A Lagrangian-Based Technique to Generate High-Resolution Sea Surface Salinity Fields from Low-Resolution Satellite Observations: A Study in the Bay of Bengal

A Lagrangian-Based Technique to Generate High-Resolution Sea Surface Salinity Fields from Low-Resolution Satellite Observations: A Study in the Bay of Bengal | 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 A Lagrangian-Based Technique to Generate High-Resolution Sea Surface Salinity Fields from Low-Resolution Satellite Observations: A Study in the Bay of Bengal Jai Kumar, Neeraj Agarwal, Rashmi Sharma This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8247078/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 7 You are reading this latest preprint version Abstract In this study, we examined salinity fields as both passive and active tracers to generate high-resolution sea surface salinity fields using a Lagrangian reconstruction technique, integrating satellite-derived data and numerical advection schemes. Specifically, we employed Version 4.0 of NASA’s Soil Moisture Active Passive (SMAP) Level-3 SSS product, which features an 8-day running mean and approximately 25 km spatial resolution. Geostrophic surface currents at comparable spatial resolution, sourced from satellite altimetry data provided by the Copernicus Marine Environment Monitoring Service (CMEMS), were utilized in this work. By applying backward and forward numerical advection schemes to the SMAP SSS fields using these altimetry-derived currents, we captured smaller-scale salinity features, enhancing spatial resolution from around 25 km down to 4 km. Utilizing salinity as a passive tracer allowed us to focus exclusively on horizontal advection without accounting for sources, sinks, or mixing. A sensitivity analysis was performed, which determined that the highest feasible resolution using this approach is 4 km, with an optimal advection integration period of 14 days Preliminary validation against ship-based thermo-salinograph observations from 2015 and 2024 demonstrates that HRSSS-PA achieves RMSEs of 0.19 to 3.09 psu with correlations of 0.12 to 0.93, generally outperforming SMAP, which shows higher errors (0.47 to 4.78 psu) and weaker or inconsistent correlations (-0.11 to 0.96). This highlights the ability of HRSSS-PA to capture both magnitude and fine-scale salinity variability. To address cases where the assumptions of passive advection break down, the framework was further extended to an active advection approach, in which salinity values were dynamically adjusted during transport to account for freshwater input and precipitation, enabling improved representation under strongly forced oceanographic conditions. Sea Surface Salinity Lagrangian Advection High-Resolution Reconstruction SMAP Geostrophic Currents Bay of Bengal Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 1. Introduction Sea surface salinity (SSS) plays a critical role in regulating upper-ocean stratification, mixed-layer dynamics, and air-sea fluxes, with downstream impacts on monsoon systems, cyclone intensities, and climate variability. This is especially evident in the Bay of Bengal (BoB), where intense monsoonal rainfall and large river discharges from systems such as the Ganga-Brahmaputra-Meghna and Irrawaddy create strong near-surface salinity gradients and highly stratified upper-ocean conditions (Mahadevan et al., 2016 ; Ratheesh et al., 2020 ). The resulting barrier layers suppress convective overturning and entrainment cooling, altering the diurnal cycle of sea surface temperature (SST) and regional air-sea interactions. Despite the dynamic significance of SSS, current satellite-derived products (e.g., from SMOS, Aquarius, and SMAP) suffer from coarse effective resolutions of ~ 50–100 km due to radiometric noise, land-sea contamination, and multi-day compositing windows (Reul et al., 2020 ; Vinogradova et al., 2019 ). These limitations hinder the detection of mesoscale (10–100 km) and sub-mesoscale (< 10 km) features such as freshwater plumes, fronts, and eddies that critically shape surface salinity distributions. In contrast, satellite SST and SSH routinely achieve finer than 10 km resolution (Chin et al., 2017 ), enabling high-fidelity studies of oceanic density fronts and mixing-processes that require co-located gradients in temperature and salinity (Lévy et al., 2001 ; Rudnick & Ferrari, 1999 ). The Bay of Bengal offers a compelling testbed for improving SSS resolution. Observations from initiatives like ASIRI–OMM reveal widespread salinity-driven