57.75 (Dec) WEST 3972.82 ± 803.29 (June) 637.88 ± 176.53 (Oct) 4893.97 ± 834.27 (June) 444.39 ± 242.80 (Oct) EAST 619.33 ± 108.65 (April) 172.83 ± 45.87 (Oct) 783.22 ± 150.12 (May) 135.17 ± 77.68 (Sep) SOUTH 2950.36 ± 536.03 (July) 371.61 ± 96.38 (Oct) 3607.12 ± 672.18 (July) 379.33 ± 159.17 (Oct) Table 3 Same as Table 2 but for city scales for the period 1980–2019. Data source Vs Cities ERA5 ERA5 IMDAA IMDAA Chandigarh 1042.57 ± 445.69 (April) 187.84 ± 99.20 (Aug) 2209.94 ± 839.05 (April) 547.08 ± 125.03 (Dec) Nagpur 2463.78 ± 979.37 (May) 372.93 ± 92.32 (Jan) 3985.45 ± 947.22 (June) 507.43 ± 202.71 (Oct) Jodhpur 4683.14 ± 1063.41 (June) 460.42 ± 284.27 (Feb) 2209.44 ± 449.21 (July) 546.13 ± 208.31 (Oct) Kolkata 2404.52 ± 584.69 (April) 319.98 ± 174.45 (Oct) 3962.14 ± 661.50 (April) 346.86 ± 231.50 (Oct) Bangalore 3584.88 ± 565.97 (July) 378.76 ± 193.28 (Oct) 4862.94 ± 1053.11 (July) 655.12 ± 404.15 (Oct) 3.2 Diurnal Variability The anthropogenic emissions of various aerosols are increasing day by day over India due to mass movement of people from rural to urban areas, degrading the AQ in urban areas (e.g. Abiye et al., 2016 ). Therefore, it is important to study the diurnal pattern of VC to understand the air quality patterns throughout the day. Figure 6 (a-e) depicts diurnal variations in VC, showing high and variable levels during the day (10:00 h to 18:00 h) and low, stable levels at night and in the morning (19:00 h to 09:00 h). VC is negatively correlated with air pollutants (e.g., particulate matter and black carbon), indicating an inverse relationship (Budakoti and Singh, 2021 ; Chan et al., 2012 ; Sujatha et al., 2016 ). Daytime emissions, driven by anthropogenic activities, are dispersed due to high solar radiation, PBLH, warm temperatures, and wind speed (Karuna et al., 2017 ; Kumar, 2019 ; Moreira et al., 2020 ; Priyanka et al., 2022 ). Conversely, at night and early morning, lower wind speed, reduced PBLH, and colder temperatures contribute to pollutant accumulation in the planetary boundary layer, resulting in elevated pollution levels (Kumar, 2019 ). This suggests that emissions, dispersed during the day, settle at night due to a stable atmosphere, leading to increased pollution levels during this period. VC begins to rise from 09:00 h until 15:00 h (local time), after which it begins to decline (Fig. 6 ). The highest VC was recorded in west India (Fig. 6 c) (IMDAA – 7706 ± 4900 m²/s, ERA5–5214 ± 3459 m²/s at 16:00 h), while the lowest VC was observed in north India (IMDAA – 151 ± 273 m²/s, ERA5–29 ± 37 m²/s at 05:00 h). The diurnal variation of VC shows the maximum (daytime) and minimum (nighttime) magnitudes, indicating greater pollutant dispersion during the daytime than at nighttime. The diurnal pattern of VC in north India (Fig. 6 a) matches with the same study conducted over the city of Bareilly in Uttar Pradesh by Karuna et al., (2019). These diurnal variations in IMDAA agree well with ERA5 in all sub-regions of the Indian sub-continent. Figure 6 (f-j) shows the diurnal patterns of VC from a metropolitan city from each sub-region. It can be seen that the diurnal pattern at the city scale (Fig. 6 f- 6 j) matches with the patterns of VC from their respective sub-regions (Fig. 6 a- 6 e). VC is high and variable during the day while low and invariable at night and morning. Overall IMDAA follows the same pattern of VC like ERA5 only the magnitude is higher, except for the Jodhpur (Fig. 6 h). For IMDAA, the highest VC was observed at Chandigarh (Fig. 6 a) with 8461 ± 7216 m 2 /s at 15:00h and the lowest VC was observed at Jodhpur (Fig. 6 h) with 743 ± 976 m 2 /s at 06:00h. For ERA5, the highest VC was observed at Jodhpur (Fig. 6 h) with 6882 ± 4996 m 2 /s at 16:00h and the lowest VC was observed at Chandigarh (Fig. 6 a) with 198 ± 526 m 2 /s at 04:00h. It is important to note that ERA5 and IMDAA have coarser resolutions, and data with higher resolution will be more useful when we focus on looking at the city scale. However, due to the limited spatial and temporal coverage of ground-based observations, it is difficult to look at the continuous long-term (40 years in our case) trends of VC. From Table 1 , it can be seen that a lot of studies have focused on VC but for a short time. On the other hand, out study (long-term analysis) provides the opportunity to extrapolate future VC (and AQ) patterns for extended period. 3.3 Spatial and Temporal Trends Studying the VC to identify ventilation corridors (regions with low pollution potential) and suitable locations for new industries is important. It also helps to assess the damage caused by existing industries and determine the locations that are most vulnerable to the pollution plume (Abiye et al., 2016 ). The VC varies across India, depending on atmospheric and land surface conditions. Additionally, VC can also be influenced by regional topography, with areas near coastlines or mountainous regions having higher ventilation coefficients due to the influence of winds and the roughness of the terrain (e.g. Kannemadugu et al., 2021 ). A better understanding of the spatial and seasonal trends of the VC over India can help improve our understanding of air pollution mitigation strategies. In this analysis, we checked the VC trend in India throughout the 40 years for each grid point. We estimated the slope and p-value of VC using linear regression. We then observed the spatial pattern of the slope and its significance at a 95% confidence interval. As shown in Fig. 7 (a-h), both IMDAA and ERA5 exhibit a similar trend during pre-monsoon and monsoon, but IMDAA (Fig. 7 b, 7 c) captures high-resolution features in some areas not shown by ERA5 (Fig. 7 f, 7 g). During winter, IMDAA and ERA5 show a similar slope pattern, except in the IGP region, where IMDAA shows an increasing trend and ERA5 shows a decreasing trend (Fig. 7 a, 7 e). Decreasing trends in VC were found over some parts of central, west, south and northeast India and also in the state of Ladakh, with significant values over the IGP region. During pre-monsoon, a decreasing trend has been found in the central, west and south Indian regions whereas an increasing trend has been found in the north and some parts of northeast India, with significant values occurring mostly over the central and northeast India region (Fig. 7 b, 7 f). During monsoon, north India and IGP region is showing an increasing trend whereas the rest of the country is showing a decreasing trend in VC, with significant values over west, central and northeast Indian regions (Fig. 7 c, 7 g). The declining trend in VC during the pre-monsoon and monsoon seasons may be linked to a decrease in PBLH during these periods (e.g., Budakoti and Singh, 2021 ). In the post-monsoon season, an increasing trend in VC is observed over the western Ghats, west, north, and northeast Indian regions, while the remaining regions show a decreasing trend. The most significant values are concentrated in central and east India (Fig. 7 h). During post-monsoon (Fig. 7 d, 7 h), IMDAA and ERA5 exhibit a similar slope pattern, except in the eastern part of IGP. The temporal trends in ERA5 and IMDAA are shown in Fig. 8 over different Indian sub-regions. The analysis clearly indicates a decreasing trend, with some of the regions in different seasons also noticing increasing VC trends. IMDAA and ERA5 agree with each other during all seasons, with some discrepancies. Figure 8 (a-t), shows that the IMDAA and ERA5 are capturing a similar pattern or trend, where IMDAA is capturing higher variability as compared to ERA5. Most of the regions are showing a decreasing trend although they are not significant, a significant trend was only observed in central India during pre-monsoon (Fig. 8 g), and central and west India during monsoon (Fig. 8 l, 8 m). North and west India are the only regions where both IMDAA and ERA5 are showing an increasing trend in VC (Fig. 8 p, 8 r). During winter, the northern, western, and southern regions exhibit a declining trend (Fig. 8 a, 8 c, 8 e), while the central and eastern regions show a decreasing trend in ERA5 (Fig. 8 b, 8 d). During pre-monsoon, the central, west, east, and south regions show a decreasing trend (Fig. 8 g, 8 h, 8 i, 8 j), whereas north India shows ERA5 increasing and IMDAA increasing (Fig. 8 f). During monsoon, central, west, and south India show a decreasing trend (Fig. 8 l, 8 m, 8 n), while east India shows ERA5 increasing and IMDAA decreasing, and north India shows ERA5 decreasing and IMDAA decreasing (Fig. 8 n, 8 k). During post-monsoon, north and west India show an increasing trend (Fig. 8 p, 8 r), south India shows a decreasing trend (Fig. 8 t), and east and central India show ERA5 decreasing and IMDAA increasing (Fig. 8 s, 8 q). It is noteworthy to mention that unfavorable meteorological conditions for the dispersion of air pollution led to an increase in PM 2.5 concentrations (e.g. Hou et al., 2018 ). Better ventilation is more favorable for the dilution and outflow of PM 2.5 , it is also a fact that PM 2.5 should negatively correlate with VC. However, there are incidences with a positive correlation between PM 2.5 and VC, suggesting that ventilation also affects the inflow of PM 2.5 from outside of that region (e.g. Hou et al., 2018 ). The trends (year − 1 ) in MERRA-2 derived PM 2.5 are given in Table 4 from 1980 to 2019. We can see that PM 2.5 has significantly increased throughout the years in all the sub-regions during all the seasons, showing that particulate pollution has increased. West India during monsoon was the only region where the trends were insignificant. This increase in PM 2.5 through the four decades is desired as there is a huge technological and industrial gap between 1980 and 2019. A decreasing VC can be one of the reasons for the same as noticed in Fig. 7 , suggesting an inverse relationship between VC and PM 2.5 . However, the increase in emissions might play a bigger role in the increasing trend of PM 2.5 than VC, as trends of VC were insignificant and inconsistent. Kannemadugu et al. ( 2023 ) also brought attention to the long-term (1980–2019) VC and PM2.5 trends on a global scale, indicating a decrease in VC over the Indian subcontinent due to a downward trend in PBLH and wind speed across the region. The study also reported an increase in PM 2.5 concentrations over Indian regions, with the Indo-Gangetic Plain (IGP) showing the highest accumulation rate of PM 2.5 . Our findings align closely with those of the study. For short-term studies, MERRA-2 have been validated with ground-based observations over multiple cities in India (Chimurka et al., 2020; Bali et al., 2021 ), although long-term validation of MERRA-2 is unfeasible due to limited temporal coverage of the in-situ PM2.5 data. Table 4 Temporal trend (slope, p-value) of MERRA2 PM 2.5 (µg/m 3 /year) for the period of 1980–2019. Trends are statistically significant at a 95% confidence level. The p-value is only mentioned when the trend is insignificant. Season Vs Sub-region DJF MAM JJAS ON NORTH m = 0.58 m = 0.30 m = 0.29 m = 0.78 CENTRAL m = 1.45 m = 0.59 m = 0.38 m = 1.55 WEST m = 0.90 m = 0.24 m = 0.14 p-value = 0.06 m = 1.04 EAST m = 1.60 m = 1.00 m = 0.34 m = 1.02 SOUTH m = 0.74 m = 0.49 m = 0.23 m = 0.64 3.6 Interannual Variability Figure 9 (a-e) shows the interannual variability of VC for the period of 1980 to 2019. Here, normalization of VC was performed for each sub-region of India, for both ERA5 and IMDAA data and correlation coefficient (R) was computed to examine the fidelity of IMDAA (with ERA5). From Fig. 9 , it can be seen that ERA5 and IMDAA follow a similar pattern for all the sub-regions. R greater than 0.8 was observed over central, west, east and south India (Fig. 9 b, 9 c, 9 d, 9 e) whereas R was 0.57 over north India (Fig. 9 a). This indicates less correlation between ERA5 and IMDAA over north India as compared to the other sub-regions, with all the regions showing a significant correlation. It is evident that the VC has decreased over the four decades, the decrease is more prominent following the year 2000 in all the sub-regions, especially in central, west, east and south India as negative values are more clustered after 2000 (Fig. 9 b, 9 c, 9 d, 9 e). In north India there is no significant VC trend (Fig. 9 a). Table 5 makes it clearer, showing the slope values for the trends in each sub-region. The highest variability was found in south India (Fig. 9 e) as the frequency of occurrence of negative or positive VC values in consecutive years is less than in the other regions. For ERA5, the highest VC was found in west India during 1987 (Fig. 9 c), which was a strong El Niño year. For IMDAA, the highest VC was found in east India during 1992 (Fig. 9 d), which was a year following a strong El Niño year (1991). The lowest VC was found in south India during 2010 for both ERA5 and IMDAA (Fig. 9 e), this was a strong La Niña year. Table 5 Annual mean trend (slope) of VC (m 2 /s/year) for the period of 1980–2019. Trends are statistically significant at a 95% confidence level. The p-value is only mentioned when the trend is insignificant. NORTH CENTRAL WEST EAST SOUTH IMDAA m = 0.00 p-value = 0.96 m = -0.04 m = -0.03 m = -0.03 m = -0.03 ERA5 m = 0.00 p-value = 0.88 m = -0.04 m = -0.05 m = -0.03 m = -0.01 p-value = 0.48 3.7 Linkages with El Niño-Southern Oscillation The El Niño-Southern Oscillation (ENSO) is an air-sea interaction and dominant tropical climate driver phenomenon that can have a significant influence on atmospheric circulation and weather patterns all over the globe, including in India. ENSO is characterized by the oscillation of sea surface temperature in the Pacific Ocean, which can alter atmospheric circulation and modulate global weather patterns (Yang et al., 2018 ). The influence of ENSO patterns on air pollution in the emerging most polluted country, India, has received much less attention, as it is complex and completely dependent on a multitude of factors. The concise impact of ENSO on the ventilation coefficient, however, will be determined by the specific conditions in the region at the time. ENSO has a significant impact on air quality in the southern regions of China (Wang et al., 2022 ). Overall, the influence of ENSO on the characteristics of ventilation coefficients over India across the seasons is an important area of research that can help improve our understanding of atmospheric dynamics in the region and inform air pollution mitigation strategies. In this work, VC spatial patterns for every season were observed for the past five strong El Niño occurrences (1982, 1987, 1991, 1997 & 2015) and for the past five strong La Niña events (1988, 1999, 2000, 2007 & 2010) as well. To examine the impact of these events, the difference between VC of El Niño and VC of La Niña was observed, and its seasonal climatology was plotted along with the significant values (p-value < 0.05) as seen in Fig. 10 (a-h). During winter, El Niño influences positively (good air quality) over some parts of south India and Maharashtra. In the case of pre-monsoon season (Figs. 10 b, 10 f), the lowest VC values are noticed over the north, northwest and some central parts of the country, which indicates stronger El Niño conditions lead to poor air quality during the season, which is also well depicted by IMDAA. A high VC was observed over Bihar and southwest coast of India (Fig. 10 b, 10 f), with significant values over IGP, north India, Gujarat, Odisha, West Bengal and east coast of Tamil Nadu (Fig. 10 f). From the La Niña perspective, we observe higher VC (more negative values) over the north, northwest, and some central parts of India, indicating that strong La Niña events lead to improved air quality in these regions during the pre-monsoon season. A recent study by Beig et al. ( 2024 ) has established a correlation between strong La Niña events and PM 2.5 concentrations in India. Their findings suggest that the strong La Niña event of 2022–2023 resulted in better air quality in north India but worsened air quality in areas around Mumbai. While our analysis supports their results on a spatial scale, there is a discrepancy in timing; we observed improved air quality due to La Niña over north India during the pre-monsoon season, whereas the reported better air quality in the same region was during the winter of 2022-23 (Beig et al., 2024 ). Therefore, there is a need to further explore the role of the effect of ENSO on air quality in India to better understand these temporal variations and their implications on air quality. Coming back to our analysis, during monsoon, El Niño conditions improve the air quality by deepening the PBLH. Moreover, a significantly higher VC was observed over most parts of the country, however, the magnitude is highest in west and south India (Fig. 10 c, 10 g). During post-monsoon, better air pollution potential was observed over west India, especially in the states of Gujarat and Rajasthan (Fig. 10 d, 10 h). The significant values were found in clusters over north, central, south and east India (Fig. 10 h). From Fig. 10 , IMDAA consistently exhibits significant values throughout all seasons across nearly all regions of the country, while in contrast, ERA5 does not show as many notable values. These differences in IMDAA could be attributed to high-resolution data which captures more local meteorological conditions. 4. Prediction of VC using Machine Learning Approach Information from the preceding seasons could be important to train the models and predict the VC which is vital for planning air pollution control in India. In this section, we attempted to predict the VC (as a proxy of air pollution indicator) using the CNN-LSTM model, over the city of Delhi. The predictions were performed on ERA5 data. From the previous analysis, we saw that IMDAA was performing better compared to ERA5. However, since IMDAA is not validated and used extensively for VC over the Indian region, we used ERA5 for the machine learning predictions. Figure 11 (a-b) shows the daily predicted values of ERA5 VC for the year 2019 and its comparison with the observed values. From Fig. 11 a, it is evident that the model is capable of capturing VC trends that are very similar to observed values. The predicted values produced MAE, RMSE, MAPE and R 2 scores, which shows the good accuracy of the model. A good (and significant) correlation coefficient (R) score of 0.96 indicates a high linear relationship between predicted and observed values and confirms the capability of the CNN-LSTM model to capture the trends of VC (Fig. 11 b). The seasonal predictions of ERA5 VC can be seen in Fig. 12 (a-d), it is noticeable that the predictions were found to be reasonably well in all the seasons. Predictions for the monsoon season were most accurate (Fig. 12 c), this was followed by post-monsoon, pre-monsoon and finally winter (Fig. 12 d, 12 b, 12 a). For the winter season, it can be seen that the predicted values follow the same pattern as the observed values but it is unable to capture the same magnitude of VC as the observed values (Fig. 12 a). The same is the case for the pre-monsoon season, although the error (MAE, RMSE and MAPE) is less as compared to winter (Fig. 12 b). For the monsoon season, the model is capturing almost the same values as the observed (Fig. 12 c). In the case of post-monsoon as well the predicted values have the same trend as observed values but it is less accurate than the monsoon season (Fig. 12 d). It is important to understand that given all the conditions, the prediction of VC by considering only one variable (univariate time series) seems to be unconventional. VC is highly dynamic, other meteorological parameters (like temperature, relative humidity, solar radiation) which have strong relationships with VC, can further be used in the prediction of VC (multivariate predictions). Such studies in future can give the reasoning for high or low VC values, and also achieve higher accuracy. Please note that only ERA5 data was used for the prediction and validation of VC. And ERA5 reanalysis data, while widely used, is not without uncertainties. One significant source of uncertainty lies in the assimilation of observations, where gaps or errors in input data can impact the accuracy of the reanalysis. Additionally, uncertainties in the representation of physical processes and the complexity of Earth's systems contribute to variability in the results. Spatial and temporal resolutions may also introduce uncertainties, especially in regions with sparse observational data. Users of ERA5 should be mindful of these uncertainties, recognizing the need for caution and validation in specific applications requiring high precision and reliability. For example, in our work, winds and PBLH (Planetary Boundary Layer Height) are crucial parameters. These variables undergo validation against observational data. Recently, several studies (e.g., Guo et al., 2021 ; Li et al., 2023 ; Zhai et al., 2022 ) have reported a strong correlation