Intra-urban variability of long-term exposure to PM2.5 and NO2 in five cities in Colombia

Intra-urban variability of long-term exposure to PM2.5 and NO2 in five cities in Colombia | 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 Intra-urban variability of long-term exposure to PM2.5 and NO2 in five cities in Colombia Laura A. Rodriguez-Villamizar, Yurley Rojas, Sara Grisales, Sonia C. Mangones, and 19 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-2988847/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 12 Dec, 2023 Read the published version in Environmental Science and Pollution Research → Version 1 posted 6 You are reading this latest preprint version Abstract Rapidly urbanizing cities in Latin America experience high levels of air pollution which are known risk factors for population health. However, the estimates of long-term exposure to air pollution are scarce in the region. We developed intraurban land use regression (LUR) models to map long-term exposure to fine particulate matter (PM 2.5 ) and nitrogen dioxide (NO 2 ) in the five largest cities in Colombia. We conducted air pollution measurement campaigns using gravimetric PM 2.5 and passive NO 2 sensors for two weeks during both the dry and rainy seasons in 2021 in the cities of Barranquilla, Bucaramanga, Bogotá, Cali, and Medellín, and combined these data with geospatial and meteorological variables. Annual models were developed using multivariable spatial regression models. The city annual PM 2.5 mean concentrations measured ranged between 12.32 𝛍g/m 3 and 15.99 𝛍g/m 3 while NO 2 concentrations ranged between 24.92 𝛍/m3) and 49.15 𝛍g/m 3 . The PM 2.5 annual models explained 82% of the variance (R 2 ) in Medellín, 77% in Bucaramanga, 73% in Barranquilla, 70% in Cali, and 44% in Bogotá. The NO 2 models explained 65% of the variance in Bucaramanga, 57% in Medellín, 44% in Cali, 40% in Bogotá, and 30% in Barranquilla. Most of the predictor variables included in the models were a combination of specific land use characteristics and roadway variables. Cross-validation suggest that PM 2.5 outperformed NO 2 models. The developed models can be used as exposure estimate in epidemiological studies, as input in hybrid models to improve personal exposure assessment, and for policy evaluation. Air pollution Fine Particulate Matter Nitrogen Dioxide Land Use Regression Models Colombia Figures Figure 1 Figure 2 Figure 3 Introduction Air pollution is recognized as one of the leading environmental risk factors for population health (GBD 2019 Risk Factors Collaborators, 2020 ). It is estimated that 99% of the world population is living in places where air pollution levels for fine particulate matter (PM 2.5 ) exceed the current safe guideline level defined by the World Health Organization (WHO), and populations from low- and middle-income countries are exposed to the highest levels (World Health Organization, 2021 ). In 2019, it was estimated that a total of 2.92 million deaths in females and 3.75 million deaths in males were attributable to ambient particulate matter and ozone air pollution. For Latin America and the Caribbean (LAC) region, and overall for low- and low-middle income countries, air pollution was the second most important risk factor (after malnutrition) that accounted for attributable disability-adjusted life-years (DALYs) rates over the past decade (GBD 2019 Risk Factors Collaborators, 2020 ). Particulate matter (PM 2.5 ), nitrogen dioxide (NO 2 ), and ozone (O 3 ) are the ambient air pollutants most strongly associated with adverse health adverse effects in the short- and long-term (World Health Organization, 2021 ). The health effects from long-term exposure to air pollution are 10-fold higher than the short-term effects represented by daily variations (Pope, 2007 ). For long-term exposure there is also evidence that there are large within-city contrasts and their effects are probably higher than the effects related to variations between cities (Crouse et al., 2015 ). Therefore, high-resolution spatial estimations of long-term exposure to air pollutants, particularly in urban setting, are critical for epidemiological research studying the association between air pollution and health and an important input for air quality management plans aimed to reduce air pollution adverse effects (Fann et al., 2011 ). There are different methods for estimating intraurban spatial variability of air pollutants. These methods include models based on proximity to monitoring stations, interpolation methods, land use regression models (LUR), and dispersion and chemical transport models combined with satellite remote sensing (Dijkema et al., 2011 ; Hoek, 2017 ; Hoek et al., 2008 ; Michael Jerrett et al., 2005 ; van Donkelaar et al., 2021 ). LUR models combined monitoring of air pollutants with the development of stochastic models using physical landscape characteristics, meteorology and population as predictor variables (Hoek et al., 2008 ). LUR models have lower computational requirements compared with dispersion or chemical transport models and are relatively easy to implement using Geographic Information Systems (GIS), which made them a method of preference in developing intraurban surfaces of air pollutant exposure (Hoek et al., 2008 ). LUR models have shown to have a high predictive value and to be a cost-effective method to estimate intraurban variations of air pollutants in different regions including North America, Europe and Asia (Allen et al., 2011 ; Chen et al., 2013 ; de Hoogh et al., 2016 ; Eeftens et al., 2012 ; Gurung et al., 2017 ; Kashima et al., 2018 ; Lee et al., 2017 ; Stafoggia et al., 2022 ). Recently LUR has been used in these regions as input data for hybrid models combining dispersion models, satellite-based observations, land use, and surface monitoring data for PM 2.5 and NO 2 (Hoek, 2017 ). Also, annual and monthly global estimates of ground level PM 2.5 and NO 2 have been developed, combining satellite remote sensing with the GEOS-Chem chemical transport model and calibration using ground-level observations (van Donkelaar et al., 2021 ). These models provide spatially fine resolutions at 0.01° × 0.01°and have shown to have a very good performance in North America and Europe but have very high uncertainty for tropical areas particularly in South America (Hoek, 2017 ; van Donkelaar et al., 2021 ). Despite LAC cities are growing rapidly and experiencing high levels of air pollution, the estimates of long-term exposure to air pollution are scarce in the region. In most cities, the ground-level measurements of atmospheric pollutants have poor consistency and coverage (Cunha-Zeri & Ometto, 2021 ). Limitations include that traditional air quality stations require high financial funding in resource-limited countries which make them logistically prohibitive since it is not cost-effective. Consequently, given the limited resources of good air quality data, modeling emerges as a possible tool to derive management measures (Agudelo-Castañeda et al., 2023 ). However, high-resolution spatial estimations of long-term exposure to air pollutants are scarce in LAC and development of LUR models for some pollutants have been reported only for the cities of Mexico, Sao Paulo, Quito, and Medellín (Alvarez-Mendoza et al., 2019 ; Habbermann & Gouveia, 2007 ; Londoño & Cañon, 2015 ; Luminati et al., 2021 ; Son et al., 2018 ). Colombia is located at the extreme north of South America with an estimated population of 52 million inhabitants (Departamento Nacional de Estadística (DANE), 2020b ) distributed across 32 departments and 1122 municipalities. The national air quality surveillance network has operated since 1993 and currently includes 22 regional surveillance systems that are distributed in 77 municipalities of 19 departments. In 2021, the national monitoring network included 131 monitoring stations for PM 2.5 and 57 for NO 2 (Instituto de Hidrologia Meteorología y Estudios Ambientales-IDEAM, 2022 ). Data from monitoring stations provide useful information for temporal daily variations of pollutants but provide limited information on the spatial variability of pollution especially in densely populated urban settings that concentrate 77% of the country´s population (Departamento Nacional de Estadística (DANE), 2020b ). Data from monitoring surveillance systems have been used in epidemiological studies assessing the short-term effects of pollutant concentrations on mortality and morbidity in the largest cities in Colombia (Blanco-Becerra et al., 2014 ; Rodriguez-Villamizar et al., 2018 ). However, there is a need for estimations of long-term spatial variation of pollutants within cities. Therefore, our objective was to develop intraurban LUR models for PM 2.5 and NO 2 in the five largest cities in Colombia to estimate of long-term population exposure to air pollution for use in air quality health assessment and mitigation. Methods Study areas The study was conducted in the urban areas of the five largest cities in Colombia: Barranquilla, Bucaramanga, Bogotá, Cali, and Medellín (Fig. 1 ). The population varies across cities, Bogotá being the most populated city with an estimated population of 7,834,167 million inhabitants in 2021. The estimated total population during 2021 was 1,297,082 for Barranquilla, 614,269 for Bucaramanga, 2,264,748 for Cali, and 2,573,220 inhabitants for Medellín (Departamento Nacional de Estadística (DANE), 2020b ). The altitude and average temperature also vary across cities, with Barranquilla being the warmer and closest to the sea level and Bogotá being the coldest and highest elevation. The physical characteristics of these cities are presented in Table S1 in Supplementary material. Similar to other capital cities in South America, the roadways networks in these cities are complex and dense, and both industrial and residential neighborhoods coexist. Air pollution measurement data PM 2.5 and NO 2 concentrations were measured in the five cities for two consecutive weeks during both the dry and the rainy season in 2021. The selection of the dry and rainy seasons for each city was defined based on the total precipitation registered in local meteorological stations between 2010 to 2019. The driest months correspond to January to March while the months with higher precipitation were April to May for most cities. The details of the sampling period for each city are presented in Table S1 in Supplementary material. For NO 2 there were 80 sampling sites for Bogotá and 40 for the other cities while for PM 2.5 there were 40 sampling sites for Bogotá and 20 for the other cities. Figure 1 shows the location of sampling sites distributed across the urban area of the cities. The density of sampling sites in the urban areas for NO 2 measurements (samplers per km 2 ) was 2.3 for Barranquilla, 0.8 for Bucaramanga, 4.4 for Bogotá, 3.5 for Cali and 3.6 for Medellín; the density of sampling for PM 2.5 was twice these values as we used half the number of monitors. The selection of sampling sites was conducted with participation of the study team and experts from the environmental and health departments of each city. The criteria for selecting the monitoring sites included: 1) the representation of traffic, residential, industrial or other areas within the cities, and 2) the heterogeneity in the characteristics of the selected sites (i.e., in terms of types of traffic, density of residential areas or particular areas for cities such as port or industrial areas). The sampling sites included one background urban site per city. The background site was located in the area of the city with the lowest concentrations of pollutants based on measurements, if they were available, or based on the experts’ knowledge of pollution within the city. In addition, sampling included 3–4 sites per city that were installed in the same location as monitoring stations from the local air quality network to facilitate instrument intercomparisons. For quality control two blank filters were used for each city. Measurement campaigns were simultaneously conducted across all sampling sites in each city for two weeks. Two trained teams of field staff were responsible for installing and uninstalling monitoring samplers across the study cities. We measured gravimetric PM 2.5 using Ultrasonic Personal Aerosol Sampler (UPAS) samplers (V2.0 Access Sensor Technologies, Fort Collins, Colorado, USA) that were installed between 2.5–3 meters above ground in all monitoring sites. The UPAS monitors have been widely used for measuring gravimetric PM 2.5 in similar and higher pollution settings (Arku et al., 2020 ) and have shown good performance for collecting airborne PM for gravimetric analysis (Leith et al., 2020 ). We adapted an environmental enclosure to protect the device during outdoor sampling and added an external battery to increase the sampling time to 7 days at 25% duty cycle at a flow rate of 1 lpm. Each monitor was loaded with a 37mm Teflon filter at the start of each measurement period. We replaced the UPAS and filters at each sampling site after 7 days to complete the two weeks monitoring period. Gravimetric analysis was conducted for all cities in a single laboratory certified for this competence (ISO/IEC 17025:1999) by the Instituto de Hidrología, Meteorología y Estudios Ambientales (IDEAM). Each filter and blank were weighted three times and the average measurement was reported for each filter. The reported limit of quantification was 0.68 𝛍g and the limit of detection was 1.36 𝛍g. The average PM 2.5 concentration of the two weekly filters from the same site and campaign was reported as the site concentration for statistical analysis. For measuring NO 2 we used passive diffusion Palmes Tubes (Gradko environmental, Hampshire, UK) that were installed for two weeks with a height of 2.5–3 meters above ground in all monitoring sites. For quality control an extra two blank tubes were deployed in each city. The processing of all tubes was conducted in the manufacturers laboratory and concentration measurements were reported as the average of duplicate measurements. The reported limit of detection was 0.031 𝛍g of NO 2 in tubes. The installation, operation, and deinstallation of the PM 2.5 and NO 2 monitoring devices including refrigeration of samples was conducted by trained personnel following the manufacturer's instructions. GIS predictor variables Predictor variables were grouped into five categories: 1) land use (areas of different land uses); 2) population (including population counts and population density); 3) roads (including total length of roads and distance from sampling sites to arterial roads); 4) traffic (including estimated average speed and traffic volume); 5) physical geography (altitude); and 6) meteorology (including average temperature, precipitation, relative humidity, and wind direction). All predictor variables were created for circular buffers with radii of 100m, 200m, and 500m and centered at the monitoring sites. These predictor variables were obtained from the intersection between buffers and GIS layers. In total, 78 independent variables were generated including variations of roadways variables. Maps were created using ESRI ArcGIS® 10.8.1 and ArcMap™ under license (ESRI® version, US). Table 1 provides the details of the predictor variables used to generate the LUR models. Land use data were obtained from the local government's planning office based on the most recent land use distribution available. Altitude was measured in sampling sites directly using an altimeter during the first deployment of monitoring devices. Population data and roads classification were obtained from the demographic and cartographic information of the census 2018 (Departamento Nacional de Estadística (DANE), 2020a ). Meteorology data were obtained from meteorological monitoring stations from the local environmental authority including 16 stations in Bogotá, 22 stations in Medellín and 8 stations in Cali. Precipitation and temperature raster surfaces were calculated using the Regnie model (Rauthe et al., 2013 ). Briefly, we used data from stations coupled with altitude from the digital terrain model (DTM) with 30 meters resolution, the slope and land exposure (the direction or azimuth angle of the inclination of the slope) to calculate spatial precipitation and temperature mean values using a linear regression model. Barranquilla and Bucaramanga had less than four local meteorological stations that did not allow for a valid spatial estimation and therefore meteorological data was not included in LUR models for these two cities. Traffic predictor variables were measured and estimated for the project. The traffic speed measurements were obtained during the same monitoring campaigns periods by using a cloud -based data method that included data pre-processing, speed computation and output data formatting. During the pre-processing, the street network vector data from Openstreetmap was edited to match the same network used by the Google Maps platform. Then, the network streets were split into 100-meter links considering the road intersections setup. Then, the speed was computed for those links using their length and travel time. Travel times at the link level were obtained from the Google Maps platform using the Distance Matrix API service, which provides predicted values at the time the service was used. Finally, the speed of each link was added to its attributes set, and the whole collection of links were used to create a GIS layer using Python scripts. To estimate traffic volumes, we used speed-density-flow functions, which describe the relationships between traffic speed, density, and flow rate on a road segment. These functions were obtained and used to estimate traffic conditions (Council, 2010 ). We computed speed-density-flow functions for urban traffic for Bogota (73 road segments) and Medellín (199 road segments) using data from sensors and traffic cameras provided by the transportation authorities. We computed and validated the functions for three different traffic regimes: interrupted, semi-interrupted, and uninterrupted flow. We tested six theoretical functional forms (Greenshields, Drew, Pipes, May&Keller, Greenberg and Underwood Model) (Gaddam & Rao, 2019 ) by using random sampling with replacement. The best model was selected based on the root mean square error (RMSE). The resulting functional forms were then used to estimate traffic volumes in the road network of Barranquilla, Cali, and Bucaramanga, taking into account the traffic regimes, and the number of lanes in each road segment. Table 1 Land Use Regression Predictor variables a Category/ Source Year Unit Variable names Land use Land use plan for each city 2014 % of land use type (square meters) Industrial-IND Residencial-RES Dotacional-DOT Central-CEN Commercial-COM Port-PORT Mixed-MIX Population National Census 2018 people per square meter Total population-POB Population Density-DEN Roads Open Street Maps 2020 kilometers Length by road type Trunk road-TRUNK Primary road-PRIM Secondary road-SEC Tertiary road-TER Local road-LOC Distance from site to road by road type DTRUNK DPRIM DSEC DTER DLOC Traffic Speed Distance Matrix API - Google Speed-density-flow functions 2021 kilometers per hour Vehicles per hour Traffic Speed- VEL Traffic volume- VOL Physical geography Altitude 2021 Meters Above Sea Level (MASL) Altitude-ALT Meteorology Monitoring Station 2021 Temperature ºC Precipitation mm Relative Humidity % Wind direction Temperature TPROM Precipitation PPROM Humidity (%) HR Wind direction WD a All predicted variables were created for buffers of 100m, 200m, and 500m. Statistical analysis We averaged pollutants' concentrations measured during both sampling campaigns to obtain annual means for each city. The comparison of measurements of the PM 2.5 sampling device with local monitoring stations was conducted for 13 monitoring stations with data available (2 in Barranquilla, 4 in Bogotá, 4 in Cali and 3 in Medellín). Comparison of concentrations were evaluated using Bland and Altman agreement coefficients and graphs (Bland & Altman, 1986 ). The average annual measurements across the monitoring sites were also compared to the average annual estimation measurements from the real-time local monitoring stations in the cities. We developed LUR models to estimate intraurban spatial variation of PM 2.5 and NO 2 within