stratification and highlight the need for integrated measurements using ships, moorings, and autonomous platforms (Wijesekera et al., 2016 ). Shipboard process cruises during the southwest monsoon have uncovered sub-mesoscale frontal networks embedded in the stratified surface ocean (Sharma et. al., 2016; Lucas et al., 2016 ; Ramachandran et al., 2018 ), while long-term mooring records show persistent near-surface stratification maintained by runoff and mesoscale stirring (Sengupta et al., 2016 ). These dynamics modulate nutrient availability, upper-ocean mixing, and freshwater redistribution, underscoring the need for high-resolution SSS fields to resolve key biophysical and climate-relevant processes (Weller et al., 2019 ; Cherian et al., 2020 ). To address the coarse resolution of satellite-derived SSS products, several reconstruction techniques have emerged. Among them, Lagrangian advection methods leverage high-resolution surface current fields (from altimetry or models) to transport coarse salinity data along fluid trajectories, thereby reconstructing fine-scale tracer patterns consistent with ocean dynamics. This passive-tracer approach was first demonstrated in the northeastern Irminger Sea, where lateral stirring by mesoscale currents sharpens filamentary structures and fronts through backward‐forward advection of salinity particles (Desprès et al., 2011 ). The method was applied in the Southern Ocean south of Tasmania, where a ~ 2-week optimal advection period balanced fine-scale detail and unmodeled processes (Dencausse et al., 2014 ), and later extended to the Pacific, confirming its robustness and consistent performance across dynamic regimes (Rogé et al., 2015 ). Regional studies in the Indian Ocean support the Lagrangian framework, highlighting upwelling and mixing in the Somali coast (Kumar et al., 2024 ) and using FSLEs to assess mixing and biological activity across the North Indian Ocean (Kumar et al., 2023 ). These studies underscore the capacity of Lagrangian diagnostics to resolve mesoscale-to-submesoscale features and their biological implications. Idealized turbulence studies further validated that Lagrangian reconstructions faithfully recover sub-mesoscale features in passive tracers (Berti & Lapeyre, 2014 ), and global implementations have since mapped sub-mesoscale salinity variability at unprecedented resolution (Drushka et al., 2019 ). Lagrangian advection, which conserves tracers and follows surface flow kinematics, provides a consistent framework for linking coarse-resolution SSS with finer SST and SSH data, enabling integrated analyses of frontogenesis and upper ocean mixing. Alongside this, recent advances in artificial intelligence have introduced data-driven methods to enhance SSS resolution (Wang et al., 2022 , Liang et al., 2025 ). While these AI-based methods offer flexible and accurate alternatives, they may trade off physical interpretability and tracer conservation for improved statistical performance. This study presents a Lagrangian-based approach to reconstruct high-resolution sea surface salinity (SSS) fields from coarse satellite data, using the Bay of Bengal as a test case. By advecting SMAP SSS with altimetry-derived currents and treating salinity as a passive tracer, the method aims to enhance the effective resolution for capturing finer-scale variability. We evaluate its performance through comparisons with existing high-resolution products and in situ observations, and further explore an adaptive advection scheme to address regional complexities. To the best of our knowledge, this is the first application of such techniques in a region with intense freshwater forcing. The paper is organized as follows: Section 2 describes the data used and the methodology; Section 3 presents the results and discussion; and Section 4 concludes with the study’s findings. 2. Data sets 2.1 Data sets used in reconstruction of high resolution sea surface salinity SMAP sea surface salinity : Version 4.0 of the Level-3 SMAP SSS product from NASA’s SMAP mission, which is an 8-day running mean gridded dataset (Meissner et al., 2019 ), has been used in this study. This product is distributed via NASA’s Physical Oceanography Distributed Active Archive Center (PO.DAAC). It is accessible from the corresponding dataset page on the PO.DAAC portal ( https://search.earthdata.nasa.gov/ ). The data are on a 25 km equal-area grid, and the effective spatial resolution of SMAP SSS is on the order of ~ 70 km for feature detection. Geostrophic currents from Altimetry : Surface ocean currents were obtained from multi-mission satellite altimeter data, distributed as a gridded absolute dynamic topography by the Copernicus Marine Environment Monitoring Service (CMEMS). We use the daily, global 1/4° (approximately 25 km) resolution geostrophic current product (a Level-4 merged altimetry product) from CMEMS ( https://data.marine.copernicus.eu/ ). These data provide the eastward (u) and northward (v) components of surface velocity, derived from the gradients of sea surface height under geostrophic approximation. 