between PBLH and winds from ERA5 and observational data. However, it is essential to note that reanalysis fields, being model-driven, inherently introduce some degree of error into the outputs. In this context, machine learning models emerge as a promising alternative (Gibson et al., 2021 ). They can potentially enhance the accuracy of these fields, especially in situations where observational data is limited. However, the use of modelled data (ERA5) to train machine learning models introduces its own set of challenges. The extensive nature of ERA5 data, as observed in our case with large VC data (daily data spanning from 1980 to 2019), presents a few challenges. A larger dataset implies longer training time, and managing such a substantial dataset can be a hindrance without access to significant computational resources. Gibson et al. ( 2021 ) emphasized the importance of employing machine learning on large climate models. Their study suggested that simulations from a machine learning model trained on a climate model could yield accurate seasonal precipitation forecasts, potentially outperforming existing climate models. It is well known that VC is considered a proxy for air quality. We further aimed to evaluate the performance of CNN-LSTM using ground-based data. As observational data for VC was unavailable, we utilized observations for PM2.5 instead. The study employed continuous PM2.5 observations from the Central Pollution Control Board (CPCB) for four metropolitan Indian cities, namely Delhi, Mumbai, Kolkata, and Bangalore. The data utilized in this step was from 2018 to 2019 for Delhi, Mumbai, Kolkata and from 2017 to 2019 for Bangalore. We used PM2.5 concentration during November and December 2019 for prediction (for testing), while the rest is used for training the model. It is noticeable that the model successfully captured the PM2.5 trend with relatively less error (MAE 0.8) for selected four metropolitan cities, as shown in Fig. 13 (a-d). Please note that machine learning models produce different outcomes for different datasets and parameters. In this study, we observed that predictions were correlated well with the observed values for both datasets (ERA5 - VC & CPCB – PM2.5). The reason for this may be that our data is univariate in nature (simpler than multivariate), therefore it is easier for this model to understand the nonlinear relationships in the data. Nevertheless, the strong performance of CNN-LSTM on both ERA5 reanalysis and in situ observations (from CPCB) implies its robust predictive capabilities. This suggests that this model could be used in future studies for forecasting, although, one should be educated and aware of its uncertainties before using it. While machine learning has shown success in prediction of both VC and PM2.5, there are various challenges while using it. Firstly, the availability and quality of data are important factors, insufficient and complex data may be difficult to work on (Ayturan et al., 2020 ; Huang and Kuo., 2018; Zhang and Li., 2022). In such cases accuracy of some models may be limited, however, the problem can be solved by exploring more machine/deep learning algorithms and finding which one works the best for that task. Furthermore, deep learning models like CNN-LSTM require high computational resources (Masood and Ahmad, 2020 ), which sometimes may not be available. Therefore, to overcome these limitations it is important to have expert knowledge of machine learning to be utilized properly and efficiently (Kanevski et al., 2008 ). 5. Summary and Conclusions We analyzed 40 years of VC data (1980–2019) derived from ERA5 and IMDAA reanalysis to study climatology, seasonal variability, diurnal variability, spatial and temporal trends of ventilation coefficient (VC), interannual variability including the effects of ENSO on VC. We also evaluated the performance of the CNN-LSTM model for time series prediction of VC. The study's conclusions are as follows: While most parts of the country showed very high pollution potential for all the seasons, high to medium pollution potential was observed only in west India during pre-monsoon, and in west and south India during monsoon seasons. All five sub-regions of India are showing high and variable VC during pre-monsoon and monsoon seasons whereas low and stable VC during post-monsoon and winter seasons. Higher values of VC during pre-monsoon season mainly because of high solar radiation which leads to an increase in temperature, further causing stronger vertical mixing and higher PBLH, which results in the dispersion of air pollutants thereby leading to good air quality. Higher values and variability of VC during monsoon season are likely because of high wind speed and convection. Variability in VC changes regionally from season to season with higher standard deviation noticed over western India and central India with IGP, northwest Rajasthan and eastern Ladakh showing the highest variability during the pre-monsoon season. Further higher variability is observed in central, western and southern India during the monsoon season. A high and variable VC was observed in the morning till the afternoon, while a low and invariable VC was observed in the evening till the early morning. A high and significant decreasing trend in VC was observed in the west, central and some parts of south India during pre-monsoon and monsoon seasons. Overall, ERA5 and IMDAA showed similar patterns in spatial trends, with contradictions only in the IGP region during winter and post-monsoon seasons. A decreasing trend in VC was observed from 1980–2019 in all sub-regions during all seasons, except for north India during pre-monsoon, east India during monsoon, and north and west India during post-monsoon. Overall, VC has decreased significantly over central, west, east and south India during the study period. And highest VC occurred during the strong El Niño year (1987) while the lowest VC was observed during the strong La Niña year (2010). ENSO had a significant impact on VC over India especially during pre-monsoon and monsoon seasons. El Niño caused low VC during pre-monsoon season over northwest, central and north India, and high VC during monsoon season over west, central and south Indian regions. Performance of the CNN-LSTM in predicting the VC is fairly well for each season and annual scale. Furthermore, CNN-LSTM is also able to capture the PM 2.5 over major cities. Overall, our findings from this long-term study could help identify ventilation corridors, which are spots with a high potential for pollutant dispersion, as well as suitable locations for new industrial operations. To summarize this study by comprehensive and region-specific approach while considering both seasonal variations and local influencing factors is crucial for effective air quality management in India. Declarations Acknowledgment The authors wish to thank ERA5 (https://rds.ncmrwf.gov.in/datasets) and IMDAA (https://cds.climate.copernicus.eu/#!/search?text=ERA5&type=dataset) database for their open access to the PBLH and wind speed (u and v component) data. We would also like to thank MERRA2 (https://disc.gsfc.nasa.gov/datasets/M2_TMAX_PM25_1/summary?keywords=PM2.5), and Central Pollution Control Board (CPCB) (https://airquality.cpcb.gov.in/ccr/#/caaqm-dashboard-all/caaqm-landing) for their open access to PM 2.5 data. Climate Data Operators was used for data processing and python programming language was used for data analysis. Funding No funding was received for conducting this study. Author information Authors and Affiliations Department of Earth and Environmental Sciences, Indian Institute of Science Education and Research Mohali, Mohali, Punjab, India: Amitabha Govande, Raju Attada, Krishna Kumar Shukla, Soumya Muralidharan, Garima Kaushik Centre for Atmospheric Sciences, Indian Institute of Technology Delhi, Delhi, India, 110016: Ravi Kumar Kunchala Department of Earth and Atmospheric Sciences, National Institute of Technology, Rourkela – 769008, India: Nagaraju Chilukoti Author contribution: AG and RA conceptualized the problem, perform the analysis and wrote the manuscript. KKS and SM prepared the tables and review the manuscript. RKK, NC and GK provided the scientific inputs on the discussions and edited the manuscript. All authors contributed toward the discussions and interpretation of the results. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Ethics declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable Competing interests The authors declare no competing interests. Data availability All the datasets used in the study are publicly available. References Abiye OE, Akinola OE, Sunmonu LA, Ajao AI, Ayoola MA (2016) Atmospheric ventilation corridors and coefficients for pollution plume released from an Industrial Facility in lle-lfe Suburb. Nigeria Afr J Environ Sci Technol 10(10):338–349 Allabakash S, Lin S (2020) Climatology of Planetary Boundary Layer Height-Controlling Meteorological Parameters Over the Korean Peninsula. Remote Sens 12:2571 Al-Salem SM, Khan AR (2010) Monitoring and Modelling the Trends of Primary and Secondary Air Pollution Precursors: The Case of the State of Kuwait. Int J Chem Eng. 879836 Apte J, Marshall J, Cohen AJ, Brauer M (2015) Addressing Global Mortality from Ambient PM2.5. Environ Sci Technol 49:8057–8066 Ayturan YA, Ayturan ZC, Altun HO, Kongoli C, Tuncez FD, Dursun S, Ozturk A (2020) Short-term prediction of PM2.5 pollution with deep learning methods. Global NEST J 22(1):126–131 Babu SS, Manoj MR, Moorthy KK, Gogoi MM, Nair VS, Kompalli SK, Satheesh SK, Niranjan K, Ramagopal K, Bhuyan PK, Singh D (2013) Trends in aerosol optical depth over Indian region: Potential causes and impact indicators. J Geophys Res Atmos. 118(20). 11,794–11,806. Bali K, Dey S, Ganguly D (2021) Diurnal patterns in ambient PM2.5 exposure over India using MERRA-2 reanalysis data. Atmos Environ 248:118180 Beig G, Anand V, Korhale N, Sobhana SB, Harshitha KM, Kripalani RH (2024) Triple dip La-Nina, unorthodox circulation and unusual spin in air quality of India. Sci Total Environ Volume 920:170963 Bran SH, Srivastava R (2017) Investigation of PM 2.5 mass concentration over India using a regional climate model. Environ Pollut 224:484–493 Budakoti S, Singh C (2021) Examining the characteristics of planetary boundary layer height and its relationship with atmospheric parameters over Indian sub-continent. Atmos Res 264:105854 Castelli M, Clemente FM, Popovic A, Silva S, Vanneschi L (2020) A Machine Learning Approach to Predict Air Quality in California. Complexity. Volume 2020: 8049504 Chan L, Qi-hong D, Wei-wei L, Bo-liang H, Ling-zhi S (2012) Characteristics of ventilation coefficient and its impact on urban air pollution. J Cent South Univ 19:615–622 Chicco D, Warrens MJ, Jurman G (2021) The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation. PeerJ Comput Sci 7:e623 Chimurkar N, Vinoj V, Pandey SK (2020) Evaluation of PM2.5 Surface Concentrations Simulated by NASA’s MERRA Version 2 Aerosol Reanalysis over India and its Relation to the Air Quality Index. Aerosol Air Qual Res 20:1329–1339 Chowdhary P, He Q, Male TL, Brune WH, Rudich Y, Pardo M (2018) Exposure of Lung Epithelial Cells to Photochemically Aged Secondary Organic Aerosol Shows Increased Toxic Effects. Environ Sci Technol Lett 5:424–430 Dey S, Di Girolamo L, van Donkelaar A, Tripathi SN, Gupta T, Mohan M (2012) Variability of outdoor fine particulate (PM2.5) concentration in the Indian Subcontinent: A remote sensing approach. Remote Sens Environ 127:153–161 Dey S, Purohit B, Balyan P, Dixit K, Bali K, Kumar A, Imam F, Chowdhury S, Ganguly D, Gargava P, Shukla VK (2020) Satellite-Based High-Resolution (1-km) Ambient PM 2.5 Database for India over Two Decades (2000–2019): Applications for Air Quality Management. Remote Sens 12(23):3872 Gao M, Sherman P, Song S, Yu Y, Wu Z, McElroy MB (2019) Seasonal prediction of Indian wintertime aerosol pollution using the ocean memory effect. Sci Adv 5:eaav4157 Gelaro R, McCarty W, Suarez MJ, Todling R, Molod A, Takacs L, Randles CA, Darmenov A, Bosilovich MG, Reichle R, Wargan K, Coy L, Cullather R, Draper C, Akella S, Buchard V, Conaty A, Da Silva AM, Gu W, Kim G, Koster R, Lucchesi R, Merkova D, Nielsen JE, Partyka G, Pawson S, Putman W, Rienecker M, Schubert SD, Sienkiewicz M, Zhao B (2017) The Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA2). J Clim 30:5419–5454 Gibson PB, Chapman WE, Altinok A, Monache LD, DeFlorio MJ, Waliser DE (2021) Training machine learning models on climate model output yields skillful interpretable seasonal precipitation forecasts. Commun Earth Environ 2:159 Gufran Beig V, Anand N, Korhale SB, Sobhana KM, Harshitha RH, Kripalani (2024) Triple dip La-Nina, unorthodox circulation and unusual spin in air quality of India. Sci Total Environ Volume 920:170963 Guo J, Zhang J, Yang K, Liao H, Zhang S, Huang K, Lv Y, Shao J, Yu T, Tong B, Li J, Su T, Yim SHL, Stoffelen A, Zhai P, Xu X (2021) Investigation of near-global daytime boundary layer height using high-resolution radiosondes: first results and comparison with ERA5, MERRA-2, JRA-55, and NCEP-2 reanalysis. Atmos Chem Phys 21:17079–17097 Hersbach H, Bell B, Berrisford P, Hirahara S, Hornayi A, Munoz-Sabater J, Nicolas J, Peubey C, Radu R, Schepers D, Simmons A, Soci C, Abdalla S, Abellan X, Balsamo G, Bechtold P, Biavati G, Bidlot J, Bonavita M, Chiara G, Dahlgren P, Dee D, Diamantakis M, Dragani R, Flemming J, Forbes R, Fuentes M, Geer A, Haimberger L, Healy S, Hogan RJ, Holm E, Janiskova M, Keeley S, Laloyaux P, Lopez P, Lupu C, Radnoti G, Rosnay P, Rozum I, Vamborg F, Villaume S, Thepaut J (2020) The ERA5 global reanalysis. Q J R Meteorol Soc 146:1999–2049 Holzworth GC (1967) Mixing Depths, Wind Speeds and Air Pollution Potential for Selected Locations in the United States. Journal of Applied Meteorology (1962–1982). 6(6), 1039–1044 Hou X, Fei D, Kang H, Zhang Y, Gao J (2018) Seasonal statistical analysis of the impact of meteorological factors on fine particle pollution in China in 2013–2017. Nat Hazards 93:677–698 Huang C, Kuo P (2018) A Deep CNN-LSTM Model for Particulate Matter (PM2.5) Forecasting in Smart Cities. Sensors 2018. 18, 2220 Iyer US, Raj PE (2012) Ventilation coefficient trends in the recent decades over four major Indian metropolitan cities. J Earth Syst Sci 122(2):537–549 Jena C, Ghude S, Kumar R, Debnath S, Govardhan G, Soni VK, Kulkarni SH, Beig G, Nanjundiah RS, Rajeevan M (2021) Performance of high resolution (400 m) PM 2.5 forecast over Delhi. Nat Sci Rep 11:4104 Jethva H, Torres O, Field RD, Lyapustin A, Gautam R, Kayetha V (2019) Connecting crop productivity, residue fires, and air quality over northern India. Sci Rep 9:16594 Kalita G, Kunchala RK, Fadnavis S, Kaskaoutis DG (2020) Long term variability of carbonaceous aerosols over Southeast Asia via reanalysis: Association with changes in vegetation cover and biomass burning. Atmos Res 245:105064 Kanevski M, Pozdnukhov A, Timonin V (2008) Machine Learning Algorithms for GeoSpatial Data. Application and Software Tools, vol 53. International Congress on Environmental Modelling and Software Kannemadugu H, Dorligjav S, Gharai B, Seshasai MVR (2021) Satellite-Based Air Pollution Climatology over India. Water Air Soil Pollution 232:365 Kannemadugu H, Dorligjav S, Taori A, Bothale RV, Chauhan P (2023) Global Air Pollution Potential: long term (1980–2019) trends and application to sustainable development, PREPRINT (Version 1) available. https://doi.org/10.21203/rs.3.rs-2888054/v1] . at Research Square Karuna MS, Upadhyay OP, Saxena DK, Swamy MM, Majumder S (2017) Air Quality Index, Ventilation Coefficient and Pollution Potential Studies Over Bareilly City, Uttar Pradesh. Int J Curr Res 9(12):63243–63255 Kumar A, Goyal P (2011) Forecasting of air quality in Delhi using principle component regression technique. Atmospheric Pollution Res 2:436–444 Kumar DKV (2019) Study of Atmospheric Boundary Layer Height from radiosonde data over a flat terrain at VBIT - Hyderabad (17.4° N – 78.5° E). Int J Appl Eng Res ISSN 14(4):0973–4562 Kumar N, Parmer K, Soni K, Garg N, Agarwal R (2015) Prediction of ventilation coefficient, using a conjunction model of wavelet-Neuro-fuzzy model: A Case Study Delhi, India. Acad J Sci Res 3(12):184–191 Li C, McLinden C, Fioletov V, Krotkov N, Carn S, Joiner J, Streets D, He H, Ren X, Li Z, Dickerson RR (2017) India is overtaking China as the World’s largest emitter of anthropogenic sulfur dioxide. Sci Rep 7:14304 Li X, Dong Y, Zhang Y, Shi Z, Yao J (2023) Climatology of Planetary Boundary Layer Height over Jiangsu, China, Based on ERA5 Reanalysis Data. Atmosphere 14:1330 Maheshwarkar P, Sunder Raman R (2021) Population exposure across central India to PM2.5 derived using remotely sensed products in a three-stage statistical model. Sci Rep 11:544 Masood A, Ahmad K (2020) A model for particulate matter (PM2.5) prediction for Delhi based machine learning approaches. Procedia Comput Sci 167:2101–2110 Moreira GA, Guerrero-Rascado JL, Bravo-Aranda JA, Foyo-Moreno I, Cazorla A, Alados I, Lyamani H, Landulfo E, Alados-Arboledas L (2020) Study of the planetary boundary layer height in an urban environment using a combination of microwave radiometer and ceilometer. Atmos Res 240:104932 Nair VS, Moorthy KK, Alappattu DP, Kunhikrishnan PK, George S, Nair PR, Babu SS, Abish B, Satheesh SK, Tripathi SN, Niranjan K, Madhavan BL, Srikant V, Dutt CBS, Badrinath KVS, Reddy RR (2007) Wintertime aerosol characteristics over the Indo-Gangetic Plain (IGP): Impacts of local boundary layer processes and long‐range transport. J Geophys Research: Atmos. 112. D13205 Nizar S, Dodamani BM (2019) Spatiotemporal distribution of aerosols over the Indian subcontinent and its dependence on prevailing meteorological conditions. Air Qual Atmos Health 12:503–217 Ojha N, Sharma A, Kumar M, Girach I, Ansari TU, Sharma SK, Singh N, Pozzer A, Gunthe SS (2020) On the widespread enhancement in fine particulate matter across the Indo-Gangetic Plain towards winter. Sci Rep 10:5862 Priyanka S, Soni K, Nair AS, Singh M (2022) Regression analysis of ventilation coefficient at a semi-arid IGP region using forward selection technique. Mausam 73(3):617–626 Ramachandran S, Kedia S, Srivastava R (2012) Aerosol optical depth trends over different regions of India. Atmos Environ 49:338–347 Rani SI, Arulalan T, George JP, Rajagopal EN, Renshaw R, Maycock A, Barker DM, Rajeevan M (2021) IMDAA: High-Resolution Satellite-Era Reanalysis for the Indian Monsoon Region. J Clim 34:5109–5133 Ratnam MV, Prasad P, Raj STA, Raman MR, Basha G (2021) Changing patterns in aerosol vertical distribution over South and East Asia. Sci Rep 11:308 Saha D, Soni K, Mohanan MN, Singh M (2019) Long-term trend of ventilation coefficient over Delhi and its potential impacts on air quality. Remote Sens Applications: Soc Environ 15:100234 Sujatha P, Mahalakshmi DV, Ramiz A, Rao PVN, Naidu CV (2016) Ventilation coefficient and boundary layer height impact on urban air quality. Cogent Environ Sci 2:1. 