the five cities. We used multivariable spatial regression models, that allow local estimations of a dependent variable 𝑍, by implementing the Ordinary Least Squares (OLS) method, in the presence of possible explanatory variables ( \({\text{Z}}_{\text{j}}\) ) at the same point \(\left({\text{x}}_{\text{i}},{\text{y}}_{\text{i}}\right)\) represented by the following equation (Londoño, 2018 ; Maantay & McLafferty, 2011 ): $$\text{Z}({\text{x}}_{\text{i}},{\text{y}}_{\text{i}})={{\beta }}_{0}+{\sum }_{\text{j}=1}^{\text{n}}{{\beta }}_{\text{i}}{\text{Z}}_{\text{j}}({\text{x}}_{\text{i}},{\text{y}}_{\text{i}})+{{\epsilon }}_{\text{j}} , {{\epsilon }}_{\text{j}}\sim\text{N}(0,{\text{v}}^{2})$$ To represent the spatial dependency structure between the features being analyzed, the best combination of explanatory variables must be determined. In a first step, we removed highly correlated variables (> 0.7) and those variables in which zero values account for more than 90% of the sampling sites. Then, all the predictors are included in the model assessing their statistical significance (p value < 0.05) and the sign for their coefficient (𝛽𝑖) (observing their agreement with the expected -theoretical. direction of effect). In addition, the selected variables must adequately specify the regression model, by evaluating the specification criteria of the OLS method. We estimated the adjusted R-squared to assess the performance of the models and the variance inflation factor to determine multicollinearity. All models were built with a combination of all the buffer variables (Eeftens et al., 2012 ; Van Nunen et al., 2017 ). We performed a Geographically Weighted Regression (GWR) with the selected equation to examine the spatial heterogeneity of the relationship between air pollutants and other spatial variables and to estimate the multiple regression model parameters. Then, we created a regular point mesh with cells spaced by 200m over the cities´ surface, where the formula obtained by each annual regression model was applied, in order to predict air pollutant levels for each point. Then, a spatial interpolation method (spline) was applied to obtain the concentration surface of the pollutant in the study area. Finally, we performed a leave-one-out cross validation (Eeftens et al., 2012 ; Wang et al., 2016 ) for each LUR model in each city and compared the set of predicted values against the observed ones. Then, the cross-validated square error and R 2 were calculated for each model. The cross-validation was conducted using the “loocv” command in Stata® version 13 (Stata Corporation). Results Pollutants´ concentrations at sampling locations There were 116 PM 2.5 sampling sites with valid measurements for both monitoring campaigns used for the estimation of the annual average concentrations. Three sites in Cali, four sites in Bogotá, and one in Medellin were excluded because they contribute only one successful measurement. The mean PM 2.5 concentrations during the dry season were slightly higher compared to the rainy season (see supplementary material Table S1 ). The annual PM 2.5 mean concentration and range in sampling sites were 16.12 𝛍g/m 3 (7.42–22.22) for Medellín, 15.90 𝛍g/m 3 (3.64–35.30) for Barranquilla, 15.79 𝛍g/m 3 (4.86–32.69) for Cali, 13.89 𝛍g/m 3 (4.39–25.52) for Bogotá, and 12.93 𝛍g/m 3 (4.90-32.23) for Bucaramanga. For NO 2 sampling, 17 out of the 240 tubes deployed were removed due to vandalism or invalid measurements, leaving 223 observations for the analyses. The mean NO 2 concentrations during the dry season were slightly higher than those in the rainy season (see supplementary material Table S1 ). The annual NO 2 mean concentration and range in sampling sites were 49.09 𝛍g/m 3 (32.38–68.31) for Medellín, 34.92 𝛍g/m 3 (12.56–64.67) for Bucaramanga, 39.12 𝛍g/m 3 (13.52–69.89) for Cali, 34.63 𝛍g/m 3 (5.09–52.19) for Bogotá, and 24.92 𝛍/m3 (7.38–51.81) for Barranquilla. The average of the differences in PM 2.5 concentrations measured using the UPAS and those reported during the same sampling period by local monitoring stations was − 1.5 𝛍g/m 3 (95%CI -6.8 to 3.9) during the dry season campaign (11 monitoring stations) and − 0.05 𝛍g/m 3 (95% CI -11.5 to 11.4) during the rainy season campaign (13 monitoring stations). During the dry season campaign, higher differences were observed for two local monitoring stations, one in Medellín and one in Cali. During the rainy season campaign, higher differences were observed for the three local monitoring stations from Medellín. Figure S1 shows the levels of agreement for PM 2.5 measurements during the two monitoring campaigns. There was only one monitoring station in downtown Medellín with valid NO 2 data for comparison of measurements obtained from Palmes tubes and local monitors. For this site-station pair the differences were 5.71 and 2.59 𝛍g/m 3 during the dry and rainy season, respectively. In Bogotá during the second campaign (rainy season) there were four sites with valid paired measurements whose average difference was 6.70 𝛍g/m 3 , which was highly influenced by the discrepancy observed in one particular station located at Carrera 7a (excluding this station the average of the difference was 2.86 𝛍g/m 3 ). The comparison of the PM 2.5 average campaign’s measurements from monitoring sites with the average annual measurements from monitoring stations during 2021 resulted in differences of -0.84 𝛍g/m 3 for Bucaramanga, -1.1 𝛍g/m 3 for Medellín, -1.7 𝛍g/m 3 for Bogotá, 1.4 𝛍g/m 3 for Cali, and 1.7 𝛍g/m 3 for Barranquilla. For NO 2 the difference between passive samplers and monitoring stations in Bogotá was 5.6 𝛍g/m 3 . LUR models The final LUR models selected for the cities explained higher variability for PM 2.5 compared with NO 2 . (Table 2 and Table 3 , respectively). The models for PM 2.5 explained between 44% (Bogotá) and 82% (Medellín) of pollutant´s spatial variability within cities. Most models showed a RMSE of approximately 1.5 𝛍g/m 3 except for Barranquilla where the error was approximately 4 𝛍g/m 3 . The contrasts between PM 2.5 measured and predicted concentrations at monitoring sites for all cities are presented in Supplementary material Figure S2. Most of the predictor variables included in the PM 2.5 LUR models were related to specific types of land uses and roadways´ attributes with predominance of 200 and 500m buffers. In Bucaramanga the LUR model only included roadways variables while Medellín was the only city where the model included a meteorological variable (see Table 2 ). There was no evidence of multicollinearity in the LUR models for both pollutants as the VIF values were all below 2.1. The maps of the predicted concentrations for PM 2.5 in the urban areas of the five cities are presented in Fig. 2 . The final selected models for NO 2 explained between 30% (Barranquilla) and 65% (Bucaramanga) of the pollutant´s spatial variability within cities. Most cities models showed a RMSE around 6 to 8 𝛍g/m 3 except for Cali where the error was close to 1.5 𝛍g/m 3 . The measured values versus the predicted values of the models in the monitoring sites for NO 2 in all cities are presented in Supplementary material Figures S3. As expected, most of the predictor variables included in the NO 2 LUR models were a combination of roadways variables with different buffers. In Bucaramanga the LUR model included population variables and in Medellín one meteorological variable (see Table 2 ). There was no collinearity in the LUR models for both pollutants as the VIF values were all below 1.7. The maps of the predicted concentrations for NO 2 in the urban areas of the five cities are presented in Fig. 3 . Cross validation Overall, the leave-one-out cross-validation R 2 s showed good stability, particularly for PM 2.5 . For PM 2.5 , the difference between the model R 2 and the validation R 2 was 19% for Barranquilla, 31% for Bucaramanga, 6% for Bogotá, 19% for Cali and 3% for Medellín. For NO 2 , the difference between the model R 2 and the validation R 2 was 11% for Barranquilla, 10% for Bucaramanga, 6% for Bogotá, 8% for Cali and 12% for Medellín. Validation R 2 s are presented in Table 2 and Table 3 for PM 2.5 and NO 2 , respectively. Table 2 Description of developed LUR models for PM 2.5 in five cities in Colombia, 2021 City LUR model No. sites Model R 2 RMSE VIF R 2 cross validation Barranquilla PM 2.5 = 19.83344–0.1489524*ALT − 0.0230902*DTRON500 + 44.43591*IND200 + 21.93109*CEN500 + 23.10317*PORT500 20 0.73 3.98 1.90 0.54 Bogotá Ln(PM 2.5 ) = 2.4713 + 3.1439*DEN100 + 1.8045*IND200–0.8418*RES500 40 0.44 1.39 1.63 0.38 Bucaramanga Ln(PM 2.5 ) = 2.199057 + 0.0014062∗SEC100 + 0.0000327∗LOC500 − 0.0012659∗DPRIM500 + 0.0215501∗VEL100 − 0.000242∗VOL200 20 0.77 1.23 1.78 0.46 Cali Ln(PM 2.5 ) = 2.3387 + 0.00001∗DOT200 + 1.0713∗PRIM200 + 0.5943∗SEC200–0.0004∗VOL100 17 0.70 1.28 2.06 0.51 Medellín PM 2.5 = 13.77207–1.357455∗PPROM − 5.589831∗DOT100 + 2.269679∗DEN200 + 70.23039∗MIX500 + 0.0043842∗VOL500 19 0.82 1.71 1.48 0.79 RMSE: Root mean square error; VIF: Variance inflation factor ALT = Altitude; CEN = Central land use; DEN = population density; DOT = Dotacional land use; DPRIM = Distance to primary roadway; DTRON = Distance to trunk roadway; IND = Industrial land use; LOC = Length local roadways; MIX: mixed land use; PORT = Port land use; PPROM = precipitation average; PRIM = Length primary roadways; RES = Residential land use; SEC = Length secondary roadways; VEL = Vehicular speed; VOL = vehicular volume. Numbers correspond to buffers of 100m, 200m, 500m. Table 3 Description of developed LUR models NO 2 in five cities in Colombia, 2021 City LUR model No. sites Model R 2 RMSE VIF R 2 cross validation Barranquilla NO 2 = 12.89591 + 25.45936∗PRIM100–0.1583713∗VEL100 + 0.0061518∗VOL500 36 0.30 8.02 1.28 0.19 Bogotá Ln(NO 2 ) = 2.8714 + 0.0001*PRIM500 + 0.0058*VEL100 + 0.2599*WPROM 73 0.40 1.26 1.19 0.34 Bucaramanga NO 2 = 13.00243 + 291.5302∗DEN100 − 0.0013283∗POB500 + 0.0025503∗TER500 + 0.0020514∗LOC500 + 0.0057464∗VOL100 40 0.65 8.08 1.66 0.55 Cali Ln(NO 2 ) = 3.47834312 + 0.49126931*PRIM200 + 0.39823891∗SEC200 + 0.36505469∗TER200–0.01475995∗VEL200 40 0.44 1.28 1.56 0.36 Medellín NO 2 = 46.06516–3.625967∗PPROM − 0.0299678∗DSEC200 + 0.0225605∗VOL500 34 0.57 5.53 1.06 0.45 RMSE: Root mean square error; VIF: Variance inflation factor DEN = population density; DOT = Dotacional land use; DSEC = Distance to secondary roadway; LOC = Length local roadways; POB = population size; PPROM = precipitation average; PRIM = Length primary roadways; SEC = Length secondary roadways; TER = Length tertiary roadways; VEL = Vehicular speed; VOL = vehicular volume, WPROM: Wind speed (mean). Numbers correspond to buffers of 100m, 200m, 500m. Discussion This is the first study to develop LUR models for multiple cities in a Latin American country, providing small-area estimations of air pollutants for use in health risk assessments, epidemiological studies of long-term exposure to air pollution and mitigation evaluation. The development of LUR models to estimate concentrations for PM 2.5 and NO 2 in five of the largest Colombian cities showed moderate to high explained variance, respectively. Generally, the models showed higher explained variance of PM 2.5 compared with NO 2 . Among the cities, the lowest explained variance was obtained for Bogotá, while the highest was recorded for Medellín and Bucaramanga. The LUR models for PM 2.5 showed relatively small errors of the predicted concentrations (RMSE < 1.7 𝛍g/m 3 ) in the cities, except for Barranquilla. Moreover, the performance of the LUR models developed for PM 2.5 was higher than that reported in previous studies in Colombia. Previous LUR models were available only for PM 10 and PM 2.5 in the city of Medellín with an explained variability of 79% for PM 10 (Londoño & Cañon, 2015 ) and monthly variations between 26% and 79% for PM 2.5 (Grisales, 2020 ), using data from 2007 and 2018, respectively. Our selected LUR model for PM 2.5 in Medellín explained 82% of the variability, the highest of the five cities, using a combination of meteorological, land use, population density and traffic volume variables. The high performance of the LUR models for PM 2.5 in Medellín compared to other cities might be explained by the wide range of estimated concentrations in the city and the influence of the topography and meteorology in the Valley of Aburrá where Medellín is located, as well as the important contribution of vehicular emissions to local concentrations as have been described in studies of PM 2.5 characterization in the city (Area Metropolitana del Valle de Aburrá & Politecnico Colombiano Jaime Isaza Cadavid, 2021). In contrast, the low performance of the LUR models for PM 2.5 in Bogotá compared to other cities might be explained partially by the lower contribution of vehicular emissions and the increased contribution of enriched fugitive dust (resuspension of crustal material and soil dust) and secondary PM (Ramírez et al., 2018 ). A similar profile has also been documented for Barranquilla with an important contribution of ocean aerosols (Nuñez Blanco, 2019 ), secondary organic aerosols and the effect of exposed land resuspension and road dust (Gómez-Plata et al., 2022 ), which was represented in the developed LUR model for this city. Additional unexplained variability in PM 2.5 concentrations in the cities might be related to regional wildfires contributions which have been substantial in northern South America and particularly in Bogotá (Ballesteros-González et al., 2020 )(Casallas et al., 2022 ). The variation in explained variability reported for the Colombian cities is comparable to that of PM 2.5 in other Latin American and European countries. In Ecuador, Alvarez et al. (Alvarez-Mendoza et al., 2019 ) developed LUR models for PM 10 using remote sensing data, and the models showed an explained variability of 68% at its highest. Sangrador et al. (Sangrador, J.T., Nuñez, M.E., Villarreal, A.B., Cadena, L.H., Jerrett, M., Romieu, 2008) developed LUR models for PM 2.5 during the rainy season in 2003 for Mexico City, which showed an explained variability of 60%. Later, Son et al. (Son et al., 2018 ) developed LUR models for the same city for different temporal scales, and the best explained variability for monthly PM 2.5 models was 76%. In Europe, the ESCAPE project developed LUR models for PM 2.5 in 20 study areas, where the explained variability varied from 35% in Manchester, UK, to 89% in Paris, France (Eeftens et al., 2012 ). As expected, the best predictor variables in our LUR models for NO 2 were road and traffic variables. However, the performance of the LUR models developed for NO 2 , however, was lower than that for PM 2.5 and the reported from previous studies in other countries. In Sao Paulo, an annual LUR developed for NO 2 explained 66% of the variability in urban concentrations, with variations for summer (75%) and winter (52%) seasons (Luminati et al., 2021 ). For the Western European countries, Vinneay et al. (Vienneau et al., 2013 ) developed LUR models for NO 2 with and without satellite-based NO 2 and obtained explained variability between 48% and 58% without satellite-based NO 2 and a modest additional improvement of 5% when adding satellite-based data. In our models for NO 2 , despite including different variables and metrics of traffic and roads, the models could not capture a higher variability in concentrations, which suggests secondary reactions might be an important source of NO 2 in the cities. Although our NO 2 LUR explained less variability compared to other reported models in cities, the LUR models explain more variability than simple road proximity metrics or interpolation methods based on data from monitoring stations and similar variability than dispersion models, which have been demonstrated in previous studies assessing exposure assessment for epidemiological studies (Allen et al., 2011 ; de Hoogh et al., 2014 ; M Jerrett et al., 2007 ). The LUR models have been used in exposure assessment and health research related to long-term exposure to air pollutants. By incorporating data on local sources of pollution, such as traffic or industrial activity, these models can provide more accurate and precise exposure estimates than traditional monitoring methods (Hoek et al., 2008 ). This is particularly important for assessing the health effects of chronic exposure to air pollution, which has been linked to a range of adverse health outcomes, including respiratory and cardiovascular disease, cancer, and neurological disorders (Chen et al., 2013 ; Herting et al., 2019 ; Knibbs et al., 2018 ; Lamichhane et al., 2017 ; Stafoggia et al., 2022 ). LUR models can also identify areas of high pollution levels and vulnerable populations, helping to inform policy and intervention strategies to reduce exposure and improve public health (Vienneau et al., 2013 ). Alternative methods for estimating surface concentrations of air pollutants have been developed recently using satellite-based models and models using mobile air pollutant measurements. A study conducted at the municipality level in Colombia compared air quality models based on satellite measurements for PM 2.5 between 2014–2019. It showed that the Copernicus Atmospheric Monitoring Service Reanalysis (CAMRA) and the Atmospheric Composition Analysis Group (ACAG) models had a low correlation and tended to overestimated surface concentrations when both models were compared to surface data from 28 cities in 2019. However, ACAG outperformed CAMSRA in terms of mean bias of the model and the spatial representation of the highest concentrations (Rodriguez-Villamizar et al., 2022 ). Using a mobile monitoring campaign in the city of Bucaramanga in 2019, estimations of within-city spatial variations in ultrafine particle and black carbon concentrations were predicted using a combination of LUR and convolutional neural networks trained using satellite and street-level images, showing the improvement of prediction when using a hybrid approach (Lloyd et al., 2021 ). Following this hybrid approach, our locally developed LUR models can be further used to develop hybrid models with satellite or mobile data and produce better spatially calibrated models for estimating long-term exposure for PM 2.5 and NO 2 in the main cities in Colombia and explore their potential transferability across cities. There are some strengths in our study that are worth mentioning. First, there was a good agreement between PM 2.5 measurements made with UPAS compared to the concentrations reported by the local monitoring stations in the cities. For NO 2 , there were few monitoring sites to conduct a valid comparison in all cities, but data from local government stations in Bogotá had a good agreement with concentrations reported from measurements with the Palmes tubes. Second, we followed the same standardized procedure for conducting measuring pollutants during the two campaigns in each city and the simultaneous measurement within cities avoid the potential error related to using measures in different time scales. Third, we included basic predictor variables for developing LUR models in the cities (land use, roads, traffic, population, and meteorology) available in the cities in Colombia and might be used further to developed multi-city models as those developed for Europe (Wang et al., 2014 ). One limitation of the LUR models developed for the cities is the limited number of sampling sites which was 20 for PM 2.5 and 40 for NO 2 , except for Bogotá which doubled the number. These numbers are below the lower range of recommended monitoring sites (between 80–100) for modeling intraurban variations in complex urban settings using LUR (Basagaña et al., 2012 ). As a result, the models developed using many predictors might have resulted in more unstable performance as was observed in the cross-validation. A second limitation of this study is the absence of valid traffic data for the cities during the campaign measurement, which has shown to improve the LUR model performance, particularly for NO 2 (Beelen et al., 2013 ). To overcome this limitation, we measured traffic speed derived from satellite instruments and used previously available traffic count data for the largest cities to calculate density functions which were then transferred to the other cities to estimated traffic density. Despite the density functions in the cities