2.2 Validation data set High-Resolution SSS for Gulf Stream : To validate the algorithm, we use a high-resolution SSS product for the Gulf Stream region produced by Barceló et al., 2021. In their study, a Lagrangian advection technique was applied to SMAP SSS and altimetry currents to reconstruct SSS at finer scale, using a 7-day backward advection. The resulting high-resolution SSS product covers the Gulf Stream region for the years 2017, 2018, and 2019. This dataset serves as an independent benchmark; since it was produced with a similar methodology by an independent group, we can directly compare reconstructed fields against theirs to assess the performance of our in-house algorithm. Thermo-Salinograph in situ SSS : We also use in situ SSS measurements from ship-based thermo-salinographs (TSG) for validation in the Bay of Bengal. Thermo-salinographs continuously measure sea surface salinity (and temperature) from water pumped through a ship’s intake, providing high temporal and spatial resolution data along the ship’s track with high accuracy. We obtained the quality-controlled TSG data from ship cruises conducted in the Bay of Bengal during September, 2015 (under the Ocean Monsoon Mixing (OMM) program (Wijesekera et. al. 2016 ) and May, 2024 (Ekamsat cruise). 2.3 Methodology 2.3.1 Passive Lagrangian Reconstruction Technique In this approach, sea surface salinity is treated as a passive tracer transported solely by horizontal advection of surface currents, with no local sources or sinks and neglecting mixing over short timescales. Based on this assumption, finer-scale salinity fields can be reconstructed by advecting observed surface salinity field with surface currents. Two approaches are commonly employed in such Lagrangian reconstructions: (i) backward advection (Barceló et al., 2021) and (ii) forward advection (Dencausse et al., 2014 ). In one-way backward advection, we begin by seeding a dense grid of virtual particles at the target time T. We then integrate each particle’s trajectory backward over the chosen advection window T to \(\:T-{\Delta\:}T\) . At that origin point, we sample salinity from the coarse SMAP field and carry that value forward with the particle to time T. Because the particles trace out fine‐scale deformations by the altimeter‐derived flow, this process effectively interpolates the 25 km SMAP salinity onto a 4 km grid, yielding a high‐resolution salinity map at T without ever “inventing” new tracer–each particle simply brings its assigned SMAP salinity forward along its trajectory. The forward advection approach, conversely, would start at \(\:{T}_{0}\) and advect particles forward to \(\:{T}_{f}\) , carrying the salinity from \(\:{T}_{0}\) to later times. In this study, we primarily use backward advection for passive advection, but forward advection in active–advection case. The equations of motion for the passive advection of fluid particles on the sphere (longitude–latitude coordinates) under the geostrophic approximation are given by: $$\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\frac{\partial\:\varphi\:\left(\varphi\:,\:\theta\:,\:t\right)}{\partial\:t}\:\:=\frac{u}{{R}_{e}cos\left(\theta\:\right)}$$ 1 $$\:\frac{\partial\:\theta\:\left(\varphi\:,\:\theta\:,\:t\right)}{\partial\:t}\:\:=\frac{v}{{R}_{e}}$$ 2 where \(\:{\upvarphi\:}\:\) is longitude, \(\:\theta\:\) is latitude, \(\:R\) is the mean radius of the Earth, and \(\:u(\varphi\:,\theta\:,t)\) and \(\:v(\varphi\:,\theta\:,t)\) are the eastward and northward components of the surface velocity field, respectively. We integrate these equations using a fourth–order Runge–Kutta numerical scheme with a time step of \(\:{\Delta\:}t\) = 0.1 day (approximately 2.4 hours). Bilinear interpolation in space and linear interpolation in time are used to estimate the velocity field at the particle’s position and time during each step. 2.3.2 Active Advection and Algorithm Modification In the standard passive scheme, salinity remains constant for each particle throughout