10.1080/23311843.2015.1125284 Shukla KK, Sarangi C, Attada R, Kumar P (2022) Characteristic dissimilarities during high aerosol loading days between western and eastern Indo-Gangetic Plain. Atmos Environ 269:118837 Venkataraman C, Brauer M, Tibrewal K, Sadavarte P, Ma Q, Cohen A, Chaliyakunnel S, Frostad J, Kilmont Z, Martin RV, Millet DB, Philip S, Walker K, Wang S (2018) Source influence on emission pathways and ambient PM2.5 pollution over India (2015–2050). Atmos Chem Phys 18:8017–8039 Wang S, Li S, Xing J, Yang J, Dong J, Qin Y, Sahu SK (2022) Evaluation of the Influence of El Nino-Southern Oscillations on air qualityin the southern China from long-term historical observations. Front EnvironmentalScience Eng 16(2):26 Xiao F, Yang M, Fan H, Fan H, Fan G, Al-qaness M (2020) An improved deep learning model for predicting daily PM2.5 concentration. Sci Rep 10:20988 Xie B, Yang Y, Wang P, Liao H (2022) Impacts of ENSO on wintertime PM2.5 pollution over China during 2014–2021. Atmospheric Ocean Sci Lett 15:100189 Yang S, Li Z, Yu J, Hu X, Dong W, He S (2018) El Niño-Southern Oscillations and its impact in the changing climate. Natl Sci Rev 5:840–857 Zhai R, Huang C, Yang W, Tang L, Zhang W (2022) Applicability Evaluation of ERA5 wind and wave reanalysis data in South China Sea. J Oceanol Limnol 41(2):495–517 Zhang J, Li S (2022) Air quality index forecast in Beijing based on CNN-LSTM multi-model. Chemosphere 308:136180 Additional Declarations No competing interests reported. Supplementary Files SupplementaryInformations.docx Cite Share Download PDF Status: Published Journal Publication published 29 Oct, 2025 Read the published version in Theoretical and Applied Climatology → Version 1 posted Editorial decision: Revision requested 25 Aug, 2025 Reviews received at journal 25 Aug, 2025 Reviews received at journal 17 Aug, 2025 Reviewers agreed at journal 15 Aug, 2025 Reviewers agreed at journal 15 Aug, 2025 Reviewers agreed at journal 15 Aug, 2025 Reviewers agreed at journal 15 Aug, 2025 Reviewers agreed at journal 14 Aug, 2025 Reviewers agreed at journal 14 Aug, 2025 Reviewers agreed at journal 13 Aug, 2025 Reviewers agreed at journal 13 Aug, 2025 Reviewers agreed at journal 13 Aug, 2025 Reviewers agreed at journal 13 Aug, 2025 Reviewers agreed at journal 04 Feb, 2025 Reviewers invited by journal 04 Feb, 2025 Editor assigned by journal 09 Jun, 2024 Submission checks completed at journal 09 Jun, 2024 First submitted to journal 08 Jun, 2024 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4551619","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":317426135,"identity":"0489e206-fd61-47f5-a3bd-9010ea16b6e3","order_by":0,"name":"Amitabha Govande","email":"","orcid":"","institution":"Indian Institute of Science Education and Research Mohali","correspondingAuthor":false,"prefix":"","firstName":"Amitabha","middleName":"","lastName":"Govande","suffix":""},{"id":317426136,"identity":"f338ed7d-2fe3-4837-adf9-ebf463883b0e","order_by":1,"name":"Raju Attada","email":"data:image/png;base64,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","orcid":"","institution":"Indian Institute of Science Education and Research Mohali","correspondingAuthor":true,"prefix":"","firstName":"Raju","middleName":"","lastName":"Attada","suffix":""},{"id":317426137,"identity":"d369295b-d401-4331-acec-b864984d53aa","order_by":2,"name":"Krishna Kumar Shukla","email":"","orcid":"","institution":"Indian Institute of Science Education and Research Mohali","correspondingAuthor":false,"prefix":"","firstName":"Krishna","middleName":"Kumar","lastName":"Shukla","suffix":""},{"id":317426138,"identity":"ff379e4e-64ed-4290-873d-f019ca4452a8","order_by":3,"name":"Soumya Muralidharan","email":"","orcid":"","institution":"Indian Institute of Science Education and Research Mohali","correspondingAuthor":false,"prefix":"","firstName":"Soumya","middleName":"","lastName":"Muralidharan","suffix":""},{"id":317426139,"identity":"c627b993-58dc-46a9-8c10-538c0cfc89a4","order_by":4,"name":"Ravi Kumar Kunchala","email":"","orcid":"","institution":"Indian Institute of Technology Delhi","correspondingAuthor":false,"prefix":"","firstName":"Ravi","middleName":"Kumar","lastName":"Kunchala","suffix":""},{"id":317426140,"identity":"aa080499-e6b4-4449-ad43-42a2a03646b2","order_by":5,"name":"Nagaraju Chilukoti","email":"","orcid":"","institution":"National Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Nagaraju","middleName":"","lastName":"Chilukoti","suffix":""},{"id":317426141,"identity":"a1f1ca3d-e547-48b8-b1f2-9ff9129635a9","order_by":6,"name":"Garima Kaushik","email":"","orcid":"","institution":"Indian Institute of Science Education and Research Mohali","correspondingAuthor":false,"prefix":"","firstName":"Garima","middleName":"","lastName":"Kaushik","suffix":""}],"badges":[],"createdAt":"2024-06-08 18:53:21","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4551619/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4551619/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s00704-025-05845-w","type":"published","date":"2025-10-29T15:58:35+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":58938270,"identity":"8f1a97d2-111a-4f8c-9e9e-f6db271bcfb7","added_by":"auto","created_at":"2024-06-24 10:38:34","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":943541,"visible":true,"origin":"","legend":"\u003cp\u003eThe spatial map of the Indian sub-continent with five different geographical sub-regions namely North (72.5°E - 80.5°E \u0026amp; 29.3°N - 37°N), Central (75.1°E - 84.9°E \u0026amp; 18.2°N - 29.2°N), West (68°E - 75°E \u0026amp; 18.2°N - 29.2°N), East (85°E - 97.5°E \u0026amp; 19°N - 29.4°N) and South (73°E - 83.5°E \u0026amp; 8°N - 18°N) adopted in this study.(The map was produced using the natural earth data, openly available at - https://www.naturalearthdata.com/ . Visualization of the map, subregions and citieswas performed using pythonprogramming language)\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-4551619/v1/6235e0dcbebcae216f52d5ed.png"},{"id":58938802,"identity":"8ac1fe3b-5dc8-4fd5-9fad-93f7bc6b1ff3","added_by":"auto","created_at":"2024-06-24 10:46:34","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":41462,"visible":true,"origin":"","legend":"\u003cp\u003eFlow chart for CNN-LSTM model used for prediction of VC.\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-4551619/v1/425218eb54319b27939b5ce8.png"},{"id":58939171,"identity":"5a1dec67-b74f-4f14-9a0a-35489ec177c5","added_by":"auto","created_at":"2024-06-24 10:54:34","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":202090,"visible":true,"origin":"","legend":"\u003cp\u003eThe spatial climatological (1980-2019) variation of seasonal mean ventilation coefficient (m\u003csup\u003e2\u003c/sup\u003e/s) of (a-d) IMDAA and (e-h) ERA5 over Indian sub-continent during winter (DJF), pre-monsoon (MAM), monsoon (JJAS), and post-monsoon (ON), respectively.\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-4551619/v1/03e560b2786e9caafe30b7e4.png"},{"id":58938271,"identity":"179f066c-b289-4181-8937-c965df7c8431","added_by":"auto","created_at":"2024-06-24 10:38:34","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":183447,"visible":true,"origin":"","legend":"\u003cp\u003eThe spatial variation of daily standard deviation of ventilation coefficient (m\u003csup\u003e2\u003c/sup\u003e/s) for (a-d) IMDAA and (e-h) ERA5 over Indian sub-continent during winter, pre-monsoon, monsoon, and post-monsoon for the period 1980-2019.\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-4551619/v1/93b70dc35f50b50a5975fa0a.png"},{"id":58938803,"identity":"bc29de67-8681-4239-944c-4a6d014012c6","added_by":"auto","created_at":"2024-06-24 10:46:34","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":188396,"visible":true,"origin":"","legend":"\u003cp\u003eAnnual cycles of VC (m\u003csup\u003e2\u003c/sup\u003e/s) over different Indian sub-regions (left panels) and metropolitan cities (right panels) from ERA5 (blue) and IMDAA (red) during 1980 to 2019. The vertical bars represent the standard deviation of VC.\u003c/p\u003e","description":"","filename":"image5.png","url":"https://assets-eu.researchsquare.com/files/rs-4551619/v1/05cad8a55015434e6a534981.png"},{"id":58938807,"identity":"316cf53e-5154-4073-83ea-9f0849460ab5","added_by":"auto","created_at":"2024-06-24 10:46:35","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":198138,"visible":true,"origin":"","legend":"\u003cp\u003eDiurnal variations of VC (m\u003csup\u003e2\u003c/sup\u003e/s) climatology (1980-2019) over different Indian sub-regions (left panels) and metropolitan cities (right panels) from ERA5 (blue) and IMDAA (red). X-axis values are in Indian Standard Time (IST) zone. The vertical bars represent the standard deviation of VC.\u003c/p\u003e","description":"","filename":"image6.png","url":"https://assets-eu.researchsquare.com/files/rs-4551619/v1/4adb441054577beca8daa7ed.png"},{"id":58938805,"identity":"8f0fa8cc-6d61-45fa-9290-aa7b1edc963d","added_by":"auto","created_at":"2024-06-24 10:46:34","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":324471,"visible":true,"origin":"","legend":"\u003cp\u003eSpatial trends of ventilation coefficient (VC; m\u003csup\u003e2\u003c/sup\u003e/s/year) for winter, pre-monsoon, monsoon and post-monsoon seasons from (a-d) IMDAA and (e-h) ERA5 for the period of 1980-2019. Statistically significant trends at 95% confidence level are marked with a black dot symbol.\u003c/p\u003e","description":"","filename":"image7.png","url":"https://assets-eu.researchsquare.com/files/rs-4551619/v1/04470eca8849a85880dea1c5.png"},{"id":58938806,"identity":"aa0b26d8-16d7-4059-a89f-8cdc156b31b6","added_by":"auto","created_at":"2024-06-24 10:46:34","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":225452,"visible":true,"origin":"","legend":"\u003cp\u003eTemporal trend over different sub-regional area-averaged VC (m\u003csup\u003e2\u003c/sup\u003e/s) from ERA5 (blue) and IMDAA (red) across the Indian subcontinent during (a-e) winter, (f-j) pre-monsoon, (k-o) monsoon and (p-t) post-monsoon seasons.\u003c/p\u003e","description":"","filename":"image8.png","url":"https://assets-eu.researchsquare.com/files/rs-4551619/v1/dcdd423158030e7f6b1e7194.png"},{"id":58938275,"identity":"1969f511-069f-42b3-b40b-a46c182fafa4","added_by":"auto","created_at":"2024-06-24 10:38:34","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":100282,"visible":true,"origin":"","legend":"\u003cp\u003eInterannual variability of VC (m\u003csup\u003e2\u003c/sup\u003e/s) over (a) North, (b) Central, (c) West, (d) East and (e) South Indian regions for the period 1980-2019 from ERA5 and IMDAA. R represents the correlation coefficient between IMDAA and ERA5, while p represents the significance (p-value) of the correlation.\u003c/p\u003e","description":"","filename":"image9.png","url":"https://assets-eu.researchsquare.com/files/rs-4551619/v1/55fefe3f4ad6ec27d9b0ce16.png"},{"id":58938281,"identity":"67dd6e32-799d-4ee2-8c91-0698e811d5dd","added_by":"auto","created_at":"2024-06-24 10:38:35","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":407566,"visible":true,"origin":"","legend":"\u003cp\u003eThe spatial distribution of seasonal mean VC (m\u003csup\u003e2\u003c/sup\u003e/s) difference between El Niño and La Nina events from (a-d) IMDAA and (e-h) ERA5 during different seasons. The statistically significant trends in VC at 95% confidence level are marked with a black dot symbol.\u003c/p\u003e","description":"","filename":"image10.png","url":"https://assets-eu.researchsquare.com/files/rs-4551619/v1/5d01e17b935e7cc170c13c89.png"},{"id":58938278,"identity":"5cec3467-ecb5-4c93-af4c-b41e987d29df","added_by":"auto","created_at":"2024-06-24 10:38:35","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":137510,"visible":true,"origin":"","legend":"\u003cp\u003e(a)Temporal variations of predicted VC usingCNN-LSTM model along with ERA5 derived VC during 2019 over the Delhi region. (b) Scatter plot between the daily mean observed (ERA5) and CNN-LSTM predicted VC during 2019 over the Delhi region.\u003c/p\u003e","description":"","filename":"image11.png","url":"https://assets-eu.researchsquare.com/files/rs-4551619/v1/acc2e85b7f78dc0003043d25.png"},{"id":58938279,"identity":"581097aa-67d8-4f1b-b757-7a2da54b8368","added_by":"auto","created_at":"2024-06-24 10:38:35","extension":"png","order_by":12,"title":"Figure 12","display":"","copyAsset":false,"role":"figure","size":156353,"visible":true,"origin":"","legend":"\u003cp\u003eSeasonal prediction of VC (m\u003csup\u003e2\u003c/sup\u003e/s) from CNN-LSTM and compared with ERA5 over the Delhi region during (a) Winter (DJF), (b) Pre-monsoon (MAM), (c) Monsoon (JJAS) and (d) Post-monsoon (ON) seasons for the year 2019.\u003c/p\u003e","description":"","filename":"image12.png","url":"https://assets-eu.researchsquare.com/files/rs-4551619/v1/3a3e49a0f407e1510039cf3e.png"},{"id":58938282,"identity":"72d8a401-3251-44d8-bee7-26b202849f96","added_by":"auto","created_at":"2024-06-24 10:38:35","extension":"png","order_by":13,"title":"Figure 13","display":"","copyAsset":false,"role":"figure","size":151221,"visible":true,"origin":"","legend":"\u003cp\u003ePrediction of PM\u003csub\u003e2.5\u003c/sub\u003e usingCNN-LSTM model and evaluated with CPCB observations the post-monsoon season only for the year 2019 over four Indian metropolitan cities−Delhi (a), Mumbai (b), Kolkata (c) and Bangalore (d).\u003c/p\u003e","description":"","filename":"image13.png","url":"https://assets-eu.researchsquare.com/files/rs-4551619/v1/96b2eef293d88d361dcf2b85.png"},{"id":95041347,"identity":"bcb1723b-d225-4cf1-adf0-4634938b4ccb","added_by":"auto","created_at":"2025-11-03 16:11:13","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4330106,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4551619/v1/5dffcec6-d1a7-48d3-8ff8-48164f16e061.pdf"},{"id":58938268,"identity":"730dbf36-5790-4258-9000-ef587dce777e","added_by":"auto","created_at":"2024-06-24 10:38:34","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":18683,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryInformations.docx","url":"https://assets-eu.researchsquare.com/files/rs-4551619/v1/2d343a0f9f005db3a9d2f3ce.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Analyzing and Predicting Ventilation Coefficient over India using Long-term Reanalysis Datasets and Hybrid Machine Learning Approach ","fulltext":[{"header":"Highlights","content":"\u003cul\u003e\n \u003cli\u003eThe spatial pattern of seasonal climatological mean VC during winter and post-monsoon seasons indicates poor air quality in India.\u003c/li\u003e\n \u003cli\u003eVC exhibits substantial diurnal variability.\u003c/li\u003e\n \u003cli\u003eENSO has been found to have an effect on air pollution dispersion capacity.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eCNN-LSTM machine learning model successfully predicts air pollution potential over India.\u0026nbsp;\u003c/li\u003e\n\u003c/ul\u003e"},{"header":"1. Introduction","content":"\u003cp\u003eAir pollution is caused due to various human activities such as waste generation, transportation, construction, coal burning, and resource exploitation, which are intensified by population growth and economic development (e.g. Saha et al., \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Hou et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Karuna et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Iyer and Raj, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). The accumulation and dispersion of pollutants depend on regional meteorological conditions. Additionally, the faster pace of urbanization of metropolitan cities of India affects the deterioration of Air Quality (AQ) rapidly, which has a detrimental effect on the environment and human health (Chowdhary et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Apte et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Altogether, the high-level exposure to air pollutants causes morbidity and mortality in any region. Air pollutant sources are generally classified into natural and anthropogenic. Natural sources release air pollutants directly into the atmosphere, including volcanoes, biogenic emissions, dust storms, and forest fires, emitting substances such as Particulate Matter (PM), SO\u003csub\u003e2\u003c/sub\u003e, CO, NOx, etc. On the other hand, anthropogenic sources release harmful particles through activities such as vehicular emissions, combustion processes, burning fossil fuels, waste management, and construction. Major air pollutants from these sources include CO, VOCs, PM, CO, and NOx. Primary pollutants in the atmosphere are directly emitted from the sources as discussed above, while secondary pollutants are formed through photochemical reactions (when primary pollutants react with other compounds) in the presence of solar energy (Castelli et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Al-Salem and Khan, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). Distinguishing between natural and anthropogenic sources and their contributions is vital for effective air quality management and pollution control efforts.\u003c/p\u003e \u003cp\u003eThe air pollutants loading shows the regional and seasonal heterogeneity due to changing background meteorology and topography over the Indian region (Babu et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Ratnam et al., \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). The dominance of different pollutant sources varies during different seasons in India which has been reported previously (Dey et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Shukla et al., \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). The Indo-Gangetic Plain (IGP) is highly populated and is also considered as a hotspot for elevated aerosol loading. The air pollutant loading across IGP is higher throughout the year than in the rest of the country. The pollutant loading is maximum during the winter season due to bio-fuel burning for heating and cooking, crop-residue burning and calm wind conditions over IGP and north India favours the accumulation of these aerosols (Chimurkar et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). The west India region is dominated by dust aerosols during pre-monsoon season due to the frequent occurrence of dust storms and long-range transport of dust from arid regions (Shukla et al., \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), while it also showed high PM loading during the post-monsoon season due to stubble burning (Jethva et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Central India receives more pollutants during winter and post-monsoon seasons due to biomass and crop-residue burning, whereas low concentrations during monsoon due to wet removal of aerosols (Maheshwarkar and Sunder Raman, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). In addition to the above, the coal burning in thermal power plants and industries is the main contributor of pollutants in eastern and peninsular India (Li et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Venkatraman et al., 2018). Anthropogenic aerosols (such as carbonaceous aerosols and sulphate) are dominant over eastern parts of India which is mainly due to biomass burning and emission from the thermal power plants located in neighbouring states (such as Bihar, Uttar Pradesh and Odisha) (Kalita et al., \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Rawat et al., 2019). The source variability of pollutants and seasonal heterogeneity highlight the complexity of air quality management in India, necessitating region-specific strategies to effectively address air pollution challenges.\u003c/p\u003e \u003cp\u003eThe Ventilation Coefficient (VC) is an essential and well-recognized factor that determines the air pollution dispersion potential and AQ over a region. VC depends on Planetary Boundary Layer Height (PBLH) and average wind speed over any region. A higher VC indicates good AQ with high dispersive capacity, while a lower VC indicates poor ventilation (or AQ). Several studies have discussed the variation of VC across the Indian sub-continent (Kannemadugu et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Kumar, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Saha et al., \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Karuna et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Sujatha et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2016\u003c/span\u003e) as well as global scale (Abiye et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Chan et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Holzworth, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e1967\u003c/span\u003e). For example, Holzworth (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e1967\u003c/span\u003e) reported that Los Angeles had low VC (poor air quality) among the seven locations studied in the United States. Abiye et al. (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2016\u003c/span\u003e) reported the highest (lowest) VC during day (night) time over an industrial site in Nigeria. Previous studies also demonstrate the influence of VC on particulate matter (PM\u003csub\u003e2.5\u003c/sub\u003e and PM\u003csub\u003e10\u003c/sub\u003e) mass concentration at various point locations in China across the seasons (Hou et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Chan et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2012\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAs far as the Indian context, the AQ deteriorates over time due to urbanization and other economic developments that impact various sectors such as human health, aviation, soil characteristics, and water quality. This underpins the importance of studying the air quality dispersion potential using VC over different sub-regions of India. Hence, some researchers have attempted to investigate the VC and its linkages with AQ over the Indian region which is well-summarized in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Recently, Kannemadugu et al. (\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) demonstrated spatial variability (2014\u0026ndash;2019) of VC across the Indian subcontinent, with high VC observed on the east coast of Andhra Pradesh and Tamil Nadu during winter and in western India and New Delhi during the summer monsoon. Iyer and Raj, (2013) showed a decreasing VC trend over some Indian metropolitan cities, leading to poor AQ during 1971\u0026ndash;2000. Saha et al. (\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) reported a rising trend in VC (70 m\u003csup\u003e2\u003c/sup\u003e/s/year) over the capital of India but for a short period from 2006 to 2014. The changes/variability in VC is a result of various regional factors. For example, the local conditions, including topography, and other meteorological parameters, have a significant impact on both the Planetary Boundary Layer Height (PBLH) and wind speed. As a result, the VC is a multifaceted phenomenon that is shaped by various factors. Budakoti and Singh, (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) reported a negative correlation between PM\u003csub\u003e2.5\u003c/sub\u003e and Planetary Boundary Layer Height (PBLH) over the Indian region. On the other hand, Sujatha et al. (\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2016\u003c/span\u003e) found that VC has a strong negative correlation with Black Carbon (BC) in the city of Hyderabad.