seemed to reflect the traffic patterns in the cities and were included as significant predictive variables, their inclusion did not help to explain a higher variability in the models for NO 2 . Third, we did not include meteorological variables in the development of LUR models for the cities of Bucaramanga and Barranquilla due to limited number of meteorological stations and data to produce a valid estimated surface. Although the models´ performance for PM 2.5 were good particularly for Bucaramanga, including meteorological variables might have increased the models´ performance as they have been reported as important predictors for intraurban variations in other countries (Cheewinsiriwat et al., 2022 ; Olvera Alvarez et al., 2018 ). Another limitation of our study is that we did not include local emission sources and regional sources (such as forest fires) in the prediction models. These variables have shown to influence the concentration of particles in the cities (Casallas et al., 2022 ). Moreover, street NO 2 levels may vary in building density or location, influencing their dispersion. Also, some atmospheric chemical reactions may reduce or transform NO 2 concentrations. In urban areas, NO 2 emitted mostly from traffic within a radius of 100-300m showed a correlation, although the high reactivity of NO 2 and rapid photodissociation may transform this pollutant in a reduced period (Agudelo-castañeda et al., 2020 ). Conclusion In this study we developed LUR models to predict PM 2.5 and NO 2 exposure in five main cities in Colombia. The LUR models showed a large intraurban variability of pollutant concentrations in all cities. The annual models for PM 2.5 outperformed the models for NO 2 and provided robust models that can be used in epidemiological studies, particularly cohort studies, assessing the effects of long-term air pollution on human health. The newly developed LUR models might be further used to create hybrid models in combination with other data sources to improve personal exposure assessment. Declarations Acknowledgements The authors thank Oscar Jiménez, Hermes Betancur, Jefferson Fernánez, Angie Rojas, Paola Barbosa, Kelly Burbano, Ronald Correa, Daniela Ortiz, Martha Mendoza, Wilmer Urango, Luz Obando, and Orlando Guaduña for their contributions during the monitoring field work in the cities. Funding: This research was funded by the Ministry of Science and Technology MINCIENCIAS in Colombia, grant number 905-2019. Competing Interests: The authors have no relevant financial or nonfinancial interests to disclose. Author Contributions Conceptualization and methodology: LR-V, SM, JC, DA-C, VH, DM, JPJ, LB-C, OR-S, JOV, SW, JB, NR; Field work and data collection: LL,OMR,MV,WS,AZO,MC,HS; Formal analysis and investigation: LR-V, YR, and SCG; Writing-original draft preparation: LR-V, YR, and SCG; Writing-review and editing: DA-C, VH, DM, JPJ, LB-C, OR-S, JOV, SW, JB, NR, LL,OMR,MV,WS,AZO,MC,HS; Funding acquisition: LR-V, SM, JC, DA-C, VH, DM, JPJ, LB-C, OR-S, JOV, NR. Ethics approval and consent to participate: Not applicable. Consent for publication: Not applicable Data availability: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. References Agudelo-Castañeda, D., Arellana, J., Morgado-Gamero, W. B., De Paoli, F., & Carla Portz, L. (2023). Linking of built environment inequalities with air quality: A case study. Transportation Research Part D: Transport and Environment, 117(March). https://doi.org/10.1016/j.trd.2023.103668 Agudelo-castañeda, D., Paoli, F. De, Morgado-gamero, W. B., Mendoza, M., Parody, A., Maturana, A. Y., & Teixeira, E. C. (2020). Assessment of the NO2 distribution and relationship with traffic load in the Caribbean coastal city. Science of the Total Environment, 720, 137675. https://doi.org/10.1016/j.scitotenv.2020.137675 Allen, R. W., Amram, O., Wheeler, A. J., & Brauer, M. (2011). The transferability of NO and NO2 land use regression models between cities and pollutants. Atmospheric Environment, 45(2), 369–378. https://doi.org/10.1016/j.atmosenv.2010.10.002 Alvarez-Mendoza, C. I., Teodoro, A. C., Torres, N., & Vivanco, V. (2019). Assessment of remote sensing data to model PM10 estimation in cities with a low number of air quality stations: A case of study in Quito, Ecuador. Environments - MDPI, 6(7). https://doi.org/10.3390/environments6070085 Area Metropolitana del Valle de Aburrá, A., & Politecnico Colombiano Jaime Isaza Cadavid, P. (2021). Aporte de fuentes y caracterización del PM2.5 en el Valle de Aburrá, Colombia, 2019-2021. Informe final proyecto ARCAL RLA7023-Convenio interadministrativo 671 de 2021. Arku, R. E., Brauer, M., Ahmed, S. H., AlHabib, K. F., Avezum, Á., Bo, J., Choudhury, T., Dans, A. M., Gupta, R., Iqbal, R., Ismail, N., Kelishadi, R., Khatib, R., Koon, T., Kumar, R., Lanas, F., Lear, S. A., Wei, L., Lopez-Jaramillo, P., … Hystad, P. (2020). Long-term exposure to outdoor and household air pollution and blood pressure in the Prospective Urban and Rural Epidemiological (PURE) study. Environmental Pollution (Barking, Essex : 1987), 262, 114197. https://doi.org/10.1016/j.envpol.2020.114197 Ballesteros-González, K., Sullivan, A. P., & Morales-Betancourt, R. (2020). Estimating the air quality and health impacts of biomass burning in northern South America using a chemical transport model. The Science of the Total Environment, 739, 139755. https://doi.org/10.1016/j.scitotenv.2020.139755 Basagaña, X., Rivera, M., Aguilera, I., Agis, D., Bouso, L., Elosua, R., Foraster, M., de Nazelle, A., Nieuwenhuijsen, M., Vila, J., & Künzli, N. (2012). Effect of the number of measurement sites on land use regression models in estimating local air pollution. Atmospheric Environment, 54, 634–642. https://doi.org/10.1016/j.atmosenv.2012.01.064 Beelen, R., Hoek, G., Vienneau, D., Eeftens, M., Dimakopoulou, K., Pedeli, X., Tsai, M. Y., Künzli, N., Schikowski, T., Marcon, A., Eriksen, K. T., Raaschou-Nielsen, O., Stephanou, E., Patelarou, E., Lanki, T., Yli-Tuomi, T., Declercq, C., Falq, G., Stempfelet, M., … de Hoogh, K. (2013). Development of NO2 and NOx land use regression models for estimating air pollution exposure in 36 study areas in Europe – The ESCAPE project. Atmospheric Environment, 72, 10–23. https://doi.org/10.1016/J.ATMOSENV.2013.02.037 Blanco-Becerra, L. C., Miranda-Soberanis, V., Hernández-Cadena, L., Barraza-Villarreal, A., Junger, W., Hurtado-Díaz, M., & Romieu, I. (2014). Effect of particulate matter less than 10µm (PM10) on mortality in Bogota, Colombia: a time-series analysis, 1998-2006. Salud Publica de Mexico, 56(4), 363–370. https://doi.org/10.21149/spm.v56i4.7356 Bland, J. M., & Altman, D. G. (1986). Statistical methods for assessing agreement between two methods of clinical measurement. Lancet (London, England), 1(8476), 307–310. Casallas, A., Castillo-Camacho, M. P., Guevara-Luna, M. A., González, Y., Sanchez, E., & Belalcazar, L. C. (2022). Spatio-temporal analysis of PM(2.5) and policies in Northwestern South America. The Science of the Total Environment, 852, 158504. https://doi.org/10.1016/j.scitotenv.2022.158504 Cheewinsiriwat, P., Duangyiwa, C., Sukitpaneenit, M., & Stettler, M. E. J. (2022). Influence of Land Use and Meteorological Factors on PM2.5 and PM10 Concentrations in Bangkok, Thailand. Sustainability (Switzerland), 14(9). https://doi.org/10.3390/su14095367 Chen, H., Goldberg, M. S., Burnett, R. T., Jerrett, M., Wheeler, A. J., & Villeneuve, P. J. (2013). Long-term exposure to traffic-related air pollution and cardiovascular mortality. Epidemiology, 24(1), 35–43. https://doi.org/10.1097/EDE.0b013e318276c005 Council, N. R. (2010). Highway capacity manual 2010. Crouse, D. L., Peters, P. A., Villeneuve, P. J., Proux, M.-O., Shin, H. H., Goldberg, M. S., Johnson, M., Wheeler, A. J., Allen, R. W., Atari, D. O., Jerrett, M., Brauer, M., Brook, J. R., Cakmak, S., & Burnett, R. T. (2015). Within- and between-city contrasts in nitrogen dioxide and mortality in 10 Canadian cities; a subset of the Canadian Census Health and Environment Cohort (CanCHEC). Journal of Exposure Science & Environmental Epidemiology, 25(5), 482–489. https://doi.org/10.1038/jes.2014.89 Cunha-Zeri, G., & Ometto, J. (2021). Nitrogen emissions in Latin America: A conceptual framework of drivers, impacts, and policy responses. Environmental Development, 38(October 2019), 100605. https://doi.org/10.1016/j.envdev.2020.100605 de Hoogh, K., Gulliver, J., van Donkelaar, A., Martin, R. V, Marshall, J. D., Bechle, M. J., Cesaroni, G., Pradas, M. C., Dedele, A., Eeftens, M., Forsberg, B., Galassi, C., Heinrich, J., Hoffmann, B., Jacquemin, B., Katsouyanni, K., Korek, M., Künzli, N., Lindley, S. J., … Hoek, G. (2016). Development of West-European PM2.5 and NO2 land use regression models incorporating satellite-derived and chemical transport modelling data. Environmental Research, 151, 1–10. https://doi.org/10.1016/j.envres.2016.07.005 de Hoogh, K., Korek, M., Vienneau, D., Keuken, M., Kukkonen, J., Nieuwenhuijsen, M. J., Badaloni, C., Beelen, R., Bolignano, A., Cesaroni, G., Pradas, M. C., Cyrys, J., Douros, J., Eeftens, M., Forastiere, F., Forsberg, B., Fuks, K., Gehring, U., Gryparis, A., … Bellander, T. (2014). Comparing land use regression and dispersion modelling to assess residential exposure to ambient air pollution for epidemiological studies. Environment International, 73, 382–392. https://doi.org/10.1016/j.envint.2014.08.011 Departamento Nacional de Estadística (DANE). (2020a). Geoportal Censo Nacional de Población y Vivienda 2018. https://geoportal.dane.gov.co/ Departamento Nacional de Estadística (DANE). (2020b). Proyecciones de Población. https://www.dane.gov.co/index.php/estadisticas-por-tema/demografia-y-poblacion/proyecciones-de-poblacion Dijkema, M. B., Gehring, U., van Strien, R. T., van der Zee, S. C., Fischer, P., Hoek, G., & Brunekreef, B. (2011). A comparison of different approaches to estimate small-scale spatial variation in outdoor NO2 concentrations. Environmental Health Perspectives, 119(2), 670–675. https://doi.org/10.1289/ehp.0901818 Eeftens, M., Beelen, R., De Hoogh, K., Bellander, T., Cesaroni, G., Cirach, M., Declercq, C., Dedele, A., Dons, E., De Nazelle, A., Dimakopoulou, K., Eriksen, K., Falq, G., Fischer, P., Galassi, C., Gražulevičiene, R., Heinrich, J., Hoffmann, B., Jerrett, M., … Hoek, G. (2012). Development of land use regression models for PM2.5, PM 2.5 absorbance, PM10 and PMcoarse in 20 European study areas; Results of the ESCAPE project. Environmental Science and Technology, 46, 11195–11205. https://doi.org/10.1021/es301948k Fann, N., Bell, M. L., Walker, K., & Hubbell, B. (2011). Improving the linkages between air pollution epidemiology and quantitative risk assessment. Environmental Health Perspectives, 119(12), 1671–1675. https://doi.org/10.1289/ehp.1103780 Gaddam, H. K., & Rao, K. R. (2019). Speed–density functional relationship for heterogeneous traffic data: a statistical and theoretical investigation. Journal of Modern Transportation, 27(1), 61–74. https://doi.org/10.1007/s40534-018-0177-7 GBD 2019 Risk Factors Collaborators. (2020). Global burden of 87 risk factors in 204 countries and territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet (London, England), 396(10258), 1223–1249. https://doi.org/10.1016/S0140-6736(20)30752-2 Gómez-Plata, L., Agudelo-Castañeda, D., Castillo, M., & Teixeira, E. C. (2022). PM10 Source Identification: A Case of a Coastal City in Colombia. Aerosol and Air Quality Research, 22(10), 1–17. https://doi.org/10.4209/aaqr.210293 Grisales, S. (2020). Modelos de regresión de usos del suelo para la caracterización espacial de la contaminación del aire por PM2.5 en la ciudad de Medellín-Colombia, 2018 [Universidad de Antioquia, Medellín, Colombia]. https://hdl.handle.net/10495/16376 Gurung, A., Levy, J. I., & Bell, M. L. (2017). Modeling the intraurban variation in nitrogen dioxide in urban areas in Kathmandu Valley, Nepal. Environmental Research, 155, 42–48. https://doi.org/10.1016/j.envres.2017.01.038 Habbermann, M., & Gouveia, N. (2007). Application of land use regression to predict the concentration of inhalable particulate matter in de São Paulo city, Brasil. Eng Sanit Ambient, 17(2), 155–162. Herting, M. M., Younan, D., Campbell, C. E., & Chen, J.-C. (2019). Outdoor Air Pollution and Brain Structure and Function From Across Childhood to Young Adulthood: A Methodological Review of Brain MRI Studies. In Frontiers in public health (Vol. 7, p. 332). https://doi.org/10.3389/fpubh.2019.00332 Hoek, G. (2017). Methods for Assessing Long-Term Exposures to Outdoor Air Pollutants. Current Environmental Health Reports, 4(4), 450–462. https://doi.org/10.1007/s40572-017-0169-5 Hoek, G., Beelen, R., de Hoogh, K., Vienneau, D., Gulliver, J., Fischer, P., & Briggs, D. (2008). A review of land-use regression models to assess spatial variation of outdoor air pollution. Atmospheric Environment, 42(33), 7561–7578. https://doi.org/10.1016/j.atmosenv.2008.05.057 Instituto de Hidrologia Meteorología y Estudios Ambientales-IDEAM. (2022). Informe del estado de la calidad del aire en Colombia 2021. Jerrett, M, Arain, M. A., Kanaroglou, P., Beckerman, B., Crouse, D., Gilbert, N. L., Brook, J. R., Finkelstein, N., & Finkelstein, M. M. (2007). Modeling the intraurban variability of ambient traffic pollution in Toronto, Canada. Journal of Toxicology and Environmental Health. Part A, 70(3–4), 200–212. https://doi.org/10.1080/15287390600883018 Jerrett, Michael, Arain, A., Kanaroglou, P., & Beckerman, B. (2005). A review and evaluation of intraurban air pollution exposure models. 185–204. https://doi.org/10.1038/sj.jea.7500388 Kashima, S., Yorifuji, T., Sawada, N., Nakaya, T., & Eboshida, A. (2018). Comparison of land use regression models for NO2 based on routine and campaign monitoring data from an urban area of Japan. The Science of the Total Environment, 631–632, 1029–1037. https://doi.org/10.1016/j.scitotenv.2018.02.334 Knibbs, L. D., Cortes de Waterman, A. M., Toelle, B. G., Guo, Y., Denison, L., Jalaludin, B., Marks, G. B., & Williams, G. M. (2018). The Australian Child Health and Air Pollution Study (ACHAPS): A national population-based cross-sectional study of long-term exposure to outdoor air pollution, asthma, and lung function. Environment International, 120, 394–403. https://doi.org/10.1016/j.envint.2018.08.025 Lamichhane, D. K., Kim, H. C., Choi, C. M., Shin, M. H., Shim, Y. M., Leem, J. H., Ryu, J. S., Nam, H. S., & Park, S. M. (2017). Lung Cancer Risk and Residential Exposure to Air Pollution: A Korean Population-Based Case-Control Study. Yonsei Medical Journal, 58(6), 1111–1118. https://doi.org/10.3349/ymj.2017.58.6.1111 Lee, M., Brauer, M., Wong, P., Tang, R., Tsui, T. H., Choi, C., Cheng, W., Lai, P.-C., Tian, L., Thach, T.-Q., Allen, R., & Barratt, B. (2017). Land use regression modelling of air pollution in high density high rise cities: A case study in Hong Kong. The Science of the Total Environment, 592, 306–315. https://doi.org/10.1016/j.scitotenv.2017.03.094 Leith, D., L’Orange, C., Mehaffy, J., & Volckens, J. (2020). Design and performance of UPAS inlets for respirable and thoracic mass sampling. Journal of Occupational and Environmental Hygiene, 17(6), 274–282. https://doi.org/10.1080/15459624.2020.1741595 Lloyd, M., Carter, E., Diaz, F. G., Magara-Gomez, K. T., Hong, K. Y., Baumgartner, J., Herrera G, V. M., & Weichenthal, S. (2021). Predicting Within-City Spatial Variations in Outdoor Ultrafine Particle and Black Carbon Concentrations in Bucaramanga, Colombia: A Hybrid Approach Using Open-Source Geographic Data and Digital Images. Environmental Science & Technology, 55(18), 12483–12492. https://doi.org/10.1021/acs.est.1c01412 Londoño, L. (2018). Metodología para caracterizar espacio-temporalmente la concentración de material particulado en Valles intramontanos con información escasa [Universidad de Antioquia, Medellín, Colombia]. https://bibliotecadigital.udea.edu.co/handle/10495/9413 Londoño, L., & Cañon, J. (2015). Metodologia para la aplicación de modelos de regresión de usos de suelo en la estimación local de la concentración mensual de PM10 en Medellín, Colombia. Revista Politécnica, 11(21), 29–40. Luminati, O., Ledebur de Antas de Campos, B., Flückiger, B., Brentani, A., Röösli, M., Fink, G., & de Hoogh, K. (2021). Land use regression modelling of NO(2) in São Paulo, Brazil. Environmental Pollution (Barking, Essex : 1987), 289, 117832. https://doi.org/10.1016/j.envpol.2021.117832 Maantay, J., & McLafferty, S. (2011). Geospatial Analysis of Environmental Health. Springer. https://doi.org/10.1007/978-94-007-0329-2 Nuñez Blanco, Y. (2019). Estimación de fuentes de material particulado atmosférico (PM10 y PM2.5) en la ciudad de Barranquilla, Colombia. Universidad de la Costa, Barranquilla, Colombia. Olvera Alvarez, H. A., Myers, O. B., Weigel, M., & Armijos, R. X. (2018). The value of using seasonality and meteorological variables to model intra-urban PM2.5 variation. Atmospheric Environment, 182(March), 1–8. https://doi.org/10.1016/j.atmosenv.2018.03.007 Pope, C. A. 3rd. (2007). Mortality effects of longer term exposures to fine particulate air pollution: review of recent epidemiological evidence. Inhalation Toxicology, 19 Suppl 1, 33–38. https://doi.org/10.1080/08958370701492961 Ramírez, O., Sánchez de la Campa, A. M., Amato, F., Catacolí, R. A., Rojas, N. Y., & de la Rosa, J. (2018). Chemical composition and source apportionment of PM(10) at an urban background site in a high-altitude Latin American megacity (Bogota, Colombia). Environmental Pollution (Barking, Essex : 1987), 233, 142–155. https://doi.org/10.1016/j.envpol.2017.10.045 Rauthe, M., Steiner, H., Riediger, U., Mazurkiewicz, A., & Gratzki, A. (2013). A Central European precipitation climatology - Part I: Generation and validation of a high-resolution gridded daily data set (HYRAS). Meteorologische Zeitschrift, 22(3), 235–256. https://doi.org/10.1127/0941-2948/2013/0436 Rodriguez-Villamizar, L. A., Belalcazar-Ceron, L. C., Castillo, M. P., Sanchez, E. R., Herrera, V., & Agudelo-Castañeda, D. M. (2022). Avoidable mortality due to long-term exposure to PM(2.5) in Colombia 2014-2019. Environmental Health : A Global Access Science Source, 21(1), 137. https://doi.org/10.1186/s12940-022-00947-8 Rodriguez-Villamizar, L. A., Rojas-Roa, N. Y., Blanco-Becerra, L. C., Herrera-Galindo, V. M., & Fernández-Niño, J. A. (2018). Short-Term Effects of Air Pollution on Respiratory and Circulatory Morbidity in Colombia 2011 − 2014: A Multi-City, Time-Series Analysis. International Journal of Environmental Research and Public Health, 15(8). https://doi.org/10.3390/ijerph15081610 Sangrador, J.T., Nuñez, M.E., Villarreal, A.B., Cadena, L.H., Jerrett, M., Romieu, I. (2008). A land use regression model for predicting PM2.5 in Mexico City. Epidemiology, 19(S259), 1. Son, Y., Osornio-Vargas, A. R., O’Neill, M. S., Hystad, P., Texcalac-Sangrador, J. L., Ohman-Strickland, P., Meng, Q., & Schwander, S. (2018). Land use regression models to assess air pollution exposure in Mexico City using finer spatial and temporal input parameters. The Science of the Total Environment, 639, 40–48. https://doi.org/10.1016/j.scitotenv.2018.05.144 Stafoggia, M., Oftedal, B., Chen, J., Rodopoulou, S., Renzi, M., Atkinson, R. W., Bauwelinck, M., Klompmaker, J. O., Mehta, A., Vienneau, D., Andersen, Z. J., Bellander, T., Brandt, J., Cesaroni, G., de Hoogh, K., Fecht, D., Gulliver, J., Hertel, O., Hoffmann, B., … Janssen, N. A. H. (2022). Long-term exposure to low ambient air pollution concentrations and mortality among 28 million people: results from seven large European cohorts within the ELAPSE project. The Lancet. Planetary Health, 6(1), e9–e18. https://doi.org/10.1016/S2542-5196(21)00277-1 van Donkelaar, A., Hammer, M. S., Bindle, L., Brauer, M., Brook, J. R., Garay, M. J., Hsu, N. C., Kalashnikova, O. V, Kahn, R. A., Lee, C., Levy, R. C., Lyapustin, A., Sayer, A. M., & Martin, R. V. (2021). Monthly Global Estimates of Fine Particulate Matter and Their Uncertainty. Environmental Science & Technology, 55(22), 15287–15300. https://doi.org/10.1021/acs.est.1c05309 Van Nunen, E., Vermeulen, R., Tsai, M. Y., Probst-Hensch, N., Ineichen, A., Davey, M., Imboden, M., Ducret-Stich, R., Naccarati, A., Raffaele, D., Ranzi, A., Ivaldi, C., Galassi, C., Nieuwenhuijsen, M., Curto, A., Donaire-Gonzalez, D., Cirach, M., Chatzi, L., Kampouri, M., … Hoek, G. (2017). Land Use Regression Models for Ultrafine Particles