the integration, fixed at the initial SMAP value, which might result in large deviations. In order to take into account this problem, the above method was modified such that the intermediate reconstructed salinity field is periodically compared with the concurrent SMAP satellite observations. Whenever deviations exceed a predefined threshold (1 psu in this study), particle salinity values are reset to the observed SMAP values. In essence, this approach limits the deviation of the reconstructed field from the satellite-observed field, preventing the algorithm from drifting too far due to neglected processes. This active advection method effectively uses a variable advection time for different water parcels: portions of the field that start to diverge significantly are re-initialized (a shorter advection), whereas other areas continue to be advected for the full period. By doing so, the algorithm can maintain more realistic salinity levels, especially in the presence of strong freshwater fluxes, at the cost of potentially smoothing out some advection-driven small scales. We applied this modified algorithm (hereafter referred to as active advection (AA)) in the Bay of Bengal to assess whether it improves the accuracy of the high-resolution SSS reconstruction relative to the original fixed-time algorithm. The chosen threshold, crucial for balancing the preservation of small-scale features against limiting error growth, directly influences the method's effectiveness. Flowchart for passive as well as active advection case is shown in Fig. 1 . 3. Results and discussion Having implemented both the passive (HRSSS-PA) and active (HRSSS-AA) reconstruction frameworks, we now present a tiered assessment of their skill and robustness. Before applying the developed technique using passive advection in the Bay of Bengal, the algorithm is verified by comparing the outputs with the high-resolution Gulf Stream product of Barceló et al., 2021, 3.1 Validation with an Existing Product (Gulf Stream) Figure 2. Reconstructed sea surface salinity fields using passive advection for (a) 7 February 2017 and (b) 14 March 2017. Panels show results from (i) Barceló et al., 2021, (ii) the proposed method, and (iii) the original SMAP salinity observations. Barceló et al., 2021 used a 7-day backward passive advection (Lagrangian technique) of SMAP data to reconstruct SSS in the Gulf Stream region. Several example periods were simulated using developed methodology for the Gulf Stream region and the high resolution salinity outputs were compared with the Barceló et al., 2021 product. Figure 2 compare these two datasets on two different dates in 2017: 7 February, and 14 March. The figures show a strong correspondence between our high-resolution SSS fields and those obtained by Barceló et al., 2021, both qualitatively in terms of spatial patterns and quantitatively in terms of range of salinity values. In particular, small-scale frontal features in the salinity field (for example, those highlighted by the black ellipse in Fig. 2) are reconstructed in almost the same manner by both methods, indicating that present implementation of the Lagrangian advection is reliable and correctly captures the salinity gradients. Furthermore, the characteristics formed as a result of passive advection within the black elliptical box strongly mirror the products produced by Barceló et al., 2021. The Root Mean Square Error (RMSE) values between the reconstructed field from the in-house developed algorithm and the reference field, evaluated at weekly intervals, are presented in Table 1 . The RMSE values ranged from a minimum of 0.138 psu on 28 February 2017, indicating the closest match, to a maximum of 0.181 psu on 14th March 2017, representing the largest deviation. Intermediate RMSE values were observed on 7 February (0.146 psu), 14 February (0.151 psu), 21 February (0.169 psu), and 7 March (0.162 psu), demonstrating consistent accuracy across the evaluated period. These low RMSE values, coupled with correlation coefficients consistently above 0.99, highlight the high reliability and excellent agreement of the in-house algorithm with established methodologies. Thereafter, this algorithm has been used to generate high resolution SSS for considered domain of interest, which is Bay of Bengal. Table 1 Correlation between reconstructed field by in-house developed algorithm and by Barceló et al., 2021 for integration time of 7 days using backward passive advection in Gulf Stream. Date Correlation Root Mean Square Error (in psu) 7 Feb, 