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eLiterature review on ventilation coefficient studies over Indian subcontinent.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRegion, point location\u003c/p\u003e \u003cp\u003eover India\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTime period\u003c/p\u003e \u003cp\u003eMonth, seasons, annual\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMethod of calculation VC using observation and reanalysis\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMain conclusions\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eReference\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMumbai, Delhi, Kolkata, Chennai\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1971\u0026ndash;2000, Winter months\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eDaily radiosonde data, obtained from IMD Pune\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eDecreasing trend in all the four regions. During Dec and Feb, VC decreased by 49 and 32 m\u003csup\u003e2\u003c/sup\u003e/s/year over Delhi, 15 m\u003csup\u003e2\u003c/sup\u003e/s/year over Mumbai and 14 and 17 m\u003csup\u003e2\u003c/sup\u003e/s/year\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eIyer and Raj, (\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2012\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDelhi\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2005\u0026ndash;2014\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePBLH -SODAR data over Delhi region - obtained by CPCB\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eIncreasing trend over Delhi with 70 m\u003csup\u003e2\u003c/sup\u003e/s/year\u003c/p\u003e \u003cp\u003eVC: monsoon\u0026thinsp;\u0026gt;\u0026thinsp;pre-monsoon\u0026thinsp;\u0026gt;\u0026thinsp;winter\u0026thinsp;\u0026gt;\u0026thinsp;post-monsoon.\u003c/p\u003e \u003cp\u003eARIMA model fit reasonably well with the data series with some discrepancies in VC during monsoon and pre-monsoon season.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSaha et al., (\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2019\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNew Delhi, Thar, Jaisalmer, Jodhpur, Patna, Kolkata, Nagpur, Rourkela, Mumbai, Hyderabad, Bangalore, Chennai\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2015\u0026ndash;2019\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePBLH - CrIS SOUMINPP satellite, wind speed - ERA5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eLow pollution potential (\u0026lt;\u0026thinsp;6000 m\u003csup\u003e2\u003c/sup\u003e/s) at east coast of Andhra Pradesh and TN during winter and western India and New Delhi during summer monsoon. During pre-monsoon, western Gujarat, south-west Rajasthan, and parts of IGP shows low pollution potential. Low pollution potential is observed at east coast of Tamil Naidu during post-monsoon.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eKannemadugu et al., (\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2021\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIVRI institute, Bareilly city, UP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2013\u0026ndash;2017\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHourly meteorological data from wunderground.com\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh VC during day and Low VC at night and early morning.\u003c/p\u003e \u003cp\u003eDuring winter max VC was found at 6000 m\u003csup\u003e2\u003c/sup\u003e/s and during summer, maximum VC was 15000 m\u003csup\u003e2\u003c/sup\u003e/s.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eKaruna et al., (\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2017\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVBIT, Hyderabad (17.4\u0026deg; N \u0026minus;\u0026thinsp;78.5\u0026deg; E)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eApril 2012 - Feb 2013,\u003c/p\u003e \u003cp\u003ePre-monsoon (April-May), monsoon (July-August), post-monsoon (Oct-Nov), winter (Dec-Jan)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh resolution GPS radiosonde flights carried out with iMet (USA) radiosonde\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eVC: monsoon\u0026thinsp;\u0026gt;\u0026thinsp;post-monsoon\u0026thinsp;\u0026gt;\u0026thinsp;pre-monsoon\u0026thinsp;\u0026gt;\u0026thinsp;winter\u003c/p\u003e \u003cp\u003eHigh wind speed during monsoon is causing high VC.\u003c/p\u003e \u003cp\u003eDiurnal pattern of VC shows high VC during the day or afternoon and low VC during night and early morning\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eKumar, (\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2019\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCSIR - National Physics Laboratory - Delhi\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMar 2019 - Feb 2020\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSODAR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eVC high and variable during day/afternoon and low and stable during night and early morning.\u003c/p\u003e \u003cp\u003eHigh VC of 30800 m\u003csup\u003e2\u003c/sup\u003e/s in March over Delhi\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003ePriyanka et al., (\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2022\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eAdditionally, global climate drivers like El Ni\u0026ntilde;o and La Ni\u0026ntilde;a also play an important role in modulating regional circulation patterns and associated meteorological conditions. In a recent study, Wang et al. (\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) examined the influence of El Ni\u0026ntilde;o\u0026ndash;Southern Oscillation (ENSO) on air quality and reported that in China good air quality is associated with El Ni\u0026ntilde;o events because of increased precipitation, while poor air quality is linked to La Ni\u0026ntilde;a because of reduced precipitation. Therefore, the interannual variability of the VC often links to the ENSO phases. In India, El Ni\u0026ntilde;o is usually associated with decreased precipitation and La Ni\u0026ntilde;a, on the other hand, induces more precipitation during summer monsoon months and this can contribute a lot to the AQ (Gao et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). These large-scale weather phenomena affect not only precipitation but also other factors like relative humidity and particulate matter and are likely to impact VC as well. The impact of ENSO on AQ and PM\u003csub\u003e2.5\u003c/sub\u003e concentrations has been the subject of a few studies (Xie et al., \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Wang et al., \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). However, understanding the role of El Ni\u0026ntilde;o and La Ni\u0026ntilde;a events on VC over India remains unexplored.\u003c/p\u003e \u003cp\u003ePrediction of VC is also extremely important for mitigating the adverse effects of air pollution in highly populated countries like India. Towards this end, it is imperative to examine the predictive models in capturing the observed VC in any given region. Artificial Intelligence and Machine Learning (AI/ML) are emerging as powerful tools to predict future weather/climate variables. It is important to emphasize that AQ prediction is quite complicated and challenging due to its quick variability at a shorter time scale over any given region (e.g., Castelli et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). AQ is a nonlinear phenomenon that is associated with numerous complexities and various meteorological factors (e.g. Castelli et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Some recent studies have highlighted the importance of AI/ML models in predicting regional weather and climate (Zhang and Li, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Castelli et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Kumar et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Traditional time series forecasting models, for example, Auto-Regresive (AR), Moving-Average (MA), Auto-Regressive Integrated Moving Average (ARIMA), etc, and machine learning-based (e.g. Multi-Layer Perceptron, Convolutional Neural Network, etc) predictions are among the currently used models for time series prediction (Zhang and Li, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). The former one has been used for the prediction of VC over Delhi by Saha et al. (\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2019\u003c/span\u003e), while the latter has been less explored for VC. Statistical models like time series forecasting are mostly based on linear correlations and therefore may not be able to comprehend the nonlinear characteristics effectively (Gibson et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Kumar et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). However, Machine learning models can be used to solve this problem. Kumar et al., (\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2015\u003c/span\u003e) has shown the successful application of Wavelet-Neuro-Fuzzy model in predicting VC derived from Sonic Detection and Ranging (SODAR) over Delhi. Masood and Ahmad., (2020) tried to predict in situ PM2.5 levels from Central Pollution Control Board (CPCB) over Delhi using Support Vector Machine (SVM) and Artificial Neural Network (ANN). Algorithms involving neural network like ANN, Recurrent Neural Network (RNN), Convolutional Neural Network (CNN), have been successful in forecasting air quality (Masood and Ahmad, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Huang and Kuo, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Xiao et al., \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Although, whenever dealing with extensive datasets sometimes features of only one algorithm are not enough to get the best possible predictions. Using a combination of two different models is a good option. Previous studies have reported that better results were observed when such hybrid models were used (Ayutran et al., 2020; Huang and Kuo, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Wang et al., \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Zhang and Li, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). One such model is CNN coupled with Long Short-Term Memory (CNN-LSTM), which has previously shown higher predictive ability over a few cities in China (Huang and Kuo, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Zhang and Li, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). CNN is well-known for extracting patterns or features even with the most complex nonlinear data (Zhang and Li, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Huang and Kuo, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). On the other hand, LSTM is known for solving long-term time dependency. Each node in the LSTM network acts as a memory cell, so it remembers every piece of information throughout time (Zhang and Li, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Huang and Kuo, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Importantly, every node in the LSTM network propagates the output backward, learns the error and continues moving forward for the correct/better predictions (e.g. Huang and Kuo, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). LSTM, when used with CNN, allows effective extraction of features and improves the accuracy of predictions.\u003c/p\u003e \u003cp\u003eThe studies on VC over India are primarily focused on a point location, short period or different seasons based on the availability of datasets. However, the long-term variations including the impact of large-scale climate modes on VC are still limited over the Indian subcontinent (Table S1 presents a comparison of our study with some previous works). Moreover, the point location observation of PBLH over India is available whereas the spatial homogeneous in-situ PBLH observations are still lacking. Given the above, the global and regional reanalysis datasets fill the gap in spatial and temporal coverage to understand the VC patterns. It provides high spatial and temporal resolution long-term data of PBLH and meteorological parameters. Thus, in this work, we first aim to investigate the long-term variability (seasonal, interannual and diurnal time scales), trends, and impact of El Ni\u0026ntilde;o events on VC over India using reanalysis datasets covering from 1980 to 2019. We have examined the VC trends for four different seasons defined by the India Meteorological Department (winter - DJF, pre-monsoon - MAM, monsoon - JJAS, and post-monsoon - ON) using ERA5 and IMDAA data. We further employed a CNN-LSTM model to construct a univariate time series model for predicting VC at seasonal and annual scales. This manuscript is organized as follows: section 2 describes the data and methods, section \u003cspan refid=\"Sec8\" class=\"InternalRef\"\u003e3\u003c/span\u003e presents results and discussions, section 4 elucidates the machine learning predictions and Section 5 summarizes the key findings of the study.\u003c/p\u003e"},{"header":"2. Data and Methodology","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Data used\u003c/h2\u003e \u003cp\u003eWe employed two high-resolution reanalysis datasets (global and regional) to understand the VC variability. The fifth-generation European Center for Medium-Range Weather Forecasts (ECMWF) reanalysis (ERA5, global) is the latest climate reanalysis with high spatial and temporal resolution available hourly at a spatial grid of 0.25\u003csup\u003eo\u003c/sup\u003e \u0026times; 0.25\u003csup\u003eo\u003c/sup\u003e on 37 pressure levels from 1979 to the present (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://cds.climate.copernicus.eu\u003c/span\u003e\u003cspan address=\"https://cds.climate.copernicus.eu\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e; Hersbach et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Additionally, Indian Monsoon Assimilation and Analysis (IMDAA, regional) provides long-term Indian regional reanalysis data for a wide range of atmospheric variables, available from 1979 to the present at a spatial grid of 12 km \u0026times; 12 km (Rani et al., 2020). This is India\u0026rsquo;s first high-resolution dataset that enhances the benefits of regional representation by accounting for better surface forcing, including orography, small-scale processes and features. It further improves the representation of the interaction between small- and large-scale processes. Therefore, the VC from the high-resolution IMDAA database provides an opportunity to study the long-term Air Pollution Climatology over India at sub-regional scales as it exhibits a regional heterogeneity over the Indian landmass. We obtained hourly PBLH, zonal (u) and meridional (v) wind datasets from IMDAA and ERA5 for 1980\u0026ndash;2019. ERA5 have previously been validated for PBLH and wind speed at a global scale (Guo et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), and used in studies related to VC (Kannemadugu et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Kannemadugu et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). And since it is a global reanalysis, our assumption is to consider ERA5 as the standard value and compare the performance of IMDAA based on it. This validation is important for the development of new generation reanalysis datasets for air quality studies. Further, we also checked the long-term seasonal PM2.5 trends, for this we used Modern-Era Retrospective analysis for Research and Applications \u0026ndash; Version 2 (MERRA-2) reanalysis (e.g. Gelaro et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2017\u003c/span\u003e) available at 0.5\u0026deg; \u0026times; 0.625\u0026deg; spatial resolution from 1980 to 2019.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Methodology\u003c/h2\u003e \u003cdiv id=\"Sec5\" class=\"Section3\"\u003e \u003ch2\u003e2.2.1 Estimation of ventilation coefficient\u003c/h2\u003e \u003cp\u003eVC is a product of PBLH and average wind speed through the mixing layer. PBLH represents the vertical mixing of pollutants (Budakoti and Singh, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Allabakash and Lim, 2020; Sujatha et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2016\u003c/span\u003e) while wind speed represents the horizontal mixing of pollutants (Chan et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). The Ventilation coefficient is given as:\u003c/p\u003e \u003cp\u003eVC = [PBLH (m) \u0026sdot; average wind speed (m/s)] --------- (1)\u003c/p\u003e \u003cp\u003eA Low (high) ventilation coefficient means less (more) dispersion potential of pollutants. The IMDAA data was re-gridded to 0.25\u0026deg; \u0026sdot; 0.25\u0026deg; using bilinear interpolation to compare the ERA5 results. The linear regression model was used to determine the best-fit line, slope, and p-value (to check statistically significant values at the confidence interval of 95% or p-value of 0.05). As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, the Indian region was further divided into five sub-regions to examine temporal variability of VC across the seasons as suggested by (Ramachandran et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Nizar and Dodamani, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Furthermore, we also extracted metropolitan cities lie in respective sub-regions of India to understand the diurnal patterns of VC.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section3\"\u003e \u003ch2\u003e2.2.2 Estimation of PM2.5 derived from MERRA-2\u003c/h2\u003e \u003cp\u003eWe estimated the PM\u003csub\u003e2.5\u003c/sub\u003e concentration (Eq.\u0026nbsp;2) using five major air pollutants (dust, black carbon (BC), organic carbon (OC), sulfate and sea salt in \u0026micro;g/m\u003csup\u003e3\u003c/sup\u003e) which are contributing to the PM\u003csub\u003e2.5\u003c/sub\u003e concentration. All of these components were derived from MERRA-2 at monthly averaged scales. The data was re-gridded to ERA5 resolution (0.25\u0026deg; \u0026sdot; 0.25\u0026deg;) using bilinear interpolation. More details regarding PM\u003csub\u003e2.5\u003c/sub\u003e mass concentration calculation can be found in Bali et al. (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) and Chimurkar et al. (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e \u003cp\u003ePM\u003csub\u003e2.5\u003c/sub\u003e = [DUST\u003csub\u003e2.5\u003c/sub\u003e] + [SS\u003csub\u003e2.5\u003c/sub\u003e] + [BC]\u0026thinsp;+\u0026thinsp;1.8 \u0026sdot; [OC]\u0026thinsp;+\u0026thinsp;1.375 \u0026sdot; [SO\u003csub\u003e4\u003c/sub\u003e] --------- (2)\u003c/p\u003e \u003cp\u003eIn Eq.\u0026nbsp;2, DUST\u003csub\u003e2.5\u003c/sub\u003e and Sea Salt (SS\u003csub\u003e2.5\u003c/sub\u003e) are particulate matter with a size less than 2.5 micrometers and BC is Black Carbon. To estimate the organic matter, OC is multiplied by a factor ranging from 1.2\u0026ndash;2.6 depending on space and time. This factor tells the contributions from other elements that are associated with organic matter (Bali et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). For the Indian region, the factor used is 1.8 (Chimurkar et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). The sulfate ion is mainly present in the form of ammonium sulfate in the atmosphere, to calculate the sulfate concentration, it is multiplied by the mass of the sulfate ion (1.375) (Bali et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). The present study estimated PM2.5 concentrations using the methodology outlined by Chimurkar et al. (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). This study also highlighted the limitations associated with this method in calculating PM2.5 concentrations from MERRA-2 data. Daily mean PM2.5 concentrations derived from MERRA-2 typically underestimate actual PM2.5 levels and exhibit a lower correlation, and unable to accurately capture PM2.5 concentrations during days of elevated pollution levels. Conversely, monthly mean PM2.5 (MERRA-2) concentrations demonstrate a high correlation with in-situ observations of PM2.5. Therefore, we used monthly data in our analysis. Furthermore, Bali et al (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) also validated hourly MERRA-2 PM2.5 across multiple sites over the country and found high correlation with the in-situ PM2.5 data from CPCB. We recommend utilizing the MERRA-2 data for analyzing a wide spatial coverage, but the data can give some discrepancies due to coarser resolution for city-scale analysis.