in Six European Areas. Environmental Science and Technology, 51, 3336–3345. https://doi.org/10.1021/acs.est.6b05920 Vienneau, D., de Hoogh, K., Bechle, M. J., Beelen, R., van Donkelaar, A., Martin, R. V, Millet, D. B., Hoek, G., & Marshall, J. D. (2013). Western European land use regression incorporating satellite- and ground-based measurements of NO2 and PM10. Environmental Science & Technology, 47(23), 13555–13564. https://doi.org/10.1021/es403089q Wang, M., Beelen, R., Bellander, T., Birk, M., Cesaroni, G., Cirach, M., Cyrys, J., de Hoogh, K., Declercq, C., Dimakopoulou, K., Eeftens, M., Eriksen, K. T., Forastiere, F., Galassi, C., Grivas, G., Heinrich, J., Hoffmann, B., Ineichen, A., Korek, M., … Brunekreef, B. (2014). Performance of multi-city land use regression models for nitrogen dioxide and fine particles. Environmental Health Perspectives, 122(8), 843–849. https://doi.org/10.1289/ehp.1307271 Wang, M., Brunekreef, B., Gehring, U., Szpiro, A., Hoek, G., & Beelen, R. (2016). A New Technique for Evaluating Land-use Regression Models and Their Impact on Health Effect Estimates. Epidemiology (Cambridge, Mass.), 27(1), 51–56. https://doi.org/10.1097/EDE.0000000000000404 World Health Organization. (2021). WHO global air quality guidelines: particulate matter (PM2.5 and PM10), ozone, nitrogen dioxide, sulfur dioxide and carbon monoxide. World Health Organization. https://apps.who.int/iris/handle/10665/345329 Supplementary Files Supplementarymaterial.docx Cite Share Download PDF Status: Published Journal Publication published 12 Dec, 2023 Read the published version in Environmental Science and Pollution Research → Version 1 posted Editorial decision: Minor Revision 17 Sep, 2023 Reviewers agreed at journal 29 Jun, 2023 Reviewers invited by journal 28 Jun, 2023 Editor invited by journal 27 Jun, 2023 Editor assigned by journal 06 Jun, 2023 First submitted to journal 31 May, 2023 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-2988847","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":213996870,"identity":"4187d116-290a-45e9-8bf5-f0739a900075","order_by":0,"name":"Laura A. Rodriguez-Villamizar","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAvklEQVRIiWNgGAWjYBACPgglIcfADKLZiNACVSNhTLIWhsQGFC5eLexnzD58bLNI33Cc/QHDh7LDDPztDQS08OQYz5zZJpG74TCPAeOMc4cZJM4cIOSwHGNmnjNgLQzMvG2HGQwkEgho4X8D1pJucJj9AfNforRIgGypkEgwACpmZiROy7NixhkVEoYzgX452HMunYegX/j5kzczfDCok+c7f/zhgx9l1nIEQwwFgIznIUH9KBgFo2AUjAJcAABIPzf+UahZ1gAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0002-5551-2586","institution":"Universidad Industrial de Santander","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"Laura","middleName":"A.","lastName":"Rodriguez-Villamizar","suffix":""},{"id":213996871,"identity":"3ffcee6a-6eca-47d7-a5ee-7b5b8954aef5","order_by":1,"name":"Yurley Rojas","email":"","orcid":"","institution":"Universidad Industrial de Santander","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Yurley","middleName":"","lastName":"Rojas","suffix":""},{"id":213996872,"identity":"e1f5acfd-1b9f-4e08-a684-d250ea4b3d57","order_by":2,"name":"Sara Grisales","email":"","orcid":"","institution":"Universidad de Antioquía: Universidad de Antioquia","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Sara","middleName":"","lastName":"Grisales","suffix":""},{"id":213996873,"identity":"c0f3c5e2-3e31-447f-9ac7-404698552943","order_by":3,"name":"Sonia C. Mangones","email":"","orcid":"","institution":"Universidad Nacional de Colombia","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Sonia","middleName":"C.","lastName":"Mangones","suffix":""},{"id":213996874,"identity":"918d3a3f-5c29-4653-8ad6-63082cd139e4","order_by":4,"name":"Jhon J. Cáceres","email":"","orcid":"","institution":"Universidad Industrial de Santander","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Jhon","middleName":"J.","lastName":"Cáceres","suffix":""},{"id":213996875,"identity":"ae918bde-467c-4122-8c14-d92cc3c222cb","order_by":5,"name":"Dayana M. Agudelo-Castañeda","email":"","orcid":"","institution":"Universidad del Norte","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Dayana","middleName":"M.","lastName":"Agudelo-Castañeda","suffix":""},{"id":213996876,"identity":"9dc4193d-f5aa-4f66-946c-2f7b08d5db26","order_by":6,"name":"Víctor Herrera","email":"","orcid":"","institution":"Universidad Industrial de Santander","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Víctor","middleName":"","lastName":"Herrera","suffix":""},{"id":213996877,"identity":"c7029a2e-598f-49fd-9b97-a8590ce29324","order_by":7,"name":"Diana Marín","email":"","orcid":"","institution":"Universidad Pontificia Bolivariana","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Diana","middleName":"","lastName":"Marín","suffix":""},{"id":213996878,"identity":"66f319e2-b5dc-4434-aaf9-0fa1ba8f8728","order_by":8,"name":"Juan G. Piñeros Jiménez","email":"","orcid":"","institution":"Universidad de Antioquía: Universidad de Antioquia","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Juan","middleName":"G. Piñeros","lastName":"Jiménez","suffix":""},{"id":213996879,"identity":"271142b1-a6a1-445a-b05a-94dd7075fb71","order_by":9,"name":"Luis C. Belalcázar-Ceron","email":"","orcid":"","institution":"Universidad Nacional de Colombia","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Luis","middleName":"C.","lastName":"Belalcázar-Ceron","suffix":""},{"id":213996880,"identity":"c76323cf-939b-46cf-9beb-bd18f0a6b8b9","order_by":10,"name":"Oscar Alberto Rojas-Sánchez","email":"","orcid":"","institution":"Instituto Nacional de Salud","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Oscar","middleName":"Alberto","lastName":"Rojas-Sánchez","suffix":""},{"id":213996881,"identity":"bab11fe1-ef08-4ece-8c84-208193b817de","order_by":11,"name":"Jonathan Ochoa Villegas","email":"","orcid":"","institution":"Universidad de San Buenaventura","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Jonathan","middleName":"Ochoa","lastName":"Villegas","suffix":""},{"id":213996882,"identity":"585a9060-47ee-48ac-ad6b-8137dbd2a9d1","order_by":12,"name":"Leandro López","email":"","orcid":"","institution":"Universidad Industrial de Santander","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Leandro","middleName":"","lastName":"López","suffix":""},{"id":213996883,"identity":"adafcbde-020a-4223-8a7e-52cfac7813ec","order_by":13,"name":"Oscar Mauricio Rojas","email":"","orcid":"","institution":"Area Metropolitana de Bucaramanga","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Oscar","middleName":"Mauricio","lastName":"Rojas","suffix":""},{"id":213996884,"identity":"226b2d0e-e6a6-4d53-88b1-f22a0cc2a9bb","order_by":14,"name":"María C. Vicini","email":"","orcid":"","institution":"Corporación para la Defensa de la Meseta de Bucaramanga","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"María","middleName":"C.","lastName":"Vicini","suffix":""},{"id":213996885,"identity":"7d75b163-bcce-404d-9a9c-d32a0fc1c6e7","order_by":15,"name":"Wilson Salas","email":"","orcid":"","institution":"Departamento Administrativo de Gestión del Medio Ambiente","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Wilson","middleName":"","lastName":"Salas","suffix":""},{"id":213996886,"identity":"6ff61d9c-de0f-4532-8eb8-c42cd93d6987","order_by":16,"name":"Ana Zuleima Orrego","email":"","orcid":"","institution":"Area Metropolitana Del Valle de Aburra","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Ana","middleName":"Zuleima","lastName":"Orrego","suffix":""},{"id":213996887,"identity":"a6a112de-0519-4de3-9dd3-994a1c31fa5f","order_by":17,"name":"Margarita Castillo","email":"","orcid":"","institution":"Barranquilla Verde","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Margarita","middleName":"","lastName":"Castillo","suffix":""},{"id":213996888,"identity":"10eccd2a-fd26-411c-83b2-510c5ed1d646","order_by":18,"name":"Hugo Sáenz","email":"","orcid":"","institution":"Secretaria Distrital de Ambiente de Bogota","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Hugo","middleName":"","lastName":"Sáenz","suffix":""},{"id":213996889,"identity":"4c578d21-fdc3-4f15-ab3e-7c0ea9f821c3","order_by":19,"name":"Luis Álvaro Hernández","email":"","orcid":"","institution":"Secretaria Distrito del Ambiente de Bogotá","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Luis","middleName":"Álvaro","lastName":"Hernández","suffix":""},{"id":213996890,"identity":"9a1b24cd-1706-407e-8dbb-37d7433eefdc","order_by":20,"name":"Scott Weichenthal","email":"","orcid":"","institution":"McGill University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Scott","middleName":"","lastName":"Weichenthal","suffix":""},{"id":213996891,"identity":"eb39098c-7ede-4c27-871f-8748ffc400b1","order_by":21,"name":"Jill Baumgartner","email":"","orcid":"","institution":"McGill University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Jill","middleName":"","lastName":"Baumgartner","suffix":""},{"id":213996892,"identity":"4c91d818-22b4-4f7e-968b-88cdf816f36f","order_by":22,"name":"Néstor Y. Rojas","email":"","orcid":"","institution":"Universidad Nacional de Colombia","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Néstor","middleName":"Y.","lastName":"Rojas","suffix":""}],"badges":[],"createdAt":"2023-05-27 11:00:01","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-2988847/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-2988847/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s11356-023-31306-w","type":"published","date":"2023-12-12T15:00:43+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":39383639,"identity":"628f76bf-7e1a-417c-8cbd-a8d537baee21","added_by":"auto","created_at":"2023-06-30 17:40:55","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":154527,"visible":true,"origin":"","legend":"\u003cp\u003eStudy areas and monitoring location within cities in Colombia\u003c/p\u003e\n\u003cp\u003eNote: Circles represent monitoring sites for both pollutants, PM\u003csub\u003e2.5\u003c/sub\u003e and NO\u003csub\u003e2,\u003c/sub\u003e and triangles represent monitoring sites for NO2 only.\u003c/p\u003e","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-2988847/v1/aa41228b39daa3ae73777f41.png"},{"id":39383640,"identity":"05c33677-ea6d-4bbd-a806-a9848e6fc3c3","added_by":"auto","created_at":"2023-06-30 17:40:55","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":261316,"visible":true,"origin":"","legend":"\u003cp\u003eAnnual predicted concentrations for PM\u003csub\u003e2.5\u003c/sub\u003e in five cities in Colombia, 2021\u003c/p\u003e","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-2988847/v1/a0d827dd0855369477fb0907.png"},{"id":39383638,"identity":"9ea00b9f-b683-489a-9223-baa4cb47ccb0","added_by":"auto","created_at":"2023-06-30 17:40:55","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":254851,"visible":true,"origin":"","legend":"\u003cp\u003eAnnual predicted concentrations for NO\u003csub\u003e2 \u003c/sub\u003ein five cities in Colombia, 2021\u003c/p\u003e","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-2988847/v1/9ab0838b0e708c80e0bb81ca.png"},{"id":48401353,"identity":"7d11c382-46a2-424b-a1b0-dd719cceeeb5","added_by":"auto","created_at":"2023-12-18 15:08:55","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2153986,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-2988847/v1/98b4e2d2-410c-4d06-9303-1acd9eee1d0d.pdf"},{"id":39383641,"identity":"f59b579b-fc92-44d2-aacc-5844734a03bc","added_by":"auto","created_at":"2023-06-30 17:40:55","extension":"docx","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":676726,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementarymaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-2988847/v1/a04cfc44e3e8077e2799b827.docx"}],"financialInterests":"","formattedTitle":"Intra-urban variability of long-term exposure to PM2.5 and NO2 in five cities in Colombia","fulltext":[{"header":"Introduction","content":"\u003cp\u003eAir pollution is recognized as one of the leading environmental risk factors for population health (GBD 2019 Risk Factors Collaborators, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). It is estimated that 99% of the world population is living in places where air pollution levels for fine particulate matter (PM\u003csub\u003e2.5\u003c/sub\u003e) exceed the current safe guideline level defined by the World Health Organization (WHO), and populations from low- and middle-income countries are exposed to the highest levels (World Health Organization, \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). In 2019, it was estimated that a total of 2.92\u0026nbsp;million deaths in females and 3.75\u0026nbsp;million deaths in males were attributable to ambient particulate matter and ozone air pollution. For Latin America and the Caribbean (LAC) region, and overall for low- and low-middle income countries, air pollution was the second most important risk factor (after malnutrition) that accounted for attributable disability-adjusted life-years (DALYs) rates over the past decade (GBD 2019 Risk Factors Collaborators, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eParticulate matter (PM\u003csub\u003e2.5\u003c/sub\u003e), nitrogen dioxide (NO\u003csub\u003e2\u003c/sub\u003e), and ozone (O\u003csub\u003e3\u003c/sub\u003e) are the ambient air pollutants most strongly associated with adverse health adverse effects in the short- and long-term (World Health Organization, \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). The health effects from long-term exposure to air pollution are 10-fold higher than the short-term effects represented by daily variations (Pope, \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2007\u003c/span\u003e). For long-term exposure there is also evidence that there are large within-city contrasts and their effects are probably higher than the effects related to variations between cities (Crouse et al., \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Therefore, high-resolution spatial estimations of long-term exposure to air pollutants, particularly in urban setting, are critical for epidemiological research studying the association between air pollution and health and an important input for air quality management plans aimed to reduce air pollution adverse effects (Fann et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). There are different methods for estimating intraurban spatial variability of air pollutants. These methods include models based on proximity to monitoring stations, interpolation methods, land use regression models (LUR), and dispersion and chemical transport models combined with satellite remote sensing (Dijkema et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; Hoek, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Hoek et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2008\u003c/span\u003e; Michael Jerrett et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2005\u003c/span\u003e; van Donkelaar et al., \u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eLUR models combined monitoring of air pollutants with the development of stochastic models using physical landscape characteristics, meteorology and population as predictor variables (Hoek et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). LUR models have lower computational requirements compared with dispersion or chemical transport models and are relatively easy to implement using Geographic Information Systems (GIS), which made them a method of preference in developing intraurban surfaces of air pollutant exposure (Hoek et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). LUR models have shown to have a high predictive value and to be a cost-effective method to estimate intraurban variations of air pollutants in different regions including North America, Europe and Asia (Allen et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; Chen et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; de Hoogh et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Eeftens et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Gurung et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Kashima et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Lee et al., \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Stafoggia et al., \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Recently LUR has been used in these regions as input data for hybrid models combining dispersion models, satellite-based observations, land use, and surface monitoring data for PM\u003csub\u003e2.5\u003c/sub\u003e and NO\u003csub\u003e2\u003c/sub\u003e (Hoek, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). Also, annual and monthly global estimates of ground level PM\u003csub\u003e2.5\u003c/sub\u003e and NO\u003csub\u003e2\u003c/sub\u003e have been developed, combining satellite remote sensing with the GEOS-Chem chemical transport model and calibration using ground-level observations (van Donkelaar et al., \u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). These models provide spatially fine resolutions at 0.01\u0026deg; \u0026times; 0.01\u0026deg;and have shown to have a very good performance in North America and Europe but have very high uncertainty for tropical areas particularly in South America (Hoek, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; van Donkelaar et al., \u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eDespite LAC cities are growing rapidly and experiencing high levels of air pollution, the estimates of long-term exposure to air pollution are scarce in the region. In most cities, the ground-level measurements of atmospheric pollutants have poor consistency and coverage (Cunha-Zeri \u0026amp; Ometto, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Limitations include that traditional air quality stations require high financial funding in resource-limited countries which make them logistically prohibitive since it is not cost-effective. Consequently, given the limited resources of good air quality data, modeling emerges as a possible tool to derive management measures (Agudelo-Casta\u0026ntilde;eda et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). However, high-resolution spatial estimations of long-term exposure to air pollutants are scarce in LAC and development of LUR models for some pollutants have been reported only for the cities of Mexico, Sao Paulo, Quito, and Medell\u0026iacute;n (Alvarez-Mendoza et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Habbermann \u0026amp; Gouveia, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2007\u003c/span\u003e; Londo\u0026ntilde;o \u0026amp; Ca\u0026ntilde;on, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Luminati et al., \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Son et al., \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e2018\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eColombia is located at the extreme north of South America with an estimated population of 52\u0026nbsp;million inhabitants (Departamento Nacional de Estad\u0026iacute;stica (DANE), \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2020b\u003c/span\u003e) distributed across 32 departments and 1122 municipalities. The national air quality surveillance network has operated since 1993 and currently includes 22 regional surveillance systems that are distributed in 77 municipalities of 19 departments. In 2021, the national monitoring network included 131 monitoring stations for PM\u003csub\u003e2.5\u003c/sub\u003e and 57 for NO\u003csub\u003e2\u003c/sub\u003e (Instituto de Hidrologia Meteorolog\u0026iacute;a y Estudios Ambientales-IDEAM, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Data from monitoring stations provide useful information for temporal daily variations of pollutants but provide limited information on the spatial variability of pollution especially in densely populated urban settings that concentrate 77% of the country\u0026acute;s population (Departamento Nacional de Estad\u0026iacute;stica (DANE), \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2020b\u003c/span\u003e). Data from monitoring surveillance systems have been used in epidemiological studies assessing the short-term effects of pollutant concentrations on mortality and morbidity in the largest cities in Colombia (Blanco-Becerra et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Rodriguez-Villamizar et al., \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). However, there is a need for estimations of long-term spatial variation of pollutants within cities. Therefore, our objective was to develop intraurban LUR models for PM\u003csub\u003e2.5\u003c/sub\u003e and NO\u003csub\u003e2\u003c/sub\u003e in the five largest cities in Colombia to estimate of long-term population exposure to air pollution for use in air quality health assessment and mitigation.