2017 0.992 0.146 14 Feb, 2017 0.990 0.151 21 Feb, 2017 0.991 0.169 28 Feb, 2017 0.992 0.138 7 Mar, 2017 0.991 0.162 14 Mar, 2017 0.991 0.181 3.2 Sensitivity study of HRSSS product with respect to spatial resolution of reconstructed product To assess the impact of spatial resolution on reconstruction accuracy, a sensitivity analysis was conducted by generating HRSSS products at four different spatial scales—25 km, 10 km, 4 km, and 2 km—through interpolation of the velocity fields. These HRSSS products, along with SMAP sea surface salinity, were compared against in situ data collected along a ship track on various days in 2024, as illustrated in Fig. 3 . The root mean square error (RMSE) was calculated between each HRSSS product and the in situ data, as well as between SMAP salinity and the in situ data. For each validation day, the TSG provided approximately 86,400 along-track measurements, enabling a robust point-by-point comparison with the HRSSS reconstructions. From Fig. 3 , it is evident that the sea surface salinity product generated using velocity fields at a spatial resolution of 25 km shows the lowest accuracy compared to in situ data across all reconstruction cases. As the spatial resolution is refined from 25 km to 10 km, and subsequently to 4 km, there is a clear and consistent improvement in the accuracy of the HRSSS product. The reduction in root mean square error (RMSE) with increasing resolution demonstrates that finer resolution allows for a more precise capture of the underlying physical processes influencing sea surface salinity. However, when the resolution is further increased from 4 km to 2 km, the improvement in accuracy becomes negligible. The lack of significant reduction in RMSE between the 4 km and 2 km resolutions suggests that the benefits of increasing resolution may plateau at some point, beyond which additional refinement does not yield proportionate gains in accuracy. This finding is critical for determining the optimal resolution for HRSSS product generation, balancing computational resources with the need for accurate data representation. Furthermore, in most cases, the RMSE between SMAP-derived sea surface salinity and in situ observations is higher than that of the reconstructed high-resolution salinity fields, highlighting better performance of the reconstruction approach. An important limitation of the current reconstruction method is its reliance on CMEMS geostrophic currents at a quarter-degree (~ 25 km) resolution. This coarse velocity field constrains the smallest flow-driven variability that can be resolved, meaning that while present algorithm produces salinity fields on a 4 km grid, the evolution of sub-mesoscale features is still governed by mesoscale currents, which miss variability below ~ 20–25 km. Consequently, the sharp gradients in reconstruction largely arise from the stretching of salinity fronts by mesoscale flows rather than explicitly resolved sub-mesoscale currents, explaining why accuracy gains diminish beyond 4 km resolution. Future improvements may come from incorporating higher-resolution currents, for instance from numerical models or upcoming missions such as SWOT, which is expected to provide ocean currents at ~ 10 km scales, thereby enhancing the fidelity of reconstructed salinity fields. 3.3 Selection of optimal advection time for reconstruction of high resolution sea surface salinity in the Bay of Bengal Previous applications of the Lagrangian advection technique in different regions have revealed that the optimal integration period varies depending on local ocean dynamics, underscoring the need to determine a region-specific value for the Bay of Bengal. For instance, in the Gulf Stream, a 7-day advection period was found to produce the best results (Barceló et al., 2021), whereas south of Tasmania an optimal period of roughly 14 days was reported (Dencausse et al., 2014 ). These variations suggest that a universal advection time cannot be assumed and must be tailored to each region's physical processes. The primary objective of reconstructing HRSSS product is to obtain tracer fields that resolve sub-mesoscale structures while remaining consistent with in situ observations. An advection period that is too short would not allow sufficient lateral stirring to develop small-scale salinity fronts, whereas an excessively long period could introduce errors as processes neglected by pure advection (e.g. air-sea fluxes, vertical and horizontal mixing, frontogenesis) begin to significantly modify the salinity field. In the