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section3\"\u003e \u003ch2\u003e2.2.3 CNN-LSTM framework\u003c/h2\u003e \u003cp\u003eIn this study, we employed CNN-LSTM deep learning model for the prediction of VC at different lead times. For this, we used VC daily data obtained from ERA5 from 1980 to 2019 on a grid of the Delhi (extracted over 77\u0026deg; E to 77.25\u0026deg; E and 28.5\u0026deg; N to 28.75\u0026deg; N grids). It is worth mentioning that when it comes to air pollution, Delhi is one of the metropolitan Indian cities which usually suffers from poor air quality for most of the months in a year (e.g. Jena et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Kumar and Goyal, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). Total, 14610 data points have been retrieved over the span of 1980 to 2019. Then, our model was trained using 97.5% of the data from 1980 to 2018 period, and our testing data were 365 days (2.5% of the total data) for the year 2019. Due to the right-skewed nature of the data, log transformation was first performed (on both training and testing set) and the data appeared to be fairly symmetrical after this procedure. The transformed data was then processed in such a way that each value was predicted based on the 15 data points that came before it. Therefore, to predict the final value, we use the previous 15 values as our variables. After running the model, the final predictions were subjected to exponential transformation to retract the values of VC over Delhi. A similar kind of chronology was adopted for seasonal forecasts as well, for the same region. Here, for every season (winter, pre-monsoon, monsoon and post-monsoon), daily data from 1980 to 2018 (97.5% of the data) were taken for training and data for the year 2019 (2.5% of the data) was used for testing. For winter and post-monsoon, the previous three values were considered for prediction and in the case of pre-monsoon and monsoon previous four values were taken to predict the final values, just because the model performed best in these conditions.\u003c/p\u003e \u003cp\u003eDetailed Schematic diagram (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e) depicts the data processing part and final structure of the CNN-LSTM model: the Conv1D layer, the Maxpooling1D layer, the LSTM layer, and a dense layer before the output. Conv1D layer identifies and extracts the feature of the time series, Maxpooling1D obtains the hidden information and reduces the dimensionality of the data and LSTM extracts the time dependence in the data. So, CNN encodes the features of the time series data, and LSTM decodes these features to determine the connection between data timing and nonlinearity (Zhang and Li, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Huang and Kuo, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2018\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eSince it is a univariate time series model, the total number of features was set to 1. The parameters settings and output shapes in the model can be found in Table S2 and Table S3 from supplementary material. Percentage Mean Absolute Error (MAE), percentage root mean squared error (RMSE) with respect to the observed mean, mean absolute percentage error (MAPE) and coefficient of determination (R\u003csup\u003e2\u003c/sup\u003e) were calculated to evaluate the model performance. It is to be noted that the lag was set to one timestep (i.e. one day) in our prediction experiments, as the CNN-LSTM model typically exhibits some degree of lag as suggested by Zhang and Li, (\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"3. Results and Discussion","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Climatology and Annual Cycles of VC\u003c/h2\u003e \u003cp\u003eThe spatial climatological (1980\u0026ndash;2019) patterns of seasonal mean VC are presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e (a-h). VC has been classified into six categories based on the air pollution potential index over India (e.g. Kannemadugu et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Kannemadugu et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Very high pollution potential occurs when VC is between 0\u0026ndash;2000 m\u003csup\u003e2\u003c/sup\u003e/s, high pollution potential occurs when VC is between 2000\u0026ndash;4000 m\u003csup\u003e2\u003c/sup\u003e/s and medium pollution potential occurs when VC is between 4000\u0026ndash;6000 m\u003csup\u003e2\u003c/sup\u003e/s. Low pollution might be considered when VC exceeds 6000 m\u003csup\u003e2\u003c/sup\u003e/s, which is not usually found over the Indian subcontinent, except few locations. During winter (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea, \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ee), extremely low VC (\u0026lt;\u0026thinsp;1000 m\u003csup\u003e2\u003c/sup\u003e/s) values are noticeable in most of the Indian regions, indicating very high pollution potential occurs during this season. This might be because of low (decreasing) PBLH and weak winds throughout this season, which are responsible for the air stagnation conditions resulting in poor air quality (Kannemadugu et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Iyer and Raj, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). Factors such as low PBLH, less solar insolation and cold temperature might also cause the trapping of air pollutants (e.g. Ramachandran et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). Further, low temperature during winter leads to denser air, because of which the vertical mixing is less, giving rise to lower PBLH and therefore poor dispersion of pollutants. Some regions of south India especially the coast of Tamil Nadu show a slightly high VC. These patterns are well depicted by IMDAA as compared to ERA5. During pre-monsoon (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb, \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ef), better dispersive capacity is observed over western parts of the country, indicating the suitability of setting up of new industries in west India during this season. These changes from winter to pre-monsoon could be attributed to elevated temperature that helps in deepening the PBLH (e.g. Ramachandran et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). The pre-monsoon conditions like high solar insolation give rise to high temperatures causing vertical mixing that leads to an increase in the PBLH resulting in the dispersion of the pollutants thereby good air quality. Furthermore, land surface response such as dry soil conditions and high surface sensible heat flux could cause an increase in the PBLH (e.g. Budakoti and Singh, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) and this can result in increased VC. During monsoon (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec, \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eg), western and southern India exhibits high to medium pollution potential ranges, these regions act as ventilation corridors during this season whereas other parts of the country show poor VC values, indicating the possibility of poor air quality. The PBLH is low during this season for most parts of the country because of high cloud cover, precipitation and strong wind speed (Budakoti and Singh, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Kannemadugu et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Nizar and Dodamani, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). However, the PBLH is high in west India and it increases gradually from north to south India, this may be because of the transfer of heat from the Bay of Bengal to northwest India, travelling over the monsoon winds that cause an increase in PBLH (e.g. Ramachandran et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). Strong convection during the season may also give rise to a better ventilation of air. IMDAA (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec) slightly overestimated these VC patterns, especially over the southern parts of India. During post-monsoon (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ed, \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eh), most of the country exhibits lower values of VC (below 2000 m\u003csup\u003e2\u003c/sup\u003e/s) suggesting lower ventilation prevails, as in winter. This may be attributed to low wind speed during this season over the Indian region. Less insolation of the earth\u0026rsquo;s surface may also be responsible for low PBLH during post-monsoon, and this forms an inversion layer which opposes the vertical mixing of air pollutants (e.g. Ramachandran et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). It is to be noted that VC over the Andhra Pradesh and Tamil Nadu appear to be slightly affected by northeast monsoon and increased wind speed due to storm and cyclonic activities. IMDAA (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eh) realistically represents the spatial patterns of the VC during the post monsoon. The spatial patterns of VC during the post-monsoon season are almost consistent with Kannemadugu et al. (\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) and Kannemadugu et al. (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe spatial climatological distribution of the daily standard deviation of VC can be seen in Figure. 4 (a-h), a high variability in VC is observed during pre-monsoon and monsoon, especially over the Indo-Gangetic Plains (IGP), west and south Indian region. During winter, all parts of the country report relatively lower variability (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea, \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee). Some high variability pockets are noticeable in the eastern parts of Ladakh (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea, \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee). During pre-monsoon, the higher standard deviation was noticed over western India and central India with IGP, northwest Rajasthan and eastern Ladakh showing the highest variability (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb, \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ef) with slightly higher VC observed by IMDAA (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb) than ERA5 (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ef). During monsoon, central, western and southern India shows very high variability, with IGP, Rajasthan and some parts of southeast India showing the highest variability (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec, \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eg). On the other hand, low variability was observed at all parts of the country (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed, \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eh), except for eastern Ladakh for IMDAA during the post-monsoon season (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed). Among IMDAA and ERA5, much higher values of VC and standard deviation of VC were observed in IMDAA than in ERA5 but the pattern looks almost similar in both, based on Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e and \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e. So, from the analysis, it is clear that IMDAA can resolve some of the features of ERA5.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe monthly mean VC climatology for five sub-regions of India during 1980\u0026ndash;2019 is illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e (a-e). The maximum VC values, with higher variability, are noticeable during the pre-monsoon and monsoon seasons whereas low VC (relatively lower variability) during the post-monsoon and winter seasons. The highest VC was observed in west India (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ec) followed by south (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ee), central (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eb), east (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ed) and north India (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea). In north India, the highest VC was observed in May (ERA5) and June (IMDAA) and the lowest VC was observed in January (ERA5) and November (IMDAA) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea). In central India, the highest VC was observed in May and the lowest VC was observed in October (ERA5) and December (IMDAA) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eb). In west India, the highest VC was observed in June and the lowest VC was observed in October (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ec). In east India, the highest VC was observed in April (ERA5) and May (IMDAA) and the lowest VC was observed in October (ERA5) and September (IMDAA) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ed). In south India, the highest VC was observed in July and the lowest VC was observed in October (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ee). The exact VC values for each case have been given in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. The same analysis was done over city scale, where a metropolitan city was picked from each of the five sub-regions of India (North India \u0026ndash; Chandigarh, Central India \u0026ndash; Nagpur, West India \u0026ndash; Jodhpur, East India \u0026ndash; Kolkata and South India - Bangalore). Figure\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e (f-j) depicts the annual cycle at city scale, it can be seen that the trend in each city follows the same trend as their respective regions, only the magnitude of VC is higher. The maximum and minimum VC values for each case have been given in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. Overall, IMDAA follows the same annual cycles as ERA5, only the magnitude of VC is higher in IMDAA and, in some instances, the peak is slightly shifted compared to ERA5.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eMaximum and Minimum values of VC (m\u003csup\u003e2\u003c/sup\u003e/s) for the annual cycle of each Indian sub-region for the period 1980\u0026ndash;2019.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eData source Vs Sub-regions\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eERA5\u003c/p\u003e \u003cp\u003eMaximum VC in m\u003csup\u003e2\u003c/sup\u003e/s (month)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eERA5\u003c/p\u003e \u003cp\u003eMinimum VC in m\u003csup\u003e2\u003c/sup\u003e/s (month)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eIMDAA\u003c/p\u003e \u003cp\u003eMaximum VC in m\u003csup\u003e2\u003c/sup\u003e/s (month)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eIMDAA\u003c/p\u003e \u003cp\u003eMinimum VC in m\u003csup\u003e2\u003c/sup\u003e/s (month)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eNORTH\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e336.08\u0026thinsp;\u0026plusmn;\u0026thinsp;70.89\u003c/p\u003e \u003cp\u003e(May)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e158.40\u0026thinsp;\u0026plusmn;\u0026thinsp;26.92\u003c/p\u003e \u003cp\u003e(Jan)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e553.52\u0026thinsp;\u0026plusmn;\u0026thinsp;284.52 (June)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e118.56\u0026thinsp;\u0026plusmn;\u0026thinsp;50.32\u003c/p\u003e \u003cp\u003e(Nov)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCENTRAL\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1729.30\u0026thinsp;\u0026plusmn;\u0026thinsp;388.72 (May)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e347.27\u0026thinsp;\u0026plusmn;\u0026thinsp;84.76\u003c/p\u003e \u003cp\u003e(Oct)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2379.63\u0026thinsp;\u0026plusmn;\u0026thinsp;763.37 (May)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e276.49\u0026thinsp;\u0026plusmn;\u0026thinsp;57.75\u003c/p\u003e \u003cp\u003e(Dec)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eWEST\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3972.82\u0026thinsp;\u0026plusmn;\u0026thinsp;803.29 (June)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e637.88\u0026thinsp;\u0026plusmn;\u0026thinsp;176.53\u003c/p\u003e \u003cp\u003e(Oct)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4893.97\u0026thinsp;\u0026plusmn;\u0026thinsp;834.27 (June)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e444.39\u0026thinsp;\u0026plusmn;\u0026thinsp;242.80\u003c/p\u003e \u003cp\u003e(Oct)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eEAST\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e619.33\u0026thinsp;\u0026plusmn;\u0026thinsp;108.65\u003c/p\u003e \u003cp\u003e(April)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e172.83\u0026thinsp;\u0026plusmn;\u0026thinsp;45.87\u003c/p\u003e \u003cp\u003e(Oct)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e783.22\u0026thinsp;\u0026plusmn;\u0026thinsp;150.12\u003c/p\u003e \u003cp\u003e(May)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e135.17\u0026thinsp;\u0026plusmn;\u0026thinsp;77.68\u003c/p\u003e \u003cp\u003e(Sep)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eSOUTH\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2950.36\u0026thinsp;\u0026plusmn;\u0026thinsp;536.03 (July)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e371.61\u0026thinsp;\u0026plusmn;\u0026thinsp;96.38\u003c/p\u003e \u003cp\u003e(Oct)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3607.12\u0026thinsp;\u0026plusmn;\u0026thinsp;672.18 (July)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e379.33\u0026thinsp;\u0026plusmn;\u0026thinsp;159.17\u003c/p\u003e \u003cp\u003e(Oct)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSame as Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e but for city scales for the period 1980\u0026ndash;2019.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eData source Vs Cities\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eERA5\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eERA5\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eIMDAA\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eIMDAA\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eChandigarh\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1042.57\u0026thinsp;\u0026plusmn;\u0026thinsp;445.69\u003c/p\u003e \u003cp\u003e(April)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e187.84\u0026thinsp;\u0026plusmn;\u0026thinsp;99.20\u003c/p\u003e \u003cp\u003e(Aug)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2209.94\u0026thinsp;\u0026plusmn;\u0026thinsp;839.05\u003c/p\u003e \u003cp\u003e(April)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e547.08\u0026thinsp;\u0026plusmn;\u0026thinsp;125.03\u003c/p\u003e \u003cp\u003e(Dec)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eNagpur\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2463.78\u0026thinsp;\u0026plusmn;\u0026thinsp;979.37 (May)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e372.93\u0026thinsp;\u0026plusmn;\u0026thinsp;92.32\u003c/p\u003e \u003cp\u003e(Jan)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3985.45\u0026thinsp;\u0026plusmn;\u0026thinsp;947.22 (June)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e507.43\u0026thinsp;\u0026plusmn;\u0026thinsp;202.71\u003c/p\u003e \u003cp\u003e(Oct)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eJodhpur\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4683.14\u0026thinsp;\u0026plusmn;\u0026thinsp;1063.41 (June)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e460.42\u0026thinsp;\u0026plusmn;\u0026thinsp;284.27\u003c/p\u003e \u003cp\u003e(Feb)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2209.44\u0026thinsp;\u0026plusmn;\u0026thinsp;449.21 (July)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e546.13\u0026thinsp;\u0026plusmn;\u0026thinsp;208.31\u003c/p\u003e \u003cp\u003e(Oct)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eKolkata\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2404.52\u0026thinsp;\u0026plusmn;\u0026thinsp;584.69\u003c/p\u003e \u003cp\u003e(April)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e319.98\u0026thinsp;\u0026plusmn;\u0026thinsp;174.45\u003c/p\u003e \u003cp\u003e(Oct)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3962.14\u0026thinsp;\u0026plusmn;\u0026thinsp;661.50\u003c/p\u003e \u003cp\u003e(April)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e346.86\u0026thinsp;\u0026plusmn;\u0026thinsp;231.50\u003c/p\u003e \u003cp\u003e(Oct)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eBangalore\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3584.88\u0026thinsp;\u0026plusmn;\u0026thinsp;565.97 (July)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e378.76\u0026thinsp;\u0026plusmn;\u0026thinsp;193.28\u003c/p\u003e \u003cp\u003e(Oct)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4862.94\u0026thinsp;\u0026plusmn;\u0026thinsp;1053.11 (July)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e655.12\u0026thinsp;\u0026plusmn;\u0026thinsp;404.15\u003c/p\u003e \u003cp\u003e(Oct)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Diurnal Variability\u003c/h2\u003e \u003cp\u003eThe anthropogenic emissions of various aerosols are increasing day by day over India due to mass movement of people from rural to urban areas, degrading the AQ in urban areas (e.g. Abiye et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Therefore, it is important to study the diurnal pattern of VC to understand the air quality patterns throughout the day. Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e (a-e) depicts diurnal variations in VC, showing high and variable levels during the day (10:00 h to 18:00 h) and low, stable levels at night and in the morning (19:00 h to 09:00 h). VC is negatively correlated with air pollutants (e.g., particulate matter and black carbon), indicating an inverse relationship (Budakoti and Singh, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Chan et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Sujatha et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Daytime emissions, driven by anthropogenic activities, are dispersed due to high solar radiation, PBLH, warm temperatures, and wind speed (Karuna et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Kumar, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Moreira et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Priyanka et al., \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Conversely, at night and early morning, lower wind speed, reduced PBLH, and colder temperatures contribute to pollutant accumulation in the planetary boundary layer, resulting in elevated pollution levels (Kumar, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). This suggests that emissions, dispersed during the day, settle at night due to a stable atmosphere, leading to increased pollution levels during this period. VC begins to rise from 09:00 h until 15:00 h (local time), after which it begins to decline (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). The highest VC was recorded in west India (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ec) (IMDAA \u0026ndash; 7706\u0026thinsp;\u0026plusmn;\u0026thinsp;4900 m\u0026sup2;/s, ERA5\u0026ndash;5214\u0026thinsp;\u0026plusmn;\u0026thinsp;3459 m\u0026sup2;/s at 16:00 h), while the lowest VC was observed in north India (IMDAA \u0026ndash; 151\u0026thinsp;\u0026plusmn;\u0026thinsp;273 m\u0026sup2;/s, ERA5\u0026ndash;29\u0026thinsp;\u0026plusmn;\u0026thinsp;37 m\u0026sup2;/s at 05:00 h). The diurnal variation of VC shows the maximum (daytime) and minimum (nighttime) magnitudes, indicating greater pollutant dispersion during the daytime than at nighttime. The diurnal pattern of VC in north India (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea) matches with the same study conducted over the city of Bareilly in Uttar Pradesh by Karuna et al., (2019). These diurnal variations in IMDAA agree well with ERA5 in all sub-regions of the Indian sub-continent. Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e (f-j) shows the diurnal patterns of VC from a metropolitan city from each sub-region. It can be seen that the diurnal pattern at the city scale (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ef-\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ej) matches with the patterns of VC from their respective sub-regions (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea-\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ee). VC is high and variable during the day while low and invariable at night and morning. Overall IMDAA follows the same pattern of VC like ERA5 only the magnitude is higher, except for the Jodhpur (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eh). For IMDAA, the highest VC was observed at Chandigarh (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea) with 8461\u0026thinsp;\u0026plusmn;\u0026thinsp;7216 m\u003csup\u003e2\u003c/sup\u003e/s at 15:00h and the lowest VC was observed at Jodhpur (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eh) with 743\u0026thinsp;\u0026plusmn;\u0026thinsp;976 m\u003csup\u003e2\u003c/sup\u003e/s at 06:00h. For ERA5, the highest VC was observed at Jodhpur (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eh) with 6882\u0026thinsp;\u0026plusmn;\u0026thinsp;4996 m\u003csup\u003e2\u003c/sup\u003e/s at 16:00h and the lowest VC was observed at Chandigarh (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea) with 198\u0026thinsp;\u0026plusmn;\u0026thinsp;526 m\u003csup\u003e2\u003c/sup\u003e/s at 04:00h.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eIt is important to note that ERA5 and IMDAA have coarser resolutions, and data with higher resolution will be more useful when we focus on looking at the city scale. However, due to the limited spatial and temporal coverage of ground-based observations, it is difficult to look at the continuous long-term (40 years in our case) trends of VC. From Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, it can be seen that a lot of studies have focused on VC but for a short time. On the other hand, out study (long-term analysis) provides the opportunity to extrapolate future VC (and AQ) patterns for extended period.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e3.3 Spatial and Temporal Trends\u003c/h2\u003e \u003cp\u003eStudying the VC to identify ventilation corridors (regions with low pollution potential) and suitable locations for new industries is important. It also helps to assess the damage caused by existing industries and determine the locations that are most vulnerable to the pollution plume (Abiye et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). The VC varies across India, depending on atmospheric and land surface conditions. Additionally, VC can also be influenced by regional topography, with areas near coastlines or mountainous regions having higher ventilation coefficients due to the influence of winds and the roughness of the terrain (e.g. Kannemadugu et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). A better understanding of the spatial and seasonal trends of the VC over India can help improve our understanding of air pollution mitigation strategies. In this analysis, we checked the VC trend in India throughout the 40 years for each grid point. We estimated the slope and p-value of VC using linear regression. We then observed the spatial pattern of the slope and its significance at a 95% confidence interval.\u003c/p\u003e \u003cp\u003eAs shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e (a-h), both IMDAA and ERA5 exhibit a similar trend during pre-monsoon and monsoon, but IMDAA (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eb, \u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003ec) captures high-resolution features in some areas not shown by ERA5 (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003ef, \u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eg). During winter, IMDAA and ERA5 show a similar slope pattern, except in the IGP region, where IMDAA shows an increasing trend and ERA5 shows a decreasing trend (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003ea, \u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003ee). Decreasing trends in VC were found over some parts of central, west, south and northeast India and also in the state of Ladakh, with significant values over the IGP region. During pre-monsoon, a decreasing trend has been found in the central, west and south Indian regions whereas an increasing trend has been found in the north and some parts of northeast India, with significant values occurring mostly over the central and northeast India region (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eb, \u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003ef). During monsoon, north India and IGP region is showing an increasing trend whereas the rest of the country is showing a decreasing trend in VC, with significant values over west, central and northeast Indian regions (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003ec, \u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eg). The declining trend in VC during the pre-monsoon and monsoon seasons may be linked to a decrease in PBLH during these periods (e.g., Budakoti and Singh, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). In the post-monsoon season, an increasing trend in VC is observed over the western Ghats, west, north, and northeast Indian regions, while the remaining regions show a decreasing trend. The most significant values are concentrated in central and east India (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eh). During post-monsoon (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003ed, \u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eh), IMDAA and ERA5 exhibit a similar slope pattern, except in the eastern part of IGP.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe temporal trends in ERA5 and IMDAA are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e over different Indian sub-regions. The analysis clearly indicates a decreasing trend, with some of the regions in different seasons also noticing increasing VC trends. IMDAA and ERA5 agree with each other during all seasons, with some discrepancies. Figure\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e (a-t), shows that the IMDAA and ERA5 are capturing a similar pattern or trend, where IMDAA is capturing higher variability as compared to ERA5. Most of the regions are showing a decreasing trend although they are not significant, a significant trend was only observed in central India during pre-monsoon (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eg), and central and west India during monsoon (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003el, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003em). North and west India are the only regions where both IMDAA and ERA5 are showing an increasing trend in VC (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ep, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003er). During winter, the northern, western, and southern regions exhibit a declining trend (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ea, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ec, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ee), while the central and eastern regions show a decreasing trend in ERA5 (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eb, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ed). During pre-monsoon, the central, west, east, and south regions show a decreasing trend (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eg, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eh, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ei, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ej), whereas north India shows ERA5 increasing and IMDAA increasing (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ef). During monsoon, central, west, and south India show a decreasing trend (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003el, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003em, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003en), while east India shows ERA5 increasing and IMDAA decreasing, and north India shows ERA5 decreasing and IMDAA decreasing (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003en, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ek). During post-monsoon, north and west India show an increasing trend (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ep, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003er), south India shows a decreasing trend (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003et), and east and central India show ERA5 decreasing and IMDAA increasing (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003es, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eq).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eIt is noteworthy to mention that unfavorable meteorological conditions for the dispersion of air pollution led to an increase in PM\u003csub\u003e2.5\u003c/sub\u003e concentrations (e.g. Hou et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Better ventilation is more favorable for the dilution and outflow of PM\u003csub\u003e2.5\u003c/sub\u003e, it is also a fact that PM\u003csub\u003e2.5\u003c/sub\u003e should negatively correlate with VC. However, there are incidences with a positive correlation between PM\u003csub\u003e2.5\u003c/sub\u003e and VC, suggesting that ventilation also affects the inflow of PM\u003csub\u003e2.5\u003c/sub\u003e from outside of that region (e.g. Hou et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). The trends (year\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e) in MERRA-2 derived PM\u003csub\u003e2.5\u003c/sub\u003e are given in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e from 1980 to 2019. We can see that PM\u003csub\u003e2.5\u003c/sub\u003e has significantly increased throughout the years in all the sub-regions during all the seasons, showing that particulate pollution has increased. West India during monsoon was the only region where the trends were insignificant. This increase in PM\u003csub\u003e2.5\u003c/sub\u003e through the four decades is desired as there is a huge technological and industrial gap between 1980 and 2019. A decreasing VC can be one of the reasons for the same as noticed in Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e, suggesting an inverse relationship between VC and PM\u003csub\u003e2.5\u003c/sub\u003e. However, the increase in emissions might play a bigger role in the increasing trend of PM\u003csub\u003e2.5\u003c/sub\u003e than VC, as trends of VC were insignificant and inconsistent. Kannemadugu et al. (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2023\u003c/span\u003e) also brought attention to the long-term (1980\u0026ndash;2019) VC and PM2.5 trends on a global scale, indicating a decrease in VC over the Indian subcontinent due to a downward trend in PBLH and wind speed across the region. The study also reported an increase in PM\u003csub\u003e2.5\u003c/sub\u003e concentrations over Indian regions, with the Indo-Gangetic Plain (IGP) showing the highest accumulation rate of PM\u003csub\u003e2.5\u003c/sub\u003e. Our findings align closely with those of the study. For short-term studies, MERRA-2 have been validated with ground-based observations over multiple cities in India (Chimurka et al., 2020; Bali et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), although long-term validation of MERRA-2 is unfeasible due to limited temporal coverage of the in-situ PM2.5 data.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eTemporal trend (slope, p-value) of MERRA2 PM\u003csub\u003e2.5\u003c/sub\u003e (\u0026micro;g/m\u003csup\u003e3\u003c/sup\u003e/year) for the period of 1980\u0026ndash;2019. Trends are statistically significant at a 95% confidence level. The p-value is only mentioned when the trend is insignificant.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSeason Vs Sub-region\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDJF\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMAM\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eJJAS\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eON\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eNORTH\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.58\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.78\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCENTRAL\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;1.45\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.59\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.38\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;1.55\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eWEST\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.24\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.14\u003c/p\u003e \u003cp\u003e\u003cb\u003ep-value\u0026thinsp;=\u0026thinsp;0.06\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;1.04\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eEAST\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;1.60\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;1.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.34\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;1.02\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eSOUTH\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.49\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.23\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.64\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e3.6 Interannual Variability\u003c/h2\u003e \u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e (a-e) shows the interannual variability of VC for the period of 1980 to 2019. Here, normalization of VC was performed for each sub-region of India, for both ERA5 and IMDAA data and correlation coefficient (R) was computed to examine the fidelity of IMDAA (with ERA5). From Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e, it can be seen that ERA5 and IMDAA follow a similar pattern for all the sub-regions. R greater than 0.8 was observed over central, west, east and south India (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eb, \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003ec, \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003ed, \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003ee) whereas R was 0.57 over north India (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003ea). This indicates less correlation between ERA5 and IMDAA over north India as compared to the other sub-regions, with all the regions showing a significant correlation. It is evident that the VC has decreased over the four decades, the decrease is more prominent following the year 2000 in all the sub-regions, especially in central, west, east and south India as negative values are more clustered after 2000 (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eb, \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003ec, \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003ed, \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003ee). In north India there is no significant VC trend (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003ea). Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e makes it clearer, showing the slope values for the trends in each sub-region. The highest variability was found in south India (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003ee) as the frequency of occurrence of negative or positive VC values in consecutive years is less than in the other regions. For ERA5, the highest VC was found in west India during 1987 (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003ec), which was a strong El Ni\u0026ntilde;o year. For IMDAA, the highest VC was found in east India during 1992 (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003ed), which was a year following a strong El Ni\u0026ntilde;o year (1991). The lowest VC was found in south India during 2010 for both ERA5 and IMDAA (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003ee), this was a strong La Ni\u0026ntilde;a year.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab8\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eAnnual mean trend (slope) of VC (m\u003csup\u003e2\u003c/sup\u003e/s/year) for the period of 1980\u0026ndash;2019. Trends are statistically significant at a 95% confidence level. The p-value is only mentioned when the trend is insignificant.