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy areas\u003c/h2\u003e \u003cp\u003eThe study was conducted in the urban areas of the five largest cities in Colombia: Barranquilla, Bucaramanga, Bogot\u0026aacute;, Cali, and Medell\u0026iacute;n (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The population varies across cities, Bogot\u0026aacute; being the most populated city with an estimated population of 7,834,167\u0026nbsp;million inhabitants in 2021. The estimated total population during 2021 was 1,297,082 for Barranquilla, 614,269 for Bucaramanga, 2,264,748 for Cali, and 2,573,220 inhabitants for Medell\u0026iacute;n (Departamento Nacional de Estad\u0026iacute;stica (DANE), \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2020b\u003c/span\u003e). The altitude and average temperature also vary across cities, with Barranquilla being the warmer and closest to the sea level and Bogot\u0026aacute; being the coldest and highest elevation. The physical characteristics of these cities are presented in Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e in Supplementary material. Similar to other capital cities in South America, the roadways networks in these cities are complex and dense, and both industrial and residential neighborhoods coexist.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eAir pollution measurement data\u003c/h2\u003e \u003cp\u003ePM\u003csub\u003e2.5\u003c/sub\u003e and NO\u003csub\u003e2\u003c/sub\u003e concentrations were measured in the five cities for two consecutive weeks during both the dry and the rainy season in 2021. The selection of the dry and rainy seasons for each city was defined based on the total precipitation registered in local meteorological stations between 2010 to 2019. The driest months correspond to January to March while the months with higher precipitation were April to May for most cities. The details of the sampling period for each city are presented in Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e in Supplementary material.\u003c/p\u003e \u003cp\u003eFor NO\u003csub\u003e2\u003c/sub\u003e there were 80 sampling sites for Bogot\u0026aacute; and 40 for the other cities while for PM\u003csub\u003e2.5\u003c/sub\u003e there were 40 sampling sites for Bogot\u0026aacute; and 20 for the other cities. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e shows the location of sampling sites distributed across the urban area of the cities. The density of sampling sites in the urban areas for NO\u003csub\u003e2\u003c/sub\u003e measurements (samplers per km\u003csup\u003e2\u003c/sup\u003e) was 2.3 for Barranquilla, 0.8 for Bucaramanga, 4.4 for Bogot\u0026aacute;, 3.5 for Cali and 3.6 for Medell\u0026iacute;n; the density of sampling for PM\u003csub\u003e2.5\u003c/sub\u003e was twice these values as we used half the number of monitors. The selection of sampling sites was conducted with participation of the study team and experts from the environmental and health departments of each city. The criteria for selecting the monitoring sites included: 1) the representation of traffic, residential, industrial or other areas within the cities, and 2) the heterogeneity in the characteristics of the selected sites (i.e., in terms of types of traffic, density of residential areas or particular areas for cities such as port or industrial areas). The sampling sites included one background urban site per city. The background site was located in the area of the city with the lowest concentrations of pollutants based on measurements, if they were available, or based on the experts\u0026rsquo; knowledge of pollution within the city. In addition, sampling included 3\u0026ndash;4 sites per city that were installed in the same location as monitoring stations from the local air quality network to facilitate instrument intercomparisons. For quality control two blank filters were used for each city.\u003c/p\u003e \u003cp\u003eMeasurement campaigns were simultaneously conducted across all sampling sites in each city for two weeks. Two trained teams of field staff were responsible for installing and uninstalling monitoring samplers across the study cities. We measured gravimetric PM\u003csub\u003e2.5\u003c/sub\u003e using Ultrasonic Personal Aerosol Sampler (UPAS) samplers (V2.0 Access Sensor Technologies, Fort Collins, Colorado, USA) that were installed between 2.5\u0026ndash;3 meters above ground in all monitoring sites. The UPAS monitors have been widely used for measuring gravimetric PM\u003csub\u003e2.5\u003c/sub\u003e in similar and higher pollution settings (Arku et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) and have shown good performance for collecting airborne PM for gravimetric analysis (Leith et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). We adapted an environmental enclosure to protect the device during outdoor sampling and added an external battery to increase the sampling time to 7 days at 25% duty cycle at a flow rate of 1 lpm. Each monitor was loaded with a 37mm Teflon filter at the start of each measurement period. We replaced the UPAS and filters at each sampling site after 7 days to complete the two weeks monitoring period. Gravimetric analysis was conducted for all cities in a single laboratory certified for this competence (ISO/IEC 17025:1999) by the Instituto de Hidrolog\u0026iacute;a, Meteorolog\u0026iacute;a y Estudios Ambientales (IDEAM). Each filter and blank were weighted three times and the average measurement was reported for each filter. The reported limit of quantification was 0.68 \u0026#120525;g and the limit of detection was 1.36 \u0026#120525;g. The average PM\u003csub\u003e2.5\u003c/sub\u003e concentration of the two weekly filters from the same site and campaign was reported as the site concentration for statistical analysis.\u003c/p\u003e \u003cp\u003eFor measuring NO\u003csub\u003e2\u003c/sub\u003e we used passive diffusion Palmes Tubes (Gradko environmental, Hampshire, UK) that were installed for two weeks with a height of 2.5\u0026ndash;3 meters above ground in all monitoring sites. For quality control an extra two blank tubes were deployed in each city. The processing of all tubes was conducted in the manufacturers laboratory and concentration measurements were reported as the average of duplicate measurements. The reported limit of detection was 0.031 \u0026#120525;g of NO\u003csub\u003e2\u003c/sub\u003e in tubes. The installation, operation, and deinstallation of the PM\u003csub\u003e2.5\u003c/sub\u003e and NO\u003csub\u003e2\u003c/sub\u003e monitoring devices including refrigeration of samples was conducted by trained personnel following the manufacturer's instructions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eGIS predictor variables\u003c/h2\u003e \u003cp\u003ePredictor variables were grouped into five categories: 1) land use (areas of different land uses); 2) population (including population counts and population density); 3) roads (including total length of roads and distance from sampling sites to arterial roads); 4) traffic (including estimated average speed and traffic volume); 5) physical geography (altitude); and 6) meteorology (including average temperature, precipitation, relative humidity, and wind direction). All predictor variables were created for circular buffers with radii of 100m, 200m, and 500m and centered at the monitoring sites. These predictor variables were obtained from the intersection between buffers and GIS layers. In total, 78 independent variables were generated including variations of roadways variables. Maps were created using ESRI ArcGIS\u0026reg; 10.8.1 and ArcMap\u0026trade; under license (ESRI\u0026reg; version, US). Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e provides the details of the predictor variables used to generate the LUR models.\u003c/p\u003e \u003cp\u003eLand use data were obtained from the local government's planning office based on the most recent land use distribution available. Altitude was measured in sampling sites directly using an altimeter during the first deployment of monitoring devices. Population data and roads classification were obtained from the demographic and cartographic information of the census 2018 (Departamento Nacional de Estad\u0026iacute;stica (DANE), \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2020a\u003c/span\u003e). Meteorology data were obtained from meteorological monitoring stations from the local environmental authority including 16 stations in Bogot\u0026aacute;, 22 stations in Medell\u0026iacute;n and 8 stations in Cali. Precipitation and temperature raster surfaces were calculated using the Regnie model (Rauthe et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Briefly, we used data from stations coupled with altitude from the digital terrain model (DTM) with 30 meters resolution, the slope and land exposure (the direction or azimuth angle of the inclination of the slope) to calculate spatial precipitation and temperature mean values using a linear regression model. Barranquilla and Bucaramanga had less than four local meteorological stations that did not allow for a valid spatial estimation and therefore meteorological data was not included in LUR models for these two cities.\u003c/p\u003e \u003cp\u003eTraffic predictor variables were measured and estimated for the project. The traffic speed measurements were obtained during the same monitoring campaigns periods by using a cloud -based data method that included data pre-processing, speed computation and output data formatting. During the pre-processing, the street network vector data from Openstreetmap was edited to match the same network used by the Google Maps platform. Then, the network streets were split into 100-meter links considering the road intersections setup. Then, the speed was computed for those links using their length and travel time. Travel times at the link level were obtained from the Google Maps platform using the Distance Matrix API service, which provides predicted values at the time the service was used. Finally, the speed of each link was added to its attributes set, and the whole collection of links were used to create a GIS layer using Python scripts.\u003c/p\u003e \u003cp\u003eTo estimate traffic volumes, we used speed-density-flow functions, which describe the relationships between traffic speed, density, and flow rate on a road segment. These functions were obtained and used to estimate traffic conditions (Council, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). We computed speed-density-flow functions for urban traffic for Bogota (73 road segments) and Medell\u0026iacute;n (199 road segments) using data from sensors and traffic cameras provided by the transportation authorities. We computed and validated the functions for three different traffic regimes: interrupted, semi-interrupted, and uninterrupted flow. We tested six theoretical functional forms (Greenshields, Drew, Pipes, May\u0026amp;Keller, Greenberg and Underwood Model) (Gaddam \u0026amp; Rao, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) by using random sampling with replacement. The best model was selected based on the root mean square error (RMSE). The resulting functional forms were then used to estimate traffic volumes in the road network of Barranquilla, Cali, and Bucaramanga, taking into account the traffic regimes, and the number of lanes in each road segment.\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\u003eLand Use Regression Predictor variables\u003csup\u003ea\u003c/sup\u003e\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=\"char\" char=\".\" 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\u003eCategory/\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSource\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eYear\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eUnit\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eVariable names\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLand use\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLand use plan for each city\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2014\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e% of land use type (square meters)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eIndustrial-IND\u003c/p\u003e \u003cp\u003eResidencial-RES\u003c/p\u003e \u003cp\u003eDotacional-DOT\u003c/p\u003e \u003cp\u003eCentral-CEN\u003c/p\u003e \u003cp\u003eCommercial-COM\u003c/p\u003e \u003cp\u003ePort-PORT\u003c/p\u003e \u003cp\u003eMixed-MIX\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePopulation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNational Census\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2018\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003epeople per square meter\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eTotal population-POB\u003c/p\u003e \u003cp\u003ePopulation Density-DEN\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRoads\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOpen Street Maps\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2020\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ekilometers\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eLength by road type\u003c/p\u003e \u003cp\u003eTrunk road-TRUNK\u003c/p\u003e \u003cp\u003ePrimary road-PRIM\u003c/p\u003e \u003cp\u003eSecondary road-SEC\u003c/p\u003e \u003cp\u003eTertiary road-TER\u003c/p\u003e \u003cp\u003eLocal road-LOC\u003c/p\u003e \u003cp\u003eDistance from site to road by road type\u003c/p\u003e \u003cp\u003eDTRUNK\u003c/p\u003e \u003cp\u003eDPRIM\u003c/p\u003e \u003cp\u003eDSEC\u003c/p\u003e \u003cp\u003eDTER\u003c/p\u003e \u003cp\u003eDLOC\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTraffic\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSpeed Distance Matrix API - Google\u003c/p\u003e \u003cp\u003eSpeed-density-flow functions\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2021\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ekilometers per hour\u003c/p\u003e \u003cp\u003eVehicles per hour\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eTraffic Speed- VEL\u003c/p\u003e \u003cp\u003eTraffic volume- VOL\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePhysical geography\u003c/p\u003e \u003cp\u003eAltitude\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2021\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMeters Above Sea Level (MASL)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAltitude-ALT\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMeteorology\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMonitoring Station\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2021\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTemperature \u0026ordm;C\u003c/p\u003e \u003cp\u003ePrecipitation mm\u003c/p\u003e \u003cp\u003eRelative Humidity %\u003c/p\u003e \u003cp\u003eWind direction\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eTemperature TPROM\u003c/p\u003e \u003cp\u003ePrecipitation PPROM\u003c/p\u003e \u003cp\u003eHumidity (%) HR\u003c/p\u003e \u003cp\u003eWind direction WD\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 \u003csup\u003ea\u003c/sup\u003eAll predicted variables were created for buffers of 100m, 200m, and 500m.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eWe averaged pollutants' concentrations measured during both sampling campaigns to obtain annual means for each city. The comparison of measurements of the PM\u003csub\u003e2.5\u003c/sub\u003e sampling device with local monitoring stations was conducted for 13 monitoring stations with data available (2 in Barranquilla, 4 in Bogot\u0026aacute;, 4 in Cali and 3 in Medell\u0026iacute;n). Comparison of concentrations were evaluated using Bland and Altman agreement coefficients and graphs (Bland \u0026amp; Altman, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e1986\u003c/span\u003e). The average annual measurements across the monitoring sites were also compared to the average annual estimation measurements from the real-time local monitoring stations in the cities.\u003c/p\u003e \u003cp\u003eWe developed LUR models to estimate intraurban spatial variation of PM\u003csub\u003e2.5\u003c/sub\u003e and NO\u003csub\u003e2\u003c/sub\u003e within the five cities. We used multivariable spatial regression models, that allow local estimations of a dependent variable \u0026#119885;, by implementing the Ordinary Least Squares (OLS) method, in the presence of possible explanatory variables (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\({\\text{Z}}_{\\text{j}}\\)\u003c/span\u003e\u003c/span\u003e) at the same point \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\left({\\text{x}}_{\\text{i}},{\\text{y}}_{\\text{i}}\\right)\\)\u003c/span\u003e\u003c/span\u003e represented by the following equation (Londo\u0026ntilde;o, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Maantay \u0026amp; McLafferty, \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2011\u003c/span\u003e):\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\text{Z}({\\text{x}}_{\\text{i}},{\\text{y}}_{\\text{i}})={{\\beta }}_{0}+{\\sum }_{\\text{j}=1}^{\\text{n}}{{\\beta }}_{\\text{i}}{\\text{Z}}_{\\text{j}}({\\text{x}}_{\\text{i}},{\\text{y}}_{\\text{i}})+{{\\epsilon }}_{\\text{j}} , {{\\epsilon }}_{\\text{j}}\\sim\\text{N}(0,{\\text{v}}^{2})$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eTo represent the spatial dependency structure between the features being analyzed, the best combination of explanatory variables must be determined. In a first step, we removed highly correlated variables (\u0026gt;\u0026thinsp;0.7) and those variables in which zero values account for more than 90% of the sampling sites. Then, all the predictors are included in the model assessing their statistical significance (p value\u0026thinsp;\u0026lt;\u0026thinsp;0.05) and the sign for their coefficient (\u0026#120573;\u0026#119894;) (observing their agreement with the expected -theoretical. direction of effect). In addition, the selected variables must adequately specify the regression model, by evaluating the specification criteria of the OLS method. We estimated the adjusted R-squared to assess the performance of the models and the variance inflation factor to determine multicollinearity. All models were built with a combination of all the buffer variables (Eeftens et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Van Nunen et al., \u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e2017\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eWe performed a Geographically Weighted Regression (GWR) with the selected equation to examine the spatial heterogeneity of the relationship between air pollutants and other spatial variables and to estimate the multiple regression model parameters. Then, we created a regular point mesh with cells spaced by 200m over the cities\u0026acute; surface, where the formula obtained by each annual regression model was applied, in order to predict air pollutant levels for each point. Then, a spatial interpolation method (spline) was applied to obtain the concentration surface of the pollutant in the study area. Finally, we performed a leave-one-out cross validation (Eeftens et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Wang et al., \u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e2016\u003c/span\u003e) for each LUR model in each city and compared the set of predicted values against the observed ones. Then, the cross-validated square error and R\u003csup\u003e2\u003c/sup\u003e were calculated for each model. The cross-validation was conducted using the \u0026ldquo;loocv\u0026rdquo; command in Stata\u0026reg; version 13 (Stata Corporation).