latter case, sub-mesoscale features might become over-emphasized and large-scale patterns could drift away from the observed state (Dencausse et al., 2014 ). With these considerations in mind, a series of tests using integration times of 7, 14, 21, and 28 days were conducted to identify the optimal advection time for high-resolution sea surface salinity (HRSSS) reconstruction in the Bay of Bengal for passive advection of the tracer field. Figure 4 presents the reconstructed high-resolution SSS fields obtained using each of these advection periods for 24 August 2015, with panels (i)–(iv) showing the 7-, 14-, 21-, and 28-day results respectively, and panel (v) showing the original SMAP SSS field for comparison. All fields in Fig. 4 are overlaid with coincident TSG measurements collected along a ship track, providing a reference for real-world salinity gradients. This visual comparison reveals that a short advection period (7 days) fails to fully develop the finer-scale fronts, as the small-scale salinity variations along the ship track are not well represented. In contrast, an overly long integration (28 days) yields salinity fronts that are displaced spatially relative to the in situ data, indicating a bias in the position of features. The intermediate durations show a more balanced outcome: notably, the 14-day advection produces fronts that align closely with the TSG observations, suggesting that this period allows sufficient development of sub-mesoscale features without incurring large errors. Besides this, SMAP sea surface salinity shows a very weak front formation. In summary, Fig. 4 (panels i–iv) indicates that an advection time of approximately 14 days captures the salinity front development most faithfully in this region, whereas shorter or longer periods either under-develop or misplace these fronts. Figure 5. Sea surface salinity generated from HRSSS (Red Line), SMAP (Orange Line) and TSG (blue line) along ship track on 24th August 2015. A quantitative comparison with in situ data further supports the choice of a 14-day advection period. Figure 5 compares the along-track salinity profiles from the reconstructed 14-day HRSSS product (red line) and the original SMAP product (orange line) against the shipboard TSG measurements (blue line) on 24 August 2015. The 14-day HRSSS reconstruction shows excellent agreement with the in situ data, closely tracing the observed salinity variations along the ship track. In contrast, there is very less variation in the SMAP salinity field and it misses some of the finer-scale fluctuations captured by the HRSSS-PA. This is reflected in the root-mean-square (RMS) differences: the HRSSS-PA product’s error relative (RMSE: 0.19) to the in situ measurements is substantially lower than that of the SMAP product (RMSE: 0.47). In effect, the reconstructed high-resolution salinity field not only matches the ship-based observations more closely, but also reproduces the variability of sea surface salinity with higher fidelity. These results confirm that an advection period on the order of two weeks is the optimal choice for the Bay of Bengal, as it yields a high-resolution salinity field that best reconciles the need for sub-mesoscale detail with consistency to direct observations. 3.4 Evaluation of High-Resolution Sea Surface Salinity Using Along-Track TSG Data from EKAMSAT 2024 under Passive Advection Conditions This section presents a detailed assessment of the reconstructed high-resolution sea surface salinity (HRSSS) fields derived from passive advection techniques using in situ salinity measurements collected during the EKAMSAT research cruise during May 2024. Figure 6 displays a three-panel visualization of sea surface salinity (SSS) along the ship track. Panel (i) shows the in situ SSS obtained from the EKAMSAT TSG, serving as the reference for evaluating satellite products. Panel (ii) illustrates the coinciding SMAP Level 3 gridded salinity, a widely used satellite product. Panel (iii) presents the HRSSS-PA estimates reconstructed using passive advection. Qualitatively, HRSSS-PA more closely follows the finer-scale structures and frontal gradients observed in the in situ data, whereas SMAP appears spatially smoother and less responsive to small-scale salinity variability. Quantitative evaluation using root mean square error (RMSE) and Pearson correlation coefficient (R) further supports these visual observations. Compared to in situ data, SMAP yields an RMSE of 2.76 psu with a correlation of 0.49, suggesting moderate agreement but limited accuracy in reproducing spatial detail. In contrast, HRSSS-PA demonstrates improved performance, achieving an RMSE of 1.97 psu and a stronger correlation of 0.60. These results suggest that the HRSSS-PA product offers a more reliable representation of SSS in coastal and dynamically active regions. Figure 7. Daily along-track profiles of sea surface salinity (SSS) retrieved from HRSSS-PA, SMAP, and in situ observations for four representative periods in April-May 2024. Each panel corresponds to a specific date range: (i) 28 & 29 April 2024, (ii) 30 April 2024, (iii) 3 & 4 May 2024, and (iv) 9 & 10 May 2024. Salinity measurements are plotted as a function of longitude along the ship track. Figure 7 extends this comparison to four selected periods during April-May 2024 to assess temporal variability in product performance. For 28–29 April 2024 [panel (i)], HRSSS-PA achieves an RMSE of 2.48 and R = 0.52, compared to SMAP's RMSE of 3.52 and a weak negative correlation (R = − 0.11), confirming HRSSS-PA's stronger consistency in both magnitude and structure. On 30 April 2024 [panel (ii)], HRSSS-PA yields an RMSE of 3.19 (R = 0.44) while SMAP records a higher RMSE of 4.78 (R = 0.32), indicating better performance by HRSSS-PA despite challenging conditions. For 3–4 May 2024 [panel (iii)], HRSSS-PA and SMAP both exhibit lower RMSEs (2.08 and 2.43, respectively), though correlation values remain weak (R = 0.12 and 0.05), suggesting limitations in capturing finer-scale salinity variability. The best agreement is observed during 9–10 May 2024 [panel (iv)], where HRSSS-PA shows an RMSE of 0.89 and a strong correlation of 0.87, while SMAP performs comparably well with an RMSE of 1.36 and a higher correlation of 0.91. Because the SMAP Level-3 product is an 8-day running mean, it smooths day-to-day variability; relative to instantaneous ship measurements, this temporal averaging can lower correlations and inflate apparent errors during rapidly evolving frontal events (Figs. 6 –7). Collectively, these results highlight the superior accuracy and effective resolution of HRSSS-PA during relatively stable periods, while performance degrades for both products under more dynamic conditions—underscoring the continuing need for high-resolution in situ observations to validate and enhance satellite SSS reconstruction algorithms. In brief, the baseline passive reconstruction (HRSSS-PA) sharpens mesoscale fronts by treating SSS as a strictly conserved tracer over the integration window and advecting SMAP with altimetry-derived currents, yielding high-resolution SSS for the Bay of Bengal. 3.5 Evaluation of High-Resolution Sea Surface Salinity Using Along-Track TSG Data from R/V Roger-Revelle 2015 under Passive Advection Conditions To further evaluate the generalizability of the HRSSS-PA reconstruction method, an independent validation was performed using historical salinity observations collected aboard the R/V Roger Revelle in August–September 2015. This complementary analysis provides a contrasting testbed featuring a different oceanographic regime and time period, thereby allowing assessment of the HRSSS-PA product's robustness under varying conditions. Figure 8 presents a side-by-side comparison of sea surface salinity (SSS) profiles retrieved from HRSSS-PA, SMAP, and in situ observations along the ship track on four selected days during the cruise. In panel (i), for 24 August 2015, the HRSSS-PA reconstruction aligns closely with the in situ salinity profile throughout the observed track. It effectively captures both the general trend and sharper gradients, particularly between longitudes 84.1° and 84.4°, where a distinct salinity drop occurs. The SMAP estimate, in contrast, displays a notably smoother gradient and underestimates the sharp salinity drop, indicating loss of spatial detail. This visual agreement with in situ data corresponds well with the reported low RMSE (0.19) and high correlation (R = 0.93) for HRSSS-PA. Although SMAP achieves a slightly higher correlation (R = 0.96), its RMSE is more than double that of HRSSS-PA (0.47), emphasizing its reduced accuracy in magnitude. In panel (ii), corresponding to 26 August 2015, HRSSS-PA again tracks the in situ salinity profile more closely, particularly in the central portion of the transect (84.2°–84.5°), where small-scale fluctuations are better represented than in the SMAP product. While SMAP maintains a consistently smooth profile, it fails to capture the local salinity troughs and peaks observed in in situ. Numerically, HRSSS-PA exhibits a slightly