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNORTH\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCENTRAL\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eWEST\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eEAST\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSOUTH\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eIMDAA\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.00\u003c/p\u003e \u003cp\u003e\u003cem\u003ep-value\u003c/em\u003e\u0026thinsp;\u003cb\u003e=\u0026thinsp;0.96\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003em = -0.04\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003em = -0.03\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003em = -0.03\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003em = -0.03\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eERA5\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003em\u0026thinsp;=\u0026thinsp;0.00\u003c/p\u003e \u003cp\u003e\u003cem\u003ep-value\u003c/em\u003e\u0026thinsp;\u003cb\u003e=\u0026thinsp;0.88\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003em = -0.04\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003em = -0.05\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003em = -0.03\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003em = -0.01\u003c/p\u003e \u003cp\u003e\u003cem\u003ep-value\u003c/em\u003e\u0026thinsp;\u003cb\u003e=\u0026thinsp;0.48\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e3.7 Linkages with El Ni\u0026ntilde;o-Southern Oscillation\u003c/h2\u003e \u003cp\u003eThe El Ni\u0026ntilde;o-Southern Oscillation (ENSO) is an air-sea interaction and dominant tropical climate driver phenomenon that can have a significant influence on atmospheric circulation and weather patterns all over the globe, including in India. ENSO is characterized by the oscillation of sea surface temperature in the Pacific Ocean, which can alter atmospheric circulation and modulate global weather patterns (Yang et al., \u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). The influence of ENSO patterns on air pollution in the emerging most polluted country, India, has received much less attention, as it is complex and completely dependent on a multitude of factors. The concise impact of ENSO on the ventilation coefficient, however, will be determined by the specific conditions in the region at the time. ENSO has a significant impact on air quality in the southern regions of China (Wang et al., \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Overall, the influence of ENSO on the characteristics of ventilation coefficients over India across the seasons is an important area of research that can help improve our understanding of atmospheric dynamics in the region and inform air pollution mitigation strategies.\u003c/p\u003e \u003cp\u003eIn this work, VC spatial patterns for every season were observed for the past five strong El Ni\u0026ntilde;o occurrences (1982, 1987, 1991, 1997 \u0026amp; 2015) and for the past five strong La Ni\u0026ntilde;a events (1988, 1999, 2000, 2007 \u0026amp; 2010) as well. To examine the impact of these events, the difference between VC of El Ni\u0026ntilde;o and VC of La Ni\u0026ntilde;a was observed, and its seasonal climatology was plotted along with the significant values (p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05) as seen in Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003e (a-h). During winter, El Ni\u0026ntilde;o influences positively (good air quality) over some parts of south India and Maharashtra. In the case of pre-monsoon season (Figs.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003eb, \u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003ef), the lowest VC values are noticed over the north, northwest and some central parts of the country, which indicates stronger El Ni\u0026ntilde;o conditions lead to poor air quality during the season, which is also well depicted by IMDAA. A high VC was observed over Bihar and southwest coast of India (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003eb, \u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003ef), with significant values over IGP, north India, Gujarat, Odisha, West Bengal and east coast of Tamil Nadu (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003ef). From the La Ni\u0026ntilde;a perspective, we observe higher VC (more negative values) over the north, northwest, and some central parts of India, indicating that strong La Ni\u0026ntilde;a events lead to improved air quality in these regions during the pre-monsoon season. A recent study by Beig et al. (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) has established a correlation between strong La Ni\u0026ntilde;a events and PM\u003csub\u003e2.5\u003c/sub\u003e concentrations in India. Their findings suggest that the strong La Ni\u0026ntilde;a event of 2022\u0026ndash;2023 resulted in better air quality in north India but worsened air quality in areas around Mumbai. While our analysis supports their results on a spatial scale, there is a discrepancy in timing; we observed improved air quality due to La Ni\u0026ntilde;a over north India during the pre-monsoon season, whereas the reported better air quality in the same region was during the winter of 2022-23 (Beig et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Therefore, there is a need to further explore the role of the effect of ENSO on air quality in India to better understand these temporal variations and their implications on air quality. Coming back to our analysis, during monsoon, El Ni\u0026ntilde;o conditions improve the air quality by deepening the PBLH. Moreover, a significantly higher VC was observed over most parts of the country, however, the magnitude is highest in west and south India (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003ec, \u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003eg). During post-monsoon, better air pollution potential was observed over west India, especially in the states of Gujarat and Rajasthan (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003ed, \u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003eh). The significant values were found in clusters over north, central, south and east India (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003eh). From Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003e, IMDAA consistently exhibits significant values throughout all seasons across nearly all regions of the country, while in contrast, ERA5 does not show as many notable values. These differences in IMDAA could be attributed to high-resolution data which captures more local meteorological conditions.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"4. Prediction of VC using Machine Learning Approach","content":"\u003cp\u003eInformation from the preceding seasons could be important to train the models and predict the VC which is vital for planning air pollution control in India. In this section, we attempted to predict the VC (as a proxy of air pollution indicator) using the CNN-LSTM model, over the city of Delhi. The predictions were performed on ERA5 data. From the previous analysis, we saw that IMDAA was performing better compared to ERA5. However, since IMDAA is not validated and used extensively for VC over the Indian region, we used ERA5 for the machine learning predictions. Figure\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e11\u003c/span\u003e (a-b) shows the daily predicted values of ERA5 VC for the year 2019 and its comparison with the observed values. From Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e11\u003c/span\u003ea, it is evident that the model is capable of capturing VC trends that are very similar to observed values. The predicted values produced MAE, RMSE, MAPE and R\u003csup\u003e2\u003c/sup\u003e scores, which shows the good accuracy of the model. A good (and significant) correlation coefficient (R) score of 0.96 indicates a high linear relationship between predicted and observed values and confirms the capability of the CNN-LSTM model to capture the trends of VC (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e11\u003c/span\u003eb). The seasonal predictions of ERA5 VC can be seen in Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e12\u003c/span\u003e (a-d), it is noticeable that the predictions were found to be reasonably well in all the seasons. Predictions for the monsoon season were most accurate (Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e12\u003c/span\u003ec), this was followed by post-monsoon, pre-monsoon and finally winter (Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e12\u003c/span\u003ed, \u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e12\u003c/span\u003eb, \u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e12\u003c/span\u003ea). For the winter season, it can be seen that the predicted values follow the same pattern as the observed values but it is unable to capture the same magnitude of VC as the observed values (Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e12\u003c/span\u003ea). The same is the case for the pre-monsoon season, although the error (MAE, RMSE and MAPE) is less as compared to winter (Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e12\u003c/span\u003eb). For the monsoon season, the model is capturing almost the same values as the observed (Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e12\u003c/span\u003ec). In the case of post-monsoon as well the predicted values have the same trend as observed values but it is less accurate than the monsoon season (Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e12\u003c/span\u003ed). It is important to understand that given all the conditions, the prediction of VC by considering only one variable (univariate time series) seems to be unconventional. VC is highly dynamic, other meteorological parameters (like temperature, relative humidity, solar radiation) which have strong relationships with VC, can further be used in the prediction of VC (multivariate predictions). Such studies in future can give the reasoning for high or low VC values, and also achieve higher accuracy.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003ePlease note that only ERA5 data was used for the prediction and validation of VC. And ERA5 reanalysis data, while widely used, is not without uncertainties. One significant source of uncertainty lies in the assimilation of observations, where gaps or errors in input data can impact the accuracy of the reanalysis. Additionally, uncertainties in the representation of physical processes and the complexity of Earth's systems contribute to variability in the results. Spatial and temporal resolutions may also introduce uncertainties, especially in regions with sparse observational data. Users of ERA5 should be mindful of these uncertainties, recognizing the need for caution and validation in specific applications requiring high precision and reliability. For example, in our work, winds and PBLH (Planetary Boundary Layer Height) are crucial parameters. These variables undergo validation against observational data. Recently, several studies (e.g., Guo et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Li et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Zhai et al., \u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) have reported a strong correlation between PBLH and winds from ERA5 and observational data. However, it is essential to note that reanalysis fields, being model-driven, inherently introduce some degree of error into the outputs. In this context, machine learning models emerge as a promising alternative (Gibson et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). They can potentially enhance the accuracy of these fields, especially in situations where observational data is limited. However, the use of modelled data (ERA5) to train machine learning models introduces its own set of challenges. The extensive nature of ERA5 data, as observed in our case with large VC data (daily data spanning from 1980 to 2019), presents a few challenges. A larger dataset implies longer training time, and managing such a substantial dataset can be a hindrance without access to significant computational resources. Gibson et al. (\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) emphasized the importance of employing machine learning on large climate models. Their study suggested that simulations from a machine learning model trained on a climate model could yield accurate seasonal precipitation forecasts, potentially outperforming existing climate models.\u003c/p\u003e \u003cp\u003eIt is well known that VC is considered a proxy for air quality. We further aimed to evaluate the performance of CNN-LSTM using ground-based data. As observational data for VC was unavailable, we utilized observations for PM2.5 instead. The study employed continuous PM2.5 observations from the Central Pollution Control Board (CPCB) for four metropolitan Indian cities, namely Delhi, Mumbai, Kolkata, and Bangalore. The data utilized in this step was from 2018 to 2019 for Delhi, Mumbai, Kolkata and from 2017 to 2019 for Bangalore. We used PM2.5 concentration during November and December 2019 for prediction (for testing), while the rest is used for training the model. It is noticeable that the model successfully captured the PM2.5 trend with relatively less error (MAE\u0026thinsp;\u0026lt;\u0026thinsp;0.15), and high accuracy (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.8) for selected four metropolitan cities, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig13\" class=\"InternalRef\"\u003e13\u003c/span\u003e (a-d). Please note that machine learning models produce different outcomes for different datasets and parameters. In this study, we observed that predictions were correlated well with the observed values for both datasets (ERA5 - VC \u0026amp; CPCB \u0026ndash; PM2.5). The reason for this may be that our data is univariate in nature (simpler than multivariate), therefore it is easier for this model to understand the nonlinear relationships in the data. Nevertheless, the strong performance of CNN-LSTM on both ERA5 reanalysis and in situ observations (from CPCB) implies its robust predictive capabilities. This suggests that this model could be used in future studies for forecasting, although, one should be educated and aware of its uncertainties before using it.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eWhile machine learning has shown success in prediction of both VC and PM2.5, there are various challenges while using it. Firstly, the availability and quality of data are important factors, insufficient and complex data may be difficult to work on (Ayturan et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Huang and Kuo., 2018; Zhang and Li., 2022). In such cases accuracy of some models may be limited, however, the problem can be solved by exploring more machine/deep learning algorithms and finding which one works the best for that task. Furthermore, deep learning models like CNN-LSTM require high computational resources (Masood and Ahmad, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2020\u003c/span\u003e), which sometimes may not be available. Therefore, to overcome these limitations it is important to have expert knowledge of machine learning to be utilized properly and efficiently (Kanevski et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2008\u003c/span\u003e).\u003c/p\u003e"},{"header":"5. Summary and Conclusions","content":"\u003cp\u003eWe analyzed 40 years of VC data (1980\u0026ndash;2019) derived from ERA5 and IMDAA reanalysis to study climatology, seasonal variability, diurnal variability, spatial and temporal trends of ventilation coefficient (VC), interannual variability including the effects of ENSO on VC. We also evaluated the performance of the CNN-LSTM model for time series prediction of VC. The study's conclusions are as follows:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eWhile most parts of the country showed very high pollution potential for all the seasons, high to medium pollution potential was observed only in west India during pre-monsoon, and in west and south India during monsoon seasons. All five sub-regions of India are showing high and variable VC during pre-monsoon and monsoon seasons whereas low and stable VC during post-monsoon and winter seasons. Higher values of VC during pre-monsoon season mainly because of high solar radiation which leads to an increase in temperature, further causing stronger vertical mixing and higher PBLH, which results in the dispersion of air pollutants thereby leading to good air quality. Higher values and variability of VC during monsoon season are likely because of high wind speed and convection.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eVariability in VC changes regionally from season to season with higher standard deviation noticed over western India and central India with IGP, northwest Rajasthan and eastern Ladakh showing the highest variability during the pre-monsoon season. Further higher variability is observed in central, western and southern India during the monsoon season.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eA high and variable VC was observed in the morning till the afternoon, while a low and invariable VC was observed in the evening till the early morning.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eA high and significant decreasing trend in VC was observed in the west, central and some parts of south India during pre-monsoon and monsoon seasons. Overall, ERA5 and IMDAA showed similar patterns in spatial trends, with contradictions only in the IGP region during winter and post-monsoon seasons.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eA decreasing trend in VC was observed from 1980\u0026ndash;2019 in all sub-regions during all seasons, except for north India during pre-monsoon, east India during monsoon, and north and west India during post-monsoon.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eOverall, VC has decreased significantly over central, west, east and south India during the study period. And highest VC occurred during the strong El Ni\u0026ntilde;o year (1987) while the lowest VC was observed during the strong La Ni\u0026ntilde;a year (2010).\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eENSO had a significant impact on VC over India especially during pre-monsoon and monsoon seasons. El Ni\u0026ntilde;o caused low VC during pre-monsoon season over northwest, central and north India, and high VC during monsoon season over west, central and south Indian regions.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003ePerformance of the CNN-LSTM in predicting the VC is fairly well for each season and annual scale. Furthermore, CNN-LSTM is also able to capture the PM\u003csub\u003e2.5\u003c/sub\u003e over major cities.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eOverall, our findings from this long-term study could help identify ventilation corridors, which are spots with a high potential for pollutant dispersion, as well as suitable locations for new industrial operations. To summarize this study by comprehensive and region-specific approach while considering both seasonal variations and local influencing factors is crucial for effective air quality management in India.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgment\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors wish to thank ERA5 (https://rds.ncmrwf.gov.in/datasets) and IMDAA (https://cds.climate.copernicus.eu/#!/search?text=ERA5\u0026amp;type=dataset) database for their open access to the PBLH and wind speed (u and v component) data. We would also like to thank MERRA2 (https://disc.gsfc.nasa.gov/datasets/M2_TMAX_PM25_1/summary?keywords=PM2.5), and Central Pollution Control Board (CPCB) (https://airquality.cpcb.gov.in/ccr/#/caaqm-dashboard-all/caaqm-landing) for their open access to PM\u003csub\u003e2.5\u003c/sub\u003e data. Climate Data Operators was used for data processing and python programming language was used for data analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNo funding was received for conducting this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor information\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors and Affiliations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDepartment of Earth and Environmental Sciences, Indian Institute of Science Education and Research Mohali, Mohali, Punjab, India: Amitabha Govande, Raju Attada, Krishna Kumar Shukla, Soumya Muralidharan, Garima Kaushik\u003c/p\u003e\n\u003cp\u003eCentre for Atmospheric Sciences, Indian Institute of Technology Delhi, Delhi, India, 110016: \u0026nbsp;Ravi Kumar Kunchala\u003c/p\u003e\n\u003cp\u003eDepartment of Earth and Atmospheric Sciences, National Institute of Technology, Rourkela \u0026ndash; 769008, India: Nagaraju Chilukoti\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contribution:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAG and RA conceptualized the problem, perform the analysis and wrote the manuscript. KKS and SM prepared the tables and review the manuscript. RKK, NC and GK provided the scientific inputs on the discussions and edited the manuscript. All authors contributed toward the discussions and interpretation of the results.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of competing interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.Ethics declarations\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll the datasets used in the study are publicly available.