\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003ePollutants\u0026acute; concentrations at sampling locations\u003c/h2\u003e \u003cp\u003eThere were 116 PM\u003csub\u003e2.5\u003c/sub\u003e sampling sites with valid measurements for both monitoring campaigns used for the estimation of the annual average concentrations. Three sites in Cali, four sites in Bogot\u0026aacute;, and one in Medellin were excluded because they contribute only one successful measurement. The mean PM\u003csub\u003e2.5\u003c/sub\u003e concentrations during the dry season were slightly higher compared to the rainy season (see supplementary material Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). The annual PM\u003csub\u003e2.5\u003c/sub\u003e mean concentration and range in sampling sites were 16.12 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e (7.42\u0026ndash;22.22) for Medell\u0026iacute;n, 15.90 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e (3.64\u0026ndash;35.30) for Barranquilla, 15.79 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e (4.86\u0026ndash;32.69) for Cali, 13.89 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e (4.39\u0026ndash;25.52) for Bogot\u0026aacute;, and 12.93 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e (4.90-32.23) for Bucaramanga.\u003c/p\u003e \u003cp\u003eFor NO\u003csub\u003e2\u003c/sub\u003e sampling, 17 out of the 240 tubes deployed were removed due to vandalism or invalid measurements, leaving 223 observations for the analyses. The mean NO\u003csub\u003e2\u003c/sub\u003e concentrations during the dry season were slightly higher than those in the rainy season (see supplementary material Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). The annual NO\u003csub\u003e2\u003c/sub\u003e mean concentration and range in sampling sites were 49.09 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e (32.38\u0026ndash;68.31) for Medell\u0026iacute;n, 34.92 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e (12.56\u0026ndash;64.67) for Bucaramanga, 39.12 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e (13.52\u0026ndash;69.89) for Cali, 34.63 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e (5.09\u0026ndash;52.19) for Bogot\u0026aacute;, and 24.92 \u0026#120525;/m3 (7.38\u0026ndash;51.81) for Barranquilla.\u003c/p\u003e \u003cp\u003eThe average of the differences in PM\u003csub\u003e2.5\u003c/sub\u003e concentrations measured using the UPAS and those reported during the same sampling period by local monitoring stations was \u0026minus;\u0026thinsp;1.5 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e (95%CI -6.8 to 3.9) during the dry season campaign (11 monitoring stations) and \u0026minus;\u0026thinsp;0.05 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e (95% CI -11.5 to 11.4) during the rainy season campaign (13 monitoring stations). During the dry season campaign, higher differences were observed for two local monitoring stations, one in Medell\u0026iacute;n and one in Cali. During the rainy season campaign, higher differences were observed for the three local monitoring stations from Medell\u0026iacute;n. Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e shows the levels of agreement for PM\u003csub\u003e2.5\u003c/sub\u003e measurements during the two monitoring campaigns. There was only one monitoring station in downtown Medell\u0026iacute;n with valid NO\u003csub\u003e2\u003c/sub\u003e data for comparison of measurements obtained from Palmes tubes and local monitors. For this site-station pair the differences were 5.71 and 2.59 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e during the dry and rainy season, respectively. In Bogot\u0026aacute; during the second campaign (rainy season) there were four sites with valid paired measurements whose average difference was 6.70 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e, which was highly influenced by the discrepancy observed in one particular station located at Carrera 7a (excluding this station the average of the difference was 2.86 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e). The comparison of the PM\u003csub\u003e2.5\u003c/sub\u003e average campaign\u0026rsquo;s measurements from monitoring sites with the average annual measurements from monitoring stations during 2021 resulted in differences of -0.84 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e for Bucaramanga, -1.1 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e for Medell\u0026iacute;n, -1.7 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e for Bogot\u0026aacute;, 1.4 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e for Cali, and 1.7 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e for Barranquilla. For NO\u003csub\u003e2\u003c/sub\u003e the difference between passive samplers and monitoring stations in Bogot\u0026aacute; was 5.6 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eLUR models\u003c/h2\u003e \u003cp\u003eThe final LUR models selected for the cities explained higher variability for PM\u003csub\u003e2.5\u003c/sub\u003e compared with NO\u003csub\u003e2\u003c/sub\u003e. (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, respectively). The models for PM\u003csub\u003e2.5\u003c/sub\u003e explained between 44% (Bogot\u0026aacute;) and 82% (Medell\u0026iacute;n) of pollutant\u0026acute;s spatial variability within cities. Most models showed a RMSE of approximately 1.5 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e except for Barranquilla where the error was approximately 4 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e. The contrasts between PM\u003csub\u003e2.5\u003c/sub\u003e measured and predicted concentrations at monitoring sites for all cities are presented in Supplementary material Figure S2. Most of the predictor variables included in the PM\u003csub\u003e2.5\u003c/sub\u003e LUR models were related to specific types of land uses and roadways\u0026acute; attributes with predominance of 200 and 500m buffers. In Bucaramanga the LUR model only included roadways variables while Medell\u0026iacute;n was the only city where the model included a meteorological variable (see Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). There was no evidence of multicollinearity in the LUR models for both pollutants as the VIF values were all below 2.1. The maps of the predicted concentrations for PM\u003csub\u003e2.5\u003c/sub\u003e in the urban areas of the five cities are presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e \u003cp\u003eThe final selected models for NO\u003csub\u003e2\u003c/sub\u003e explained between 30% (Barranquilla) and 65% (Bucaramanga) of the pollutant\u0026acute;s spatial variability within cities. Most cities models showed a RMSE around 6 to 8 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e except for Cali where the error was close to 1.5 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e. The measured values versus the predicted values of the models in the monitoring sites for NO\u003csub\u003e2\u003c/sub\u003e in all cities are presented in Supplementary material Figures S3. As expected, most of the predictor variables included in the NO\u003csub\u003e2\u003c/sub\u003e LUR models were a combination of roadways variables with different buffers. In Bucaramanga the LUR model included population variables and in Medell\u0026iacute;n one meteorological variable (see Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). There was no collinearity in the LUR models for both pollutants as the VIF values were all below 1.7. The maps of the predicted concentrations for NO\u003csub\u003e2\u003c/sub\u003e in the urban areas of the five cities are presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eCross validation\u003c/h2\u003e \u003cp\u003eOverall, the leave-one-out cross-validation R\u003csup\u003e2\u003c/sup\u003es showed good stability, particularly for PM\u003csub\u003e2.5\u003c/sub\u003e. For PM\u003csub\u003e2.5\u003c/sub\u003e, the difference between the model R\u003csup\u003e2\u003c/sup\u003e and the validation R\u003csup\u003e2\u003c/sup\u003e was 19% for Barranquilla, 31% for Bucaramanga, 6% for Bogot\u0026aacute;, 19% for Cali and 3% for Medell\u0026iacute;n. For NO\u003csub\u003e2\u003c/sub\u003e, the difference between the model R\u003csup\u003e2\u003c/sup\u003e and the validation R\u003csup\u003e2\u003c/sup\u003e was 11% for Barranquilla, 10% for Bucaramanga, 6% for Bogot\u0026aacute;, 8% for Cali and 12% for Medell\u0026iacute;n. Validation R\u003csup\u003e2\u003c/sup\u003es are presented in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e for PM\u003csub\u003e2.5\u003c/sub\u003e and NO\u003csub\u003e2\u003c/sub\u003e, respectively.\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\u003eDescription of developed LUR models for PM\u003csub\u003e2.5\u003c/sub\u003e in five cities in Colombia, 2021\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\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=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLUR model\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNo.\u003c/p\u003e \u003cp\u003esites\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eModel R\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eRMSE\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eVIF\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eR\u003csup\u003e2\u003c/sup\u003e cross validation\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBarranquilla\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePM\u003csub\u003e2.5\u003c/sub\u003e= 19.83344\u0026ndash;0.1489524*ALT \u0026minus;\u0026thinsp;0.0230902*DTRON500\u0026thinsp;+\u0026thinsp;44.43591*IND200\u0026thinsp;+\u0026thinsp;21.93109*CEN500\u0026thinsp;+\u0026thinsp;23.10317*PORT500\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.73\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e3.98\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.54\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBogot\u0026aacute;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLn(PM\u003csub\u003e2.5\u003c/sub\u003e)\u0026thinsp;=\u0026thinsp;2.4713 +\u003c/p\u003e \u003cp\u003e3.1439*DEN100\u0026thinsp;+\u0026thinsp;1.8045*IND200\u0026ndash;0.8418*RES500\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.44\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.39\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.63\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.38\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBucaramanga\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLn(PM\u003csub\u003e2.5\u003c/sub\u003e)\u0026thinsp;=\u0026thinsp;2.199057\u0026thinsp;+\u0026thinsp;0.0014062\u0026lowast;SEC100 + 0.0000327\u0026lowast;LOC500\u003c/p\u003e \u003cp\u003e\u0026minus;\u0026thinsp;0.0012659\u0026lowast;DPRIM500 + 0.0215501\u0026lowast;VEL100 \u0026minus;\u0026thinsp;0.000242\u0026lowast;VOL200\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.23\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.46\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCali\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLn(PM\u003csub\u003e2.5\u003c/sub\u003e)\u0026thinsp;=\u0026thinsp;2.3387\u0026thinsp;+\u0026thinsp;0.00001\u0026lowast;DOT200\u0026thinsp;+\u0026thinsp;1.0713\u0026lowast;PRIM200\u0026thinsp;+\u0026thinsp;0.5943\u0026lowast;SEC200\u0026ndash;0.0004\u0026lowast;VOL100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e17\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.70\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e2.06\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.51\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedell\u0026iacute;n\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePM\u003csub\u003e2.5\u003c/sub\u003e = 13.77207\u0026ndash;1.357455\u0026lowast;PPROM \u0026minus;\u0026thinsp;5.589831\u0026lowast;DOT100 + 2.269679\u0026lowast;DEN200 + 70.23039\u0026lowast;MIX500 + 0.0043842\u0026lowast;VOL500\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.71\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.48\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.79\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"7\"\u003eRMSE: Root mean square error; VIF: Variance inflation factor\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eALT\u0026thinsp;=\u0026thinsp;Altitude; CEN\u0026thinsp;=\u0026thinsp;Central land use; DEN\u0026thinsp;=\u0026thinsp;population density; DOT\u0026thinsp;=\u0026thinsp;Dotacional land use; DPRIM\u0026thinsp;=\u0026thinsp;Distance to primary roadway; DTRON\u0026thinsp;=\u0026thinsp;Distance to trunk roadway; IND\u0026thinsp;=\u0026thinsp;Industrial land use; LOC\u0026thinsp;=\u0026thinsp;Length local roadways; MIX: mixed land use; PORT\u0026thinsp;=\u0026thinsp;Port land use; PPROM\u0026thinsp;=\u0026thinsp;precipitation average; PRIM\u0026thinsp;=\u0026thinsp;Length primary roadways; RES\u0026thinsp;=\u0026thinsp;Residential land use; SEC\u0026thinsp;=\u0026thinsp;Length secondary roadways; VEL\u0026thinsp;=\u0026thinsp;Vehicular speed; VOL\u0026thinsp;=\u0026thinsp;vehicular volume. Numbers correspond to buffers of 100m, 200m, 500m.\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\u003eDescription of developed LUR models NO\u003csub\u003e2\u003c/sub\u003e in five cities in Colombia, 2021\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\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=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLUR model\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNo. sites\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eModel R\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eRMSE\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eVIF\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eR\u003csup\u003e2\u003c/sup\u003e cross validation\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBarranquilla\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNO\u003csub\u003e2\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;12.89591\u0026thinsp;+\u0026thinsp;25.45936\u0026lowast;PRIM100\u0026ndash;0.1583713\u0026lowast;VEL100\u0026thinsp;+\u0026thinsp;0.0061518\u0026lowast;VOL500\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e36\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e8.02\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.19\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBogot\u0026aacute;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLn(NO\u003csub\u003e2\u003c/sub\u003e)\u0026thinsp;=\u0026thinsp;2.8714\u0026thinsp;+\u0026thinsp;0.0001*PRIM500\u0026thinsp;+\u0026thinsp;0.0058*VEL100\u0026thinsp;+\u0026thinsp;0.2599*WPROM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e73\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.26\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.34\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBucaramanga\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNO\u003csub\u003e2\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;13.00243\u0026thinsp;+\u0026thinsp;291.5302\u0026lowast;DEN100 \u0026minus;\u0026thinsp;0.0013283\u0026lowast;POB500 + 0.0025503\u0026lowast;TER500\u0026thinsp;+\u0026thinsp;0.0020514\u0026lowast;LOC500 + 0.0057464\u0026lowast;VOL100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e8.08\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.66\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.55\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCali\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLn(NO\u003csub\u003e2\u003c/sub\u003e)\u0026thinsp;=\u0026thinsp;3.47834312\u0026thinsp;+\u0026thinsp;0.49126931*PRIM200\u003c/p\u003e \u003cp\u003e+\u0026thinsp;0.39823891\u0026lowast;SEC200\u0026thinsp;+\u0026thinsp;0.36505469\u0026lowast;TER200\u0026ndash;0.01475995\u0026lowast;VEL200\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.44\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.36\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedell\u0026iacute;n\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNO\u003csub\u003e2\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;46.06516\u0026ndash;3.625967\u0026lowast;PPROM \u0026minus;\u0026thinsp;0.0299678\u0026lowast;DSEC200 + 0.0225605\u0026lowast;VOL500\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e34\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.57\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e5.53\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.06\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.45\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"7\"\u003eRMSE: Root mean square error; VIF: Variance inflation factor\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eDEN\u0026thinsp;=\u0026thinsp;population density; DOT\u0026thinsp;=\u0026thinsp;Dotacional land use; DSEC\u0026thinsp;=\u0026thinsp;Distance to secondary roadway; LOC\u0026thinsp;=\u0026thinsp;Length local roadways; POB\u0026thinsp;=\u0026thinsp;population size; PPROM\u0026thinsp;=\u0026thinsp;precipitation average; PRIM\u0026thinsp;=\u0026thinsp;Length primary roadways; SEC\u0026thinsp;=\u0026thinsp;Length secondary roadways; TER\u0026thinsp;=\u0026thinsp;Length tertiary roadways; VEL\u0026thinsp;=\u0026thinsp;Vehicular speed; VOL\u0026thinsp;=\u0026thinsp;vehicular volume, WPROM: Wind speed (mean). Numbers correspond to buffers of 100m, 200m, 500m.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis is the first study to develop LUR models for multiple cities in a Latin American country, providing small-area estimations of air pollutants for use in health risk assessments, epidemiological studies of long-term exposure to air pollution and mitigation evaluation. The development of LUR models to estimate concentrations for PM\u003csub\u003e2.5\u003c/sub\u003e and NO\u003csub\u003e2\u003c/sub\u003e in five of the largest Colombian cities showed moderate to high explained variance, respectively. Generally, the models showed higher explained variance of PM\u003csub\u003e2.5\u003c/sub\u003e compared with NO\u003csub\u003e2\u003c/sub\u003e. Among the cities, the lowest explained variance was obtained for Bogot\u0026aacute;, while the highest was recorded for Medell\u0026iacute;n and Bucaramanga.\u003c/p\u003e \u003cp\u003eThe LUR models for PM\u003csub\u003e2.5\u003c/sub\u003e showed relatively small errors of the predicted concentrations (RMSE\u0026thinsp;\u0026lt;\u0026thinsp;1.7 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e) in the cities, except for Barranquilla. Moreover, the performance of the LUR models developed for PM\u003csub\u003e2.5\u003c/sub\u003e was higher than that reported in previous studies in Colombia. Previous LUR models were available only for PM\u003csub\u003e10\u003c/sub\u003e and PM\u003csub\u003e2.5\u003c/sub\u003e in the city of Medell\u0026iacute;n with an explained variability of 79% for PM\u003csub\u003e10\u003c/sub\u003e (Londo\u0026ntilde;o \u0026amp; Ca\u0026ntilde;on, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2015\u003c/span\u003e) and monthly variations between 26% and 79% for PM\u003csub\u003e2.5\u003c/sub\u003e (Grisales, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2020\u003c/span\u003e), using data from 2007 and 2018, respectively. Our selected LUR model for PM\u003csub\u003e2.5\u003c/sub\u003e in Medell\u0026iacute;n explained 82% of the variability, the highest of the five cities, using a combination of meteorological, land use, population density and traffic volume variables. The high performance of the LUR models for PM\u003csub\u003e2.5\u003c/sub\u003e in Medell\u0026iacute;n compared to other cities might be explained by the wide range of estimated concentrations in the city and the influence of the topography and meteorology in the Valley of Aburr\u0026aacute; where Medell\u0026iacute;n is located, as well as the important contribution of vehicular emissions to local concentrations as have been described in studies of PM\u003csub\u003e2.5\u003c/sub\u003e characterization in the city (Area Metropolitana del Valle de Aburr\u0026aacute; \u0026amp; Politecnico Colombiano Jaime Isaza Cadavid, 2021). In contrast, the low performance of the LUR models for PM\u003csub\u003e2.5\u003c/sub\u003e in Bogot\u0026aacute; compared to other cities might be explained partially by the lower contribution of vehicular emissions and the increased contribution of enriched fugitive dust (resuspension of crustal material and soil dust) and secondary PM (Ram\u0026iacute;rez et al., \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). A similar profile has also been documented for Barranquilla with an important contribution of ocean aerosols (Nu\u0026ntilde;ez Blanco, \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2019\u003c/span\u003e), secondary organic aerosols and the effect of exposed land resuspension and road dust (G\u0026oacute;mez-Plata et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), which was represented in the developed LUR model for this city. Additional unexplained variability in PM\u003csub\u003e2.5\u003c/sub\u003e concentrations in the cities might be related to regional wildfires contributions which have been substantial in northern South America and particularly in Bogot\u0026aacute; (Ballesteros-Gonz\u0026aacute;lez et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2020\u003c/span\u003e)(Casallas et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe variation in explained variability reported for the Colombian cities is comparable to that of PM\u003csub\u003e2.5\u003c/sub\u003e in other Latin American and European countries. In Ecuador, Alvarez et al. (Alvarez-Mendoza et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) developed LUR models for PM\u003csub\u003e10\u003c/sub\u003e using remote sensing data, and the models showed an explained variability of 68% at its highest. Sangrador et al. (Sangrador, J.T., Nu\u0026ntilde;ez, M.E., Villarreal, A.B., Cadena, L.H., Jerrett, M., Romieu, 2008) developed LUR models for PM\u003csub\u003e2.5\u003c/sub\u003e during the rainy season in 2003 for Mexico City, which showed an explained variability of 60%. Later, Son et al. (Son et al., \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e2018\u003c/span\u003e) developed LUR models for the same city for different temporal scales, and the best explained variability for monthly PM\u003csub\u003e2.5\u003c/sub\u003e models was 76%. In Europe, the ESCAPE project developed LUR models for PM\u003csub\u003e2.5\u003c/sub\u003e in 20 study areas, where the explained variability varied from 35% in Manchester, UK, to 89% in Paris, France (Eeftens et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2012\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAs expected, the best predictor variables in our LUR models for NO\u003csub\u003e2\u003c/sub\u003e were road and traffic variables. However, the performance of the LUR models developed for NO\u003csub\u003e2\u003c/sub\u003e, however, was lower than that for PM\u003csub\u003e2.5\u003c/sub\u003e and the reported from previous studies in other countries. In Sao Paulo, an annual LUR developed for NO\u003csub\u003e2\u003c/sub\u003e explained 66% of the variability in urban concentrations, with variations for summer (75%) and winter (52%) seasons (Luminati et al., \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). For the Western European countries, Vinneay et al. (Vienneau et al., \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2013\u003c/span\u003e) developed LUR models for NO\u003csub\u003e2\u003c/sub\u003e with and without satellite-based NO\u003csub\u003e2\u003c/sub\u003e and obtained explained variability between 48% and 58% without satellite-based NO\u003csub\u003e2\u003c/sub\u003e and a modest additional improvement of 5% when adding satellite-based data. In our models for NO\u003csub\u003e2\u003c/sub\u003e, despite including different variables and metrics of traffic and roads, the models could not capture a higher variability in concentrations, which suggests secondary reactions might be an important source of NO\u003csub\u003e2\u003c/sub\u003e in the cities. Although our NO\u003csub\u003e2\u003c/sub\u003e LUR explained less variability compared to other reported models in cities, the LUR models explain more variability than simple road proximity metrics or interpolation methods based on data from monitoring stations and similar variability than dispersion models, which have been demonstrated in previous studies assessing exposure assessment for epidemiological studies (Allen et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; de Hoogh et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; M Jerrett et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2007\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe LUR models have been used in exposure assessment and health research related to long-term exposure to air pollutants. By incorporating data on local sources of pollution, such as traffic or industrial activity, these models can provide more accurate and precise exposure estimates than traditional monitoring methods (Hoek et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). This is particularly important for assessing the health effects of chronic exposure to air pollution, which has been linked to a range of adverse health outcomes, including respiratory and cardiovascular disease, cancer, and neurological disorders (Chen et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Herting et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Knibbs et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Lamichhane et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Stafoggia et al., \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). LUR models can also identify areas of high pollution levels and vulnerable populations, helping to inform policy and intervention strategies to reduce exposure and improve public health (Vienneau et al., \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2013\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAlternative methods for estimating surface concentrations of air pollutants have been developed recently using satellite-based models and models using mobile air pollutant measurements. A study conducted at the municipality level in Colombia compared air quality models based on satellite measurements for PM\u003csub\u003e2.5\u003c/sub\u003e between 2014\u0026ndash;2019. It showed that the Copernicus Atmospheric Monitoring Service Reanalysis (CAMRA) and the Atmospheric Composition Analysis Group (ACAG) models had a low correlation and tended to overestimated surface concentrations when both models were compared to surface data from 28 cities in 2019. However, ACAG outperformed CAMSRA in terms of mean bias of the model and the spatial representation of the highest concentrations (Rodriguez-Villamizar et al., \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Using a mobile monitoring campaign in the city of Bucaramanga in 2019, estimations of within-city spatial variations in ultrafine particle and black carbon concentrations were predicted using a combination of LUR and convolutional neural networks trained using satellite and street-level images, showing the improvement of prediction when using a hybrid approach (Lloyd et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Following this hybrid approach, our locally developed LUR models can be further used to develop hybrid models with satellite or mobile data and produce better spatially calibrated models for estimating long-term exposure for PM\u003csub\u003e2.5\u003c/sub\u003e and NO\u003csub\u003e2\u003c/sub\u003e in the main cities in Colombia and explore their potential transferability across cities.\u003c/p\u003e \u003cp\u003eThere are some strengths in our study that are worth mentioning. First, there was a good agreement between PM\u003csub\u003e2.5\u003c/sub\u003e measurements made with UPAS compared to the concentrations reported by the local monitoring stations in the cities. For NO\u003csub\u003e2\u003c/sub\u003e, there were few monitoring sites to conduct a valid comparison in all cities, but data from local government stations in Bogot\u0026aacute; had a good agreement with concentrations reported from measurements with the Palmes tubes. Second, we followed the same standardized procedure for conducting measuring pollutants during the two campaigns in each city and the simultaneous measurement within cities avoid the potential error related to using measures in different time scales. Third, we included basic predictor variables for developing LUR models in the cities (land use, roads, traffic, population, and meteorology) available in the cities in Colombia and might be used further to developed multi-city models as those developed for Europe (Wang et al., \u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e2014\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eOne limitation of the LUR models developed for the cities is the limited number of sampling sites which was 20 for PM\u003csub\u003e2.5\u003c/sub\u003e and 40 for NO\u003csub\u003e2\u003c/sub\u003e, except for Bogot\u0026aacute; which doubled the number. These numbers are below the lower range of recommended monitoring sites (between 80\u0026ndash;100) for modeling intraurban variations in complex urban settings using LUR (Basaga\u0026ntilde;a et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). As a result, the models developed using many predictors might have resulted in more unstable performance as was observed in the cross-validation. A second limitation of this study is the absence of valid traffic data for the cities during the campaign measurement, which has shown to improve the LUR model performance, particularly for NO\u003csub\u003e2\u003c/sub\u003e (Beelen et al., \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). To overcome this limitation, we measured traffic speed derived from satellite instruments and used previously available traffic count data for the largest cities to calculate density functions which were then transferred to the other cities to estimated traffic density. Despite the density functions in the cities seemed to reflect the traffic patterns in the cities and were included as significant predictive variables, their inclusion did not help to explain a higher variability in the models for NO\u003csub\u003e2\u003c/sub\u003e. Third, we did not include meteorological variables in the development of LUR models for the cities of Bucaramanga and Barranquilla due to limited number of meteorological stations and data to produce a valid estimated surface. Although the models\u0026acute; performance for PM\u003csub\u003e2.5\u003c/sub\u003e were good particularly for Bucaramanga, including meteorological variables might have increased the models\u0026acute; performance as they have been reported as important predictors for intraurban variations in other countries (Cheewinsiriwat et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Olvera Alvarez et al., \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Another limitation of our study is that we did not include local emission sources and regional sources (such as forest fires) in the prediction models. These variables have shown to influence the concentration of particles in the cities (Casallas et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Moreover, street NO\u003csub\u003e2\u003c/sub\u003e levels may vary in building density or location, influencing their dispersion. Also, some atmospheric chemical reactions may reduce or transform NO\u003csub\u003e2\u003c/sub\u003e concentrations. In urban areas, NO\u003csub\u003e2\u003c/sub\u003e emitted mostly from traffic within a radius of 100-300m showed a correlation, although the high reactivity of NO\u003csub\u003e2\u003c/sub\u003e and rapid photodissociation may transform this pollutant in a reduced period (Agudelo-casta\u0026ntilde;eda et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIn this study we developed LUR models to predict PM\u003csub\u003e2.5\u003c/sub\u003e and NO\u003csub\u003e2\u003c/sub\u003e exposure in five main cities in Colombia. The LUR models showed a large intraurban variability of pollutant concentrations in all cities. The annual models for PM\u003csub\u003e2.5\u003c/sub\u003e outperformed the models for NO\u003csub\u003e2\u003c/sub\u003e and provided robust models that can be used in epidemiological studies, particularly cohort studies, assessing the effects of long-term air pollution on human health. The newly developed LUR models might be further used to create hybrid models in combination with other data sources to improve personal exposure assessment.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors thank Oscar Jim\u0026eacute;nez, Hermes Betancur, \u0026nbsp;Jefferson Fern\u0026aacute;nez, Angie Rojas, Paola Barbosa, Kelly Burbano, Ronald Correa, Daniela Ortiz, Martha Mendoza, Wilmer Urango, Luz Obando, and Orlando Guadu\u0026ntilde;a for their contributions during the monitoring field work in the cities.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u003c/strong\u003e This research was funded by the Ministry of Science and Technology MINCIENCIAS in Colombia, grant number 905-2019.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interests:\u003c/strong\u003e The authors have no relevant financial or nonfinancial interests to disclose.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConceptualization and methodology: LR-V, SM, JC, DA-C, VH, DM, JPJ, LB-C, OR-S, JOV, SW, JB, NR; Field work and data collection: LL,OMR,MV,WS,AZO,MC,HS; Formal analysis and investigation: LR-V, YR, and SCG; Writing-original draft preparation: LR-V, YR, and SCG; Writing-review and editing: DA-C, VH, DM, JPJ, LB-C, OR-S, JOV, SW, JB, NR, LL,OMR,MV,WS,AZO,MC,HS;\u003c/p\u003e\n\u003cp\u003eFunding acquisition: LR-V, SM, JC, DA-C, VH, DM, JPJ, LB-C, OR-S, JOV, NR.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate:\u003c/strong\u003e Not applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication:\u003c/strong\u003e Not applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability:\u003c/strong\u003e The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAgudelo-Casta\u0026ntilde;eda, D., Arellana, J., Morgado-Gamero, W. B., De Paoli, F., \u0026amp; Carla Portz, L. (2023). Linking of built environment inequalities with air quality: A case study. Transportation Research Part D: Transport and Environment, 117(March). https://doi.org/10.1016/j.trd.2023.103668\u003c/li\u003e\n\u003cli\u003eAgudelo-casta\u0026ntilde;eda, D., Paoli, F. De, Morgado-gamero, W. B., Mendoza, M., Parody, A., Maturana, A. Y., \u0026amp; Teixeira, E. C. (2020). Assessment of the NO2 distribution and relationship with traffic load in the Caribbean coastal city. Science of the Total Environment, 720, 137675. https://doi.org/10.1016/j.scitotenv.2020.137675\u003c/li\u003e\n\u003cli\u003eAllen, R. W., Amram, O., Wheeler, A. J., \u0026amp; Brauer, M. (2011). The transferability of NO and NO2 land use regression models between cities and pollutants. Atmospheric Environment, 45(2), 369\u0026ndash;378. https://doi.org/10.1016/j.atmosenv.2010.10.002\u003c/li\u003e\n\u003cli\u003eAlvarez-Mendoza, C. I., Teodoro, A. C., Torres, N., \u0026amp; Vivanco, V. (2019). Assessment of remote sensing data to model PM10 estimation in cities with a low number of air quality stations: A case of study in Quito, Ecuador. Environments - MDPI, 6(7). https://doi.org/10.3390/environments6070085\u003c/li\u003e\n\u003cli\u003eArea Metropolitana del Valle de Aburr\u0026aacute;, A., \u0026amp; Politecnico Colombiano Jaime Isaza Cadavid, P. (2021). Aporte de fuentes y caracterizaci\u0026oacute;n del PM2.5 en el Valle de Aburr\u0026aacute;, Colombia, 2019-2021. Informe final proyecto ARCAL RLA7023-Convenio interadministrativo 671 de 2021.\u003c/li\u003e\n\u003cli\u003eArku, R. E., Brauer, M., Ahmed, S. H., AlHabib, K. F., Avezum, \u0026Aacute;., Bo, J., Choudhury, T., Dans, A. M., Gupta, R., Iqbal, R., Ismail, N., Kelishadi, R., Khatib, R., Koon, T., Kumar, R., Lanas, F., Lear, S. A., Wei, L., Lopez-Jaramillo, P., \u0026hellip; Hystad, P. (2020). Long-term exposure to outdoor and household air pollution and blood pressure in the Prospective Urban and Rural Epidemiological (PURE) study. Environmental Pollution (Barking, Essex : 1987), 262, 114197. https://doi.org/10.1016/j.envpol.2020.114197\u003c/li\u003e\n\u003cli\u003eBallesteros-Gonz\u0026aacute;lez, K., Sullivan, A. P., \u0026amp; Morales-Betancourt, R. (2020). Estimating the air quality and health impacts of biomass burning in northern South America using a chemical transport model. The Science of the Total Environment, 739, 139755. https://doi.org/10.1016/j.scitotenv.2020.139755\u003c/li\u003e\n\u003cli\u003eBasaga\u0026ntilde;a, X., Rivera, M., Aguilera, I., Agis, D., Bouso, L., Elosua, R., Foraster, M., de Nazelle, A., Nieuwenhuijsen, M., Vila, J., \u0026amp; K\u0026uuml;nzli, N. (2012). Effect of the number of measurement sites on land use regression models in estimating local air pollution. Atmospheric Environment, 54, 634\u0026ndash;642. https://doi.org/10.1016/j.atmosenv.2012.01.064\u003c/li\u003e\n\u003cli\u003eBeelen, R., Hoek, G., Vienneau, D., Eeftens, M., Dimakopoulou, K., Pedeli, X., Tsai, M. Y., K\u0026uuml;nzli, N., Schikowski, T., Marcon, A., Eriksen, K. T., Raaschou-Nielsen, O., Stephanou, E., Patelarou, E., Lanki, T., Yli-Tuomi, T., Declercq, C., Falq, G., Stempfelet, M., \u0026hellip; de Hoogh, K. (2013). Development of NO2 and NOx land use regression models for estimating air pollution exposure in 36 study areas in Europe \u0026ndash; The ESCAPE project. Atmospheric Environment, 72, 10\u0026ndash;23. https://doi.org/10.1016/J.ATMOSENV.2013.02.037\u003c/li\u003e\n\u003cli\u003eBlanco-Becerra, L. C., Miranda-Soberanis, V., Hern\u0026aacute;ndez-Cadena, L., Barraza-Villarreal, A., Junger, W., Hurtado-D\u0026iacute;az, M., \u0026amp; Romieu, I. (2014). Effect of particulate matter less than 10\u0026micro;m (PM10) on mortality in Bogota, Colombia: a time-series analysis, 1998-2006. Salud Publica de Mexico, 56(4), 363\u0026ndash;370. https://doi.org/10.21149/spm.v56i4.7356\u003c/li\u003e\n\u003cli\u003eBland, J. M., \u0026amp; Altman, D. G. (1986). Statistical methods for assessing agreement between two methods of clinical measurement. Lancet (London, England), 1(8476), 307\u0026ndash;310.\u003c/li\u003e\n\u003cli\u003eCasallas, A., Castillo-Camacho, M. P., Guevara-Luna, M. A., Gonz\u0026aacute;lez, Y., Sanchez, E., \u0026amp; Belalcazar, L. C. (2022). Spatio-temporal analysis of PM(2.5) and policies in Northwestern South America. The Science of the Total Environment, 852, 158504. https://doi.org/10.1016/j.scitotenv.2022.158504\u003c/li\u003e\n\u003cli\u003eCheewinsiriwat, P., Duangyiwa, C., Sukitpaneenit, M., \u0026amp; Stettler, M. E. J. (2022). Influence of Land Use and Meteorological Factors on PM2.5 and PM10 Concentrations in Bangkok, Thailand. Sustainability (Switzerland), 14(9). https://doi.org/10.3390/su14095367\u003c/li\u003e\n\u003cli\u003eChen, H., Goldberg, M. S., Burnett, R. T., Jerrett, M., Wheeler, A. J., \u0026amp; Villeneuve, P. J. (2013). Long-term exposure to traffic-related air pollution and cardiovascular mortality. Epidemiology, 24(1), 35\u0026ndash;43. https://doi.org/10.1097/EDE.0b013e318276c005\u003c/li\u003e\n\u003cli\u003eCouncil, N. R. (2010). Highway capacity manual 2010.\u003c/li\u003e\n\u003cli\u003eCrouse, D. L., Peters, P. A., Villeneuve, P. J., Proux, M.-O., Shin, H. H., Goldberg, M. S., Johnson, M., Wheeler, A. J., Allen, R. W., Atari, D. O., Jerrett, M., Brauer, M., Brook, J. R., Cakmak, S., \u0026amp; Burnett, R. T. (2015). Within- and between-city contrasts in nitrogen dioxide and mortality in 10 Canadian cities; a subset of the Canadian Census Health and Environment Cohort (CanCHEC). Journal of Exposure Science \u0026amp; Environmental Epidemiology, 25(5), 482\u0026ndash;489. https://doi.org/10.1038/jes.2014.89\u003c/li\u003e\n\u003cli\u003eCunha-Zeri, G., \u0026amp; Ometto, J. (2021). Nitrogen emissions in Latin America: A conceptual framework of drivers, impacts, and policy responses. Environmental Development, 38(October 2019), 100605. https://doi.org/10.1016/j.envdev.2020.100605\u003c/li\u003e\n\u003cli\u003ede Hoogh, K., Gulliver, J., van Donkelaar, A., Martin, R. V, Marshall, J. D., Bechle, M. J., Cesaroni, G., Pradas, M. C., Dedele, A., Eeftens, M., Forsberg, B., Galassi, C., Heinrich, J., Hoffmann, B., Jacquemin, B., Katsouyanni, K., Korek, M., K\u0026uuml;nzli, N., Lindley, S. J., \u0026hellip; Hoek, G. (2016). Development of West-European PM2.5 and NO2 land use regression models incorporating satellite-derived and chemical transport modelling data. Environmental Research, 151, 1\u0026ndash;10. https://doi.org/10.1016/j.envres.2016.07.005\u003c/li\u003e\n\u003cli\u003ede Hoogh, K., Korek, M., Vienneau, D., Keuken, M., Kukkonen, J., Nieuwenhuijsen, M. J., Badaloni, C., Beelen, R., Bolignano, A., Cesaroni, G., Pradas, M. C., Cyrys, J., Douros, J., Eeftens, M., Forastiere, F., Forsberg, B., Fuks, K., Gehring, U., Gryparis, A., \u0026hellip; Bellander, T. (2014). Comparing land use regression and dispersion modelling to assess residential exposure to ambient air pollution for epidemiological studies. Environment International, 73, 382\u0026ndash;392. https://doi.org/10.1016/j.envint.2014.08.011\u003c/li\u003e\n\u003cli\u003eDepartamento Nacional de Estad\u0026iacute;stica (DANE). (2020a). Geoportal Censo Nacional de Poblaci\u0026oacute;n y Vivienda 2018. https://geoportal.dane.gov.co/\u003c/li\u003e\n\u003cli\u003eDepartamento Nacional de Estad\u0026iacute;stica (DANE). (2020b). Proyecciones de Poblaci\u0026oacute;n. https://www.dane.gov.co/index.php/estadisticas-por-tema/demografia-y-poblacion/proyecciones-de-poblacion\u003c/li\u003e\n\u003cli\u003eDijkema, M. B., Gehring, U., van Strien, R. T., van der Zee, S. C., Fischer, P., Hoek, G., \u0026amp; Brunekreef, B. (2011). A comparison of different approaches to estimate small-scale spatial variation in outdoor NO2 concentrations. Environmental Health Perspectives, 119(2), 670\u0026ndash;675. https://doi.org/10.1289/ehp.0901818\u003c/li\u003e\n\u003cli\u003eEeftens, M., Beelen, R., De Hoogh, K., Bellander, T., Cesaroni, G., Cirach, M., Declercq, C., Dedele, A., Dons, E., De Nazelle, A., Dimakopoulou, K., Eriksen, K., Falq, G., Fischer, P., Galassi, C., Gražulevičiene, R., Heinrich, J., Hoffmann, B., Jerrett, M., \u0026hellip; Hoek, G. (2012). Development of land use regression models for PM2.5, PM 2.5 absorbance, PM10 and PMcoarse in 20 European study areas; Results of the ESCAPE project. Environmental Science and Technology, 46, 11195\u0026ndash;11205. https://doi.org/10.1021/es301948k\u003c/li\u003e\n\u003cli\u003eFann, N., Bell, M. L., Walker, K., \u0026amp; Hubbell, B. (2011). Improving the linkages between air pollution epidemiology and quantitative risk assessment. Environmental Health Perspectives, 119(12), 1671\u0026ndash;1675. https://doi.org/10.1289/ehp.1103780\u003c/li\u003e\n\u003cli\u003eGaddam, H. K., \u0026amp; Rao, K. R. (2019). Speed\u0026ndash;density functional relationship for heterogeneous traffic data: a statistical and theoretical investigation. Journal of Modern Transportation, 27(1), 61\u0026ndash;74. https://doi.org/10.1007/s40534-018-0177-7\u003c/li\u003e\n\u003cli\u003eGBD 2019 Risk Factors Collaborators. (2020). Global burden of 87 risk factors in 204 countries and territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet (London, England), 396(10258), 1223\u0026ndash;1249. https://doi.org/10.1016/S0140-6736(20)30752-2\u003c/li\u003e\n\u003cli\u003eG\u0026oacute;mez-Plata, L., Agudelo-Casta\u0026ntilde;eda, D., Castillo, M., \u0026amp; Teixeira, E. C. (2022). PM10 Source Identification: A Case of a Coastal City in Colombia. Aerosol and Air Quality Research, 22(10), 1\u0026ndash;17. https://doi.org/10.4209/aaqr.210293\u003c/li\u003e\n\u003cli\u003eGrisales, S. (2020). Modelos de regresi\u0026oacute;n de usos del suelo para la caracterizaci\u0026oacute;n espacial de la contaminaci\u0026oacute;n del aire por PM2.5 en la ciudad de Medell\u0026iacute;n-Colombia, 2018 [Universidad de Antioquia, Medell\u0026iacute;n, Colombia]. https://hdl.handle.net/10495/16376\u003c/li\u003e\n\u003cli\u003eGurung, A., Levy, J. I., \u0026amp; Bell, M. L. (2017). Modeling the intraurban variation in nitrogen dioxide in urban areas in Kathmandu Valley, Nepal. Environmental Research, 155, 42\u0026ndash;48. https://doi.org/10.1016/j.envres.2017.01.038\u003c/li\u003e\n\u003cli\u003eHabbermann, M., \u0026amp; Gouveia, N. (2007). Application of land use regression to predict the concentration of inhalable particulate matter in de S\u0026atilde;o Paulo city, Brasil. Eng Sanit Ambient, 17(2), 155\u0026ndash;162.\u003c/li\u003e\n\u003cli\u003eHerting, M. M., Younan, D., Campbell, C. E., \u0026amp; Chen, J.-C. (2019). Outdoor Air Pollution and Brain Structure and Function From Across Childhood to Young Adulthood: A Methodological Review of Brain MRI Studies. In Frontiers in public health (Vol. 7, p. 332). https://doi.org/10.3389/fpubh.2019.00332\u003c/li\u003e\n\u003cli\u003eHoek, G. (2017). Methods for Assessing Long-Term Exposures to Outdoor Air Pollutants. Current Environmental Health Reports, 4(4), 450\u0026ndash;462. https://doi.org/10.1007/s40572-017-0169-5\u003c/li\u003e\n\u003cli\u003eHoek, G., Beelen, R., de Hoogh, K., Vienneau, D., Gulliver, J., Fischer, P., \u0026amp; Briggs, D. (2008). A review of land-use regression models to assess spatial variation of outdoor air pollution. Atmospheric Environment, 42(33), 7561\u0026ndash;7578. https://doi.org/10.1016/j.atmosenv.2008.05.057\u003c/li\u003e\n\u003cli\u003eInstituto de Hidrologia Meteorolog\u0026iacute;a y Estudios Ambientales-IDEAM. (2022). Informe del estado de la calidad del aire en Colombia 2021.\u003c/li\u003e\n\u003cli\u003eJerrett, M, Arain, M. A., Kanaroglou, P., Beckerman, B., Crouse, D., Gilbert, N. L., Brook, J. R., Finkelstein, N., \u0026amp; Finkelstein, M. M. (2007). Modeling the intraurban variability of ambient traffic pollution in Toronto, Canada. Journal of Toxicology and Environmental Health. Part A, 70(3\u0026ndash;4), 200\u0026ndash;212. https://doi.org/10.1080/15287390600883018\u003c/li\u003e\n\u003cli\u003eJerrett, Michael, Arain, A., Kanaroglou, P., \u0026amp; Beckerman, B. (2005). A review and evaluation of intraurban air pollution exposure models. 185\u0026ndash;204. https://doi.org/10.1038/sj.jea.7500388\u003c/li\u003e\n\u003cli\u003eKashima, S., Yorifuji, T., Sawada, N., Nakaya, T., \u0026amp; Eboshida, A. (2018). Comparison of land use regression models for NO2 based on routine and campaign monitoring data from an urban area of Japan. The Science of the Total Environment, 631\u0026ndash;632, 1029\u0026ndash;1037. https://doi.org/10.1016/j.scitotenv.2018.02.334\u003c/li\u003e\n\u003cli\u003eKnibbs, L. D., Cortes de Waterman, A. M., Toelle, B. G., Guo, Y., Denison, L., Jalaludin, B., Marks, G. B., \u0026amp; Williams, G. M. (2018). The Australian Child Health and Air Pollution Study (ACHAPS): A national population-based cross-sectional study of long-term exposure to outdoor air pollution, asthma, and lung function. Environment International, 120, 394\u0026ndash;403. https://doi.org/10.1016/j.envint.2018.08.025\u003c/li\u003e\n\u003cli\u003eLamichhane, D. K., Kim, H. C., Choi, C. M., Shin, M. H., Shim, Y. M., Leem, J. H., Ryu, J. S., Nam, H. S., \u0026amp; Park, S. M. (2017). Lung Cancer Risk and Residential Exposure to Air Pollution: A Korean Population-Based Case-Control Study. Yonsei Medical Journal, 58(6), 1111\u0026ndash;1118. https://doi.org/10.3349/ymj.2017.58.6.1111\u003c/li\u003e\n\u003cli\u003eLee, M., Brauer, M., Wong, P., Tang, R., Tsui, T. H., Choi, C., Cheng, W., Lai, P.-C., Tian, L., Thach, T.-Q., Allen, R., \u0026amp; Barratt, B. (2017). Land use regression modelling of air pollution in high density high rise cities: A case study in Hong Kong. The Science of the Total Environment, 592, 306\u0026ndash;315. https://doi.org/10.1016/j.scitotenv.2017.03.094\u003c/li\u003e\n\u003cli\u003eLeith, D., L\u0026rsquo;Orange, C., Mehaffy, J., \u0026amp; Volckens, J. (2020). Design and performance of UPAS inlets for respirable and thoracic mass sampling. Journal of Occupational and Environmental Hygiene, 17(6), 274\u0026ndash;282. https://doi.org/10.1080/15459624.2020.1741595\u003c/li\u003e\n\u003cli\u003eLloyd, M., Carter, E., Diaz, F. G., Magara-Gomez, K. T., Hong, K. Y., Baumgartner, J., Herrera G, V. M., \u0026amp; Weichenthal, S. (2021). Predicting Within-City Spatial Variations in Outdoor Ultrafine Particle and Black Carbon Concentrations in Bucaramanga, Colombia: A Hybrid Approach Using Open-Source Geographic Data and Digital Images. Environmental Science \u0026amp; Technology, 55(18), 12483\u0026ndash;12492. https://doi.org/10.1021/acs.est.1c01412\u003c/li\u003e\n\u003cli\u003eLondo\u0026ntilde;o, L. (2018). Metodolog\u0026iacute;a para caracterizar espacio-temporalmente la concentraci\u0026oacute;n de material particulado en Valles intramontanos con informaci\u0026oacute;n escasa [Universidad de Antioquia, Medell\u0026iacute;n, Colombia]. https://bibliotecadigital.udea.edu.co/handle/10495/9413\u003c/li\u003e\n\u003cli\u003eLondo\u0026ntilde;o, L., \u0026amp; Ca\u0026ntilde;on, J. (2015). Metodologia para la aplicaci\u0026oacute;n de modelos de regresi\u0026oacute;n de usos de suelo en la estimaci\u0026oacute;n local de la concentraci\u0026oacute;n mensual de PM10 en Medell\u0026iacute;n, Colombia. Revista Polit\u0026eacute;cnica, 11(21), 29\u0026ndash;40.\u003c/li\u003e\n\u003cli\u003eLuminati, O., Ledebur de Antas de Campos, B., Fl\u0026uuml;ckiger, B., Brentani, A., R\u0026ouml;\u0026ouml;sli, M., Fink, G., \u0026amp; de Hoogh, K. (2021). Land use regression modelling of NO(2) in S\u0026atilde;o Paulo, Brazil. Environmental Pollution (Barking, Essex : 1987), 289, 117832. https://doi.org/10.1016/j.envpol.2021.117832\u003c/li\u003e\n\u003cli\u003eMaantay, J., \u0026amp; McLafferty, S. (2011). Geospatial Analysis of Environmental Health. Springer. https://doi.org/10.1007/978-94-007-0329-2\u003c/li\u003e\n\u003cli\u003eNu\u0026ntilde;ez Blanco, Y. (2019). Estimaci\u0026oacute;n de fuentes de material particulado atmosf\u0026eacute;rico (PM10 y PM2.5) en la ciudad de Barranquilla, Colombia. Universidad de la Costa, Barranquilla, Colombia.\u003c/li\u003e\n\u003cli\u003eOlvera Alvarez, H. A., Myers, O. B., Weigel, M., \u0026amp; Armijos, R. X. (2018). The value of using seasonality and meteorological variables to model intra-urban PM2.5 variation. Atmospheric Environment, 182(March), 1\u0026ndash;8. https://doi.org/10.1016/j.atmosenv.2018.03.007\u003c/li\u003e\n\u003cli\u003ePope, C. A. 3rd. (2007). Mortality effects of longer term exposures to fine particulate air pollution: review of recent epidemiological evidence. Inhalation Toxicology, 19 Suppl 1, 33\u0026ndash;38. https://doi.org/10.1080/08958370701492961\u003c/li\u003e\n\u003cli\u003eRam\u0026iacute;rez, O., S\u0026aacute;nchez de la Campa, A. M., Amato, F., Catacol\u0026iacute;, R. A., Rojas, N. Y., \u0026amp; de la Rosa, J. (2018). Chemical composition and source apportionment of PM(10) at an urban background site in a high-altitude Latin American megacity (Bogota, Colombia). Environmental Pollution (Barking, Essex : 1987), 233, 142\u0026ndash;155. https://doi.org/10.1016/j.envpol.2017.10.045\u003c/li\u003e\n\u003cli\u003eRauthe, M., Steiner, H., Riediger, U., Mazurkiewicz, A., \u0026amp; Gratzki, A. (2013). A Central European precipitation climatology - Part I: Generation and validation of a high-resolution gridded daily data set (HYRAS). Meteorologische Zeitschrift, 22(3), 235\u0026ndash;256. https://doi.org/10.1127/0941-2948/2013/0436\u003c/li\u003e\n\u003cli\u003eRodriguez-Villamizar, L. A., Belalcazar-Ceron, L. C., Castillo, M. P., Sanchez, E. R., Herrera, V., \u0026amp; Agudelo-Casta\u0026ntilde;eda, D. M. (2022). Avoidable mortality due to long-term exposure to PM(2.5) in Colombia 2014-2019. Environmental Health : A Global Access Science Source, 21(1), 137. https://doi.org/10.1186/s12940-022-00947-8\u003c/li\u003e\n\u003cli\u003eRodriguez-Villamizar, L. A., Rojas-Roa, N. Y., Blanco-Becerra, L. C., Herrera-Galindo, V. M., \u0026amp; Fern\u0026aacute;ndez-Ni\u0026ntilde;o, J. A. (2018). Short-Term Effects of Air Pollution on Respiratory and Circulatory Morbidity in Colombia 2011\u003csup\u003e\u0026minus;\u003c/sup\u003e2014: A Multi-City, Time-Series Analysis. International Journal of Environmental Research and Public Health, 15(8). https://doi.org/10.3390/ijerph15081610\u003c/li\u003e\n\u003cli\u003eSangrador, J.T., Nu\u0026ntilde;ez, M.E., Villarreal, A.B., Cadena, L.H., Jerrett, M., Romieu, I. (2008). A land use regression model for predicting PM2.5 in Mexico City. Epidemiology, 19(S259), 1.\u003c/li\u003e\n\u003cli\u003eSon, Y., Osornio-Vargas, A. R., O\u0026rsquo;Neill, M. S., Hystad, P., Texcalac-Sangrador, J. L., Ohman-Strickland, P., Meng, Q., \u0026amp; Schwander, S. (2018). Land use regression models to assess air pollution exposure in Mexico City using finer spatial and temporal input parameters. The Science of the Total Environment, 639, 40\u0026ndash;48. https://doi.org/10.1016/j.scitotenv.2018.05.144\u003c/li\u003e\n\u003cli\u003eStafoggia, M., Oftedal, B., Chen, J., Rodopoulou, S., Renzi, M., Atkinson, R. W., Bauwelinck, M., Klompmaker, J. O., Mehta, A., Vienneau, D., Andersen, Z. J., Bellander, T., Brandt, J., Cesaroni, G., de Hoogh, K., Fecht, D., Gulliver, J., Hertel, O., Hoffmann, B., \u0026hellip; Janssen, N. A. H. (2022). Long-term exposure to low ambient air pollution concentrations and mortality among 28 million people: results from seven large European cohorts within the ELAPSE project. The Lancet. Planetary Health, 6(1), e9\u0026ndash;e18. https://doi.org/10.1016/S2542-5196(21)00277-1\u003c/li\u003e\n\u003cli\u003evan Donkelaar, A., Hammer, M. S., Bindle, L., Brauer, M., Brook, J. R., Garay, M. J., Hsu, N. C., Kalashnikova, O. V, Kahn, R. A., Lee, C., Levy, R. C., Lyapustin, A., Sayer, A. M., \u0026amp; Martin, R. V. (2021). Monthly Global Estimates of Fine Particulate Matter and Their Uncertainty. Environmental Science \u0026amp; Technology, 55(22), 15287\u0026ndash;15300. https://doi.org/10.1021/acs.est.1c05309\u003c/li\u003e\n\u003cli\u003eVan Nunen, E., Vermeulen, R., Tsai, M. Y., Probst-Hensch, N., Ineichen, A., Davey, M., Imboden, M., Ducret-Stich, R., Naccarati, A., Raffaele, D., Ranzi, A., Ivaldi, C., Galassi, C., Nieuwenhuijsen, M., Curto, A., Donaire-Gonzalez, D., Cirach, M., Chatzi, L., Kampouri, M., \u0026hellip; Hoek, G. (2017). Land Use Regression Models for Ultrafine Particles in Six European Areas. Environmental Science and Technology, 51, 3336\u0026ndash;3345. https://doi.org/10.1021/acs.est.6b05920\u003c/li\u003e\n\u003cli\u003eVienneau, D., de Hoogh, K., Bechle, M. J., Beelen, R., van Donkelaar, A., Martin, R. V, Millet, D. B., Hoek, G., \u0026amp; Marshall, J. D. (2013). Western European land use regression incorporating satellite- and ground-based measurements of NO2 and PM10. Environmental Science \u0026amp; Technology, 47(23), 13555\u0026ndash;13564. https://doi.org/10.1021/es403089q\u003c/li\u003e\n\u003cli\u003eWang, M., Beelen, R., Bellander, T., Birk, M., Cesaroni, G., Cirach, M., Cyrys, J., de Hoogh, K., Declercq, C., Dimakopoulou, K., Eeftens, M., Eriksen, K. T., Forastiere, F., Galassi, C., Grivas, G., Heinrich, J., Hoffmann, B., Ineichen, A., Korek, M., \u0026hellip; Brunekreef, B. (2014). Performance of multi-city land use regression models for nitrogen dioxide and fine particles. Environmental Health Perspectives, 122(8), 843\u0026ndash;849. https://doi.org/10.1289/ehp.1307271\u003c/li\u003e\n\u003cli\u003eWang, M., Brunekreef, B., Gehring, U., Szpiro, A., Hoek, G., \u0026amp; Beelen, R. (2016). A New Technique for Evaluating Land-use Regression Models and Their Impact on Health Effect Estimates. Epidemiology (Cambridge, Mass.), 27(1), 51\u0026ndash;56. https://doi.org/10.1097/EDE.0000000000000404\u003c/li\u003e\n\u003cli\u003eWorld Health Organization. (2021). WHO global air quality guidelines: particulate matter (PM2.5 and PM10), ozone, nitrogen dioxide, sulfur dioxide and carbon monoxide. World Health Organization. https://apps.who.int/iris/handle/10665/345329\u003c/li\u003e\n\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":true,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"environmental-science-and-pollution-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"espr","sideBox":"Learn more about [Environmental Science and Pollution Research](https://www.springer.com/journal/11356)","snPcode":"11356","submissionUrl":"https://submission.nature.com/new-submission/11356/3","title":"Environmental Science and Pollution Research","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Air pollution, Fine Particulate Matter, Nitrogen Dioxide, Land Use Regression Models, Colombia","lastPublishedDoi":"10.21203/rs.3.rs-2988847/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-2988847/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eRapidly urbanizing cities in Latin America experience high levels of air pollution which are known risk factors for population health. However, the estimates of long-term exposure to air pollution are scarce in the region. We developed intraurban land use regression (LUR) models to map long-term exposure to fine particulate matter (PM\u003csub\u003e2.5\u003c/sub\u003e) and nitrogen dioxide (NO\u003csub\u003e2\u003c/sub\u003e) in the five largest cities in Colombia. We conducted air pollution measurement campaigns using gravimetric PM\u003csub\u003e2.5\u003c/sub\u003e and passive NO\u003csub\u003e2\u003c/sub\u003e sensors for two weeks during both the dry and rainy seasons in 2021 in the cities of Barranquilla, Bucaramanga, Bogot\u0026aacute;, Cali, and Medell\u0026iacute;n, and combined these data with geospatial and meteorological variables. Annual models were developed using multivariable spatial regression models. The city annual PM\u003csub\u003e2.5\u003c/sub\u003e mean concentrations measured ranged between 12.32 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e and 15.99 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e while NO\u003csub\u003e2\u003c/sub\u003e concentrations ranged between 24.92 \u0026#120525;/m3) and 49.15 \u0026#120525;g/m\u003csup\u003e3\u003c/sup\u003e. The PM\u003csub\u003e2.5\u003c/sub\u003e annual models explained 82% of the variance (R\u003csup\u003e2\u003c/sup\u003e) in Medell\u0026iacute;n, 77% in Bucaramanga, 73% in Barranquilla, 70% in Cali, and 44% in Bogot\u0026aacute;. The NO\u003csub\u003e2\u003c/sub\u003e models explained 65% of the variance in Bucaramanga, 57% in Medell\u0026iacute;n, 44% in Cali, 40% in Bogot\u0026aacute;, and 30% in Barranquilla. Most of the predictor variables included in the models were a combination of specific land use characteristics and roadway variables. Cross-validation suggest that PM\u003csub\u003e2.5\u003c/sub\u003e outperformed NO\u003csub\u003e2\u003c/sub\u003e models. The developed models can be used as exposure estimate in epidemiological studies, as input in hybrid models to improve personal exposure assessment, and for policy evaluation.\u003c/p\u003e","manuscriptTitle":"Intra-urban variability of long-term exposure to PM2.5 and NO2 in five cities in Colombia","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2023-06-30 17:40:50","doi":"10.21203/rs.3.rs-2988847/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Minor Revision","date":"2023-09-18T03:13:41+00:00","index":"","fulltext":""},{"type":"reviewerAgreed","content":"","date":"2023-06-29T19:41:08+00:00","index":0,"fulltext":""},{"type":"reviewersInvited","content":"","date":"2023-06-28T12:07:09+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"Environmental Science and Pollution Research","date":"2023-06-27T13:21:39+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2023-06-06T04:18:18+00:00","index":"","fulltext":""},{"type":"submitted","content":"Environmental Science and Pollution Research","date":"2023-05-31T07:20:49+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"environmental-science-and-pollution-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"espr","sideBox":"Learn more about [Environmental Science and Pollution Research](https://www.springer.com/journal/11356)","snPcode":"11356","submissionUrl":"https://submission.nature.com/new-submission/11356/3","title":"Environmental Science and Pollution Research","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"5270eb41-31b9-4d96-a485-79ce628f3690","owner":[],"postedDate":"June 30th, 2023","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2023-12-18T15:04:42+00:00","versionOfRecord":{"articleIdentity":"rs-2988847","link":"https://doi.org/10.1007/s11356-023-31306-w","journal":{"identity":"environmental-science-and-pollution-research","isVorOnly":false,"title":"Environmental Science and Pollution Research"},"publishedOn":"2023-12-12 15:00:43","publishedOnDateReadable":"December 12th, 2023"},"versionCreatedAt":"2023-06-30 17:40:50","video":"","vorDoi":"10.1007/s11356-023-31306-w","vorDoiUrl":"https://doi.org/10.1007/s11356-023-31306-w","workflowStages":[]},"version":"v1","identity":"rs-2988847","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-2988847","identity":"rs-2988847","version":["v1"]},"buildId":"cBFmMYwuxLRRLfASyISRj","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}