\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAbiye OE, Akinola OE, Sunmonu LA, Ajao AI, Ayoola MA (2016) Atmospheric ventilation corridors and coefficients for pollution plume released from an Industrial Facility in lle-lfe Suburb. Nigeria Afr J Environ Sci Technol 10(10):338\u0026ndash;349\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAllabakash S, Lin S (2020) Climatology of Planetary Boundary Layer Height-Controlling Meteorological Parameters Over the Korean Peninsula. Remote Sens 12:2571\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAl-Salem SM, Khan AR (2010) Monitoring and Modelling the Trends of Primary and Secondary Air Pollution Precursors: The Case of the State of Kuwait. Int J Chem Eng. 879836\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eApte J, Marshall J, Cohen AJ, Brauer M (2015) Addressing Global Mortality from Ambient PM2.5. Environ Sci Technol 49:8057\u0026ndash;8066\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAyturan YA, Ayturan ZC, Altun HO, Kongoli C, Tuncez FD, Dursun S, Ozturk A (2020) Short-term prediction of PM2.5 pollution with deep learning methods. Global NEST J 22(1):126\u0026ndash;131\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBabu SS, Manoj MR, Moorthy KK, Gogoi MM, Nair VS, Kompalli SK, Satheesh SK, Niranjan K, Ramagopal K, Bhuyan PK, Singh D (2013) Trends in aerosol optical depth over Indian region: Potential causes and impact indicators. J Geophys Res Atmos. 118(20). 11,794\u0026ndash;11,806.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBali K, Dey S, Ganguly D (2021) Diurnal patterns in ambient PM2.5 exposure over India using MERRA-2 reanalysis data. Atmos Environ 248:118180\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBeig G, Anand V, Korhale N, Sobhana SB, Harshitha KM, Kripalani RH (2024) Triple dip La-Nina, unorthodox circulation and unusual spin in air quality of India. Sci Total Environ Volume 920:170963\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBran SH, Srivastava R (2017) Investigation of PM\u003csub\u003e2.5\u003c/sub\u003e mass concentration over India using a regional climate model. Environ Pollut 224:484\u0026ndash;493\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBudakoti S, Singh C (2021) Examining the characteristics of planetary boundary layer height and its relationship with atmospheric parameters over Indian sub-continent. Atmos Res 264:105854\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCastelli M, Clemente FM, Popovic A, Silva S, Vanneschi L (2020) A Machine Learning Approach to Predict Air Quality in California. Complexity. Volume 2020: 8049504\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChan L, Qi-hong D, Wei-wei L, Bo-liang H, Ling-zhi S (2012) Characteristics of ventilation coefficient and its impact on urban air pollution. J Cent South Univ 19:615\u0026ndash;622\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChicco D, Warrens MJ, Jurman G (2021) The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation. PeerJ Comput Sci 7:e623\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChimurkar N, Vinoj V, Pandey SK (2020) Evaluation of PM2.5 Surface Concentrations Simulated by NASA\u0026rsquo;s MERRA Version 2 Aerosol Reanalysis over India and its Relation to the Air Quality Index. Aerosol Air Qual Res 20:1329\u0026ndash;1339\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChowdhary P, He Q, Male TL, Brune WH, Rudich Y, Pardo M (2018) Exposure of Lung Epithelial Cells to Photochemically Aged Secondary Organic Aerosol Shows Increased Toxic Effects. Environ Sci Technol Lett 5:424\u0026ndash;430\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDey S, Di Girolamo L, van Donkelaar A, Tripathi SN, Gupta T, Mohan M (2012) Variability of outdoor fine particulate (PM2.5) concentration in the Indian Subcontinent: A remote sensing approach. Remote Sens Environ 127:153\u0026ndash;161\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDey S, Purohit B, Balyan P, Dixit K, Bali K, Kumar A, Imam F, Chowdhury S, Ganguly D, Gargava P, Shukla VK (2020) Satellite-Based High-Resolution (1-km) Ambient PM\u003csub\u003e2.5\u003c/sub\u003e Database for India over Two Decades (2000\u0026ndash;2019): Applications for Air Quality Management. Remote Sens 12(23):3872\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGao M, Sherman P, Song S, Yu Y, Wu Z, McElroy MB (2019) Seasonal prediction of Indian wintertime aerosol pollution using the ocean memory effect. Sci Adv 5:eaav4157\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGelaro R, McCarty W, Suarez MJ, Todling R, Molod A, Takacs L, Randles CA, Darmenov A, Bosilovich MG, Reichle R, Wargan K, Coy L, Cullather R, Draper C, Akella S, Buchard V, Conaty A, Da Silva AM, Gu W, Kim G, Koster R, Lucchesi R, Merkova D, Nielsen JE, Partyka G, Pawson S, Putman W, Rienecker M, Schubert SD, Sienkiewicz M, Zhao B (2017) The Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA2). J Clim 30:5419\u0026ndash;5454\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGibson PB, Chapman WE, Altinok A, Monache LD, DeFlorio MJ, Waliser DE (2021) Training machine learning models on climate model output yields skillful interpretable seasonal precipitation forecasts. Commun Earth Environ 2:159\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGufran Beig V, Anand N, Korhale SB, Sobhana KM, Harshitha RH, Kripalani (2024) Triple dip La-Nina, unorthodox circulation and unusual spin in air quality of India. Sci Total Environ Volume 920:170963\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuo J, Zhang J, Yang K, Liao H, Zhang S, Huang K, Lv Y, Shao J, Yu T, Tong B, Li J, Su T, Yim SHL, Stoffelen A, Zhai P, Xu X (2021) Investigation of near-global daytime boundary layer height using high-resolution radiosondes: first results and comparison with ERA5, MERRA-2, JRA-55, and NCEP-2 reanalysis. Atmos Chem Phys 21:17079\u0026ndash;17097\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHersbach H, Bell B, Berrisford P, Hirahara S, Hornayi A, Munoz-Sabater J, Nicolas J, Peubey C, Radu R, Schepers D, Simmons A, Soci C, Abdalla S, Abellan X, Balsamo G, Bechtold P, Biavati G, Bidlot J, Bonavita M, Chiara G, Dahlgren P, Dee D, Diamantakis M, Dragani R, Flemming J, Forbes R, Fuentes M, Geer A, Haimberger L, Healy S, Hogan RJ, Holm E, Janiskova M, Keeley S, Laloyaux P, Lopez P, Lupu C, Radnoti G, Rosnay P, Rozum I, Vamborg F, Villaume S, Thepaut J (2020) The ERA5 global reanalysis. Q J R Meteorol Soc 146:1999\u0026ndash;2049\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHolzworth GC (1967) Mixing Depths, Wind Speeds and Air Pollution Potential for Selected Locations in the United States. Journal of Applied Meteorology (1962\u0026ndash;1982). 6(6), 1039\u0026ndash;1044\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHou X, Fei D, Kang H, Zhang Y, Gao J (2018) Seasonal statistical analysis of the impact of meteorological factors on fine particle pollution in China in 2013\u0026ndash;2017. Nat Hazards 93:677\u0026ndash;698\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang C, Kuo P (2018) A Deep CNN-LSTM Model for Particulate Matter (PM2.5) Forecasting in Smart Cities. Sensors 2018. 18, 2220\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIyer US, Raj PE (2012) Ventilation coefficient trends in the recent decades over four major Indian metropolitan cities. J Earth Syst Sci 122(2):537\u0026ndash;549\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJena C, Ghude S, Kumar R, Debnath S, Govardhan G, Soni VK, Kulkarni SH, Beig G, Nanjundiah RS, Rajeevan M (2021) Performance of high resolution (400 m) PM\u003csub\u003e2.5\u003c/sub\u003e forecast over Delhi. Nat Sci Rep 11:4104\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJethva H, Torres O, Field RD, Lyapustin A, Gautam R, Kayetha V (2019) Connecting crop productivity, residue fires, and air quality over northern India. Sci Rep 9:16594\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKalita G, Kunchala RK, Fadnavis S, Kaskaoutis DG (2020) Long term variability of carbonaceous aerosols over Southeast Asia via reanalysis: Association with changes in vegetation cover and biomass burning. Atmos Res 245:105064\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKanevski M, Pozdnukhov A, Timonin V (2008) Machine Learning Algorithms for GeoSpatial Data. Application and Software Tools, vol 53. International Congress on Environmental Modelling and Software\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKannemadugu H, Dorligjav S, Gharai B, Seshasai MVR (2021) Satellite-Based Air Pollution Climatology over India. Water Air Soil Pollution 232:365\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKannemadugu H, Dorligjav S, Taori A, Bothale RV, Chauhan P (2023) Global Air Pollution Potential: long term (1980\u0026ndash;2019) trends and application to sustainable development, PREPRINT (Version 1) available. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.21203/rs.3.rs-2888054/v1]\u003c/span\u003e\u003cspan address=\"10.21203/rs.3.rs-2888054/v1]\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. at Research Square\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaruna MS, Upadhyay OP, Saxena DK, Swamy MM, Majumder S (2017) Air Quality Index, Ventilation Coefficient and Pollution Potential Studies Over Bareilly City, Uttar Pradesh. Int J Curr Res 9(12):63243\u0026ndash;63255\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKumar A, Goyal P (2011) Forecasting of air quality in Delhi using principle component regression technique. Atmospheric Pollution Res 2:436\u0026ndash;444\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKumar DKV (2019) Study of Atmospheric Boundary Layer Height from radiosonde data over a flat terrain at VBIT - Hyderabad (17.4\u0026deg; N \u0026ndash;\u0026thinsp;78.5\u0026deg; E). Int J Appl Eng Res ISSN 14(4):0973\u0026ndash;4562\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKumar N, Parmer K, Soni K, Garg N, Agarwal R (2015) Prediction of ventilation coefficient, using a conjunction model of wavelet-Neuro-fuzzy model: A Case Study Delhi, India. Acad J Sci Res 3(12):184\u0026ndash;191\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi C, McLinden C, Fioletov V, Krotkov N, Carn S, Joiner J, Streets D, He H, Ren X, Li Z, Dickerson RR (2017) India is overtaking China as the World\u0026rsquo;s largest emitter of anthropogenic sulfur dioxide. Sci Rep 7:14304\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi X, Dong Y, Zhang Y, Shi Z, Yao J (2023) Climatology of Planetary Boundary Layer Height over Jiangsu, China, Based on ERA5 Reanalysis Data. Atmosphere 14:1330\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMaheshwarkar P, Sunder Raman R (2021) Population exposure across central India to PM2.5 derived using remotely sensed products in a three-stage statistical model. Sci Rep 11:544\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMasood A, Ahmad K (2020) A model for particulate matter (PM2.5) prediction for Delhi based machine learning approaches. Procedia Comput Sci 167:2101\u0026ndash;2110\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMoreira GA, Guerrero-Rascado JL, Bravo-Aranda JA, Foyo-Moreno I, Cazorla A, Alados I, Lyamani H, Landulfo E, Alados-Arboledas L (2020) Study of the planetary boundary layer height in an urban environment using a combination of microwave radiometer and ceilometer. Atmos Res 240:104932\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNair VS, Moorthy KK, Alappattu DP, Kunhikrishnan PK, George S, Nair PR, Babu SS, Abish B, Satheesh SK, Tripathi SN, Niranjan K, Madhavan BL, Srikant V, Dutt CBS, Badrinath KVS, Reddy RR (2007) Wintertime aerosol characteristics over the Indo-Gangetic Plain (IGP): Impacts of local boundary layer processes and long‐range transport. J Geophys Research: Atmos. 112. D13205\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNizar S, Dodamani BM (2019) Spatiotemporal distribution of aerosols over the Indian subcontinent and its dependence on prevailing meteorological conditions. Air Qual Atmos Health 12:503\u0026ndash;217\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOjha N, Sharma A, Kumar M, Girach I, Ansari TU, Sharma SK, Singh N, Pozzer A, Gunthe SS (2020) On the widespread enhancement in fine particulate matter across the Indo-Gangetic Plain towards winter. Sci Rep 10:5862\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePriyanka S, Soni K, Nair AS, Singh M (2022) Regression analysis of ventilation coefficient at a semi-arid IGP region using forward selection technique. Mausam 73(3):617\u0026ndash;626\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRamachandran S, Kedia S, Srivastava R (2012) Aerosol optical depth trends over different regions of India. Atmos Environ 49:338\u0026ndash;347\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRani SI, Arulalan T, George JP, Rajagopal EN, Renshaw R, Maycock A, Barker DM, Rajeevan M (2021) IMDAA: High-Resolution Satellite-Era Reanalysis for the Indian Monsoon Region. J Clim 34:5109\u0026ndash;5133\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRatnam MV, Prasad P, Raj STA, Raman MR, Basha G (2021) Changing patterns in aerosol vertical distribution over South and East Asia. Sci Rep 11:308\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSaha D, Soni K, Mohanan MN, Singh M (2019) Long-term trend of ventilation coefficient over Delhi and its potential impacts on air quality. Remote Sens Applications: Soc Environ 15:100234\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSujatha P, Mahalakshmi DV, Ramiz A, Rao PVN, Naidu CV (2016) Ventilation coefficient and boundary layer height impact on urban air quality. Cogent Environ Sci 2:1. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/23311843.2015.1125284\u003c/span\u003e\u003cspan address=\"10.1080/23311843.2015.1125284\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShukla KK, Sarangi C, Attada R, Kumar P (2022) Characteristic dissimilarities during high aerosol loading days between western and eastern Indo-Gangetic Plain. Atmos Environ 269:118837\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVenkataraman C, Brauer M, Tibrewal K, Sadavarte P, Ma Q, Cohen A, Chaliyakunnel S, Frostad J, Kilmont Z, Martin RV, Millet DB, Philip S, Walker K, Wang S (2018) Source influence on emission pathways and ambient PM2.5 pollution over India (2015\u0026ndash;2050). Atmos Chem Phys 18:8017\u0026ndash;8039\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang S, Li S, Xing J, Yang J, Dong J, Qin Y, Sahu SK (2022) Evaluation of the Influence of El Nino-Southern Oscillations on air qualityin the southern China from long-term historical observations. Front EnvironmentalScience Eng 16(2):26\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXiao F, Yang M, Fan H, Fan H, Fan G, Al-qaness M (2020) An improved deep learning model for predicting daily PM2.5 concentration. Sci Rep 10:20988\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXie B, Yang Y, Wang P, Liao H (2022) Impacts of ENSO on wintertime PM2.5 pollution over China during 2014\u0026ndash;2021. Atmospheric Ocean Sci Lett 15:100189\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang S, Li Z, Yu J, Hu X, Dong W, He S (2018) El Ni\u0026ntilde;o-Southern Oscillations and its impact in the changing climate. Natl Sci Rev 5:840\u0026ndash;857\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhai R, Huang C, Yang W, Tang L, Zhang W (2022) Applicability Evaluation of ERA5 wind and wave reanalysis data in South China Sea. J Oceanol Limnol 41(2):495\u0026ndash;517\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang J, Li S (2022) Air quality index forecast in Beijing based on CNN-LSTM multi-model. Chemosphere 308:136180\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"theoretical-and-applied-climatology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"taac","sideBox":"Learn more about [Theoretical and Applied Climatology](https://www.springer.com/journal/704)","snPcode":"704","submissionUrl":"https://submission.nature.com/new-submission/704/3","title":"Theoretical and Applied Climatology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Air quality, Ventilation Coefficient, Variability, Trends, CNN-LSTM model","lastPublishedDoi":"10.21203/rs.3.rs-4551619/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4551619/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe concentrations of atmospheric pollutants are a serious concern due to their adverse impacts on human health. The ventilation coefficient (VC) is an indicator that measures the dispersion capacity of air pollutants (air pollution potential) in the atmosphere, providing insights into air quality. In this study, we aim to investigate the spatio-temporal variation and trends of VC over the Indian subcontinent using India’s first high-resolution regional reanalysis (IMDAA) and global reanalysis datasets (ERA5) for the period 1980-2019. The spatial pattern of the seasonal climatological mean ERA5 and IMDAA derived VC shows a lower magnitude during winter and post-monsoon seasons, indicating poor air quality over the Indian region, especially in the northern parts of India. We noticed a gradual declination of VC during different seasons, implying increasing surface-level air pollutants and worsening air quality over India. The study further investigates the changes of VC during strong phases of El Niño and La Niña events. The results reveal that El Niño significantly impacts air quality over northern and western parts of India during pre-monsoon and monsoon seasons. At the diurnal scale, the VC exhibits the highest magnitude and variability during daytime due to increased dispersion of pollutants and higher human activities, while remaining low and stable during night due to stagnant atmospheric conditions. These essential characteristics of VC are well represented in IMDAA, albeit with some discrepancies. Furthermore, we have examined the fidelity of a machine learning model-Convolutional Neural Network and Long Short-Term Memory (CNN-LSTM), in predicting the VC for the year 2019 over Delhi city. Various statistical metrics are computed to evaluate the performance of the CNN-LSTM model. The results confirm that the model successfully predicts the VC compared to observations from ERA5.\u003c/p\u003e","manuscriptTitle":"Analyzing and Predicting Ventilation Coefficient over India using Long-term Reanalysis Datasets and Hybrid Machine Learning Approach ","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-06-24 10:38:29","doi":"10.21203/rs.3.rs-4551619/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-08-25T17:57:32+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-25T09:16:50+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-17T16:46:00+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"286531802201303942078745614674268248378","date":"2025-08-15T18:11:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"174188505414775203441113524498439496450","date":"2025-08-15T15:58:11+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"205689230547277424597444932502070786851","date":"2025-08-15T14:48:23+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"171527860933351750951667172261199546451","date":"2025-08-15T07:44:03+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"306051456351001145400467272994169599095","date":"2025-08-14T10:10:36+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"234855708688822169152981280664692478222","date":"2025-08-14T06:04:31+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"87885163257346021662280540193019020502","date":"2025-08-13T18:36:32+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"261907447568862213255319278687079766892","date":"2025-08-13T11:35:27+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"46202152565549367236188351333831952342","date":"2025-08-13T08:33:44+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"55389615882187434096459921042309143800","date":"2025-08-13T07:23:23+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"170209068184399784585606669901798204891","date":"2025-02-04T11:38:34+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-02-04T10:55:50+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-06-09T22:30:19+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-06-09T22:29:54+00:00","index":"","fulltext":""},{"type":"submitted","content":"Theoretical and Applied Climatology","date":"2024-06-08T18:41:20+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"theoretical-and-applied-climatology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"taac","sideBox":"Learn more about [Theoretical and Applied Climatology](https://www.springer.com/journal/704)","snPcode":"704","submissionUrl":"https://submission.nature.com/new-submission/704/3","title":"Theoretical and Applied Climatology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"cdc39481-62a4-4fb4-aa30-009e5a6f1c36","owner":[],"postedDate":"June 24th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-11-03T16:09:50+00:00","versionOfRecord":{"articleIdentity":"rs-4551619","link":"https://doi.org/10.1007/s00704-025-05845-w","journal":{"identity":"theoretical-and-applied-climatology","isVorOnly":false,"title":"Theoretical and Applied Climatology"},"publishedOn":"2025-10-29 15:58:35","publishedOnDateReadable":"October 29th, 2025"},"versionCreatedAt":"2024-06-24 10:38:29","video":"","vorDoi":"10.1007/s00704-025-05845-w","vorDoiUrl":"https://doi.org/10.1007/s00704-025-05845-w","workflowStages":[]},"version":"v1","identity":"rs-4551619","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4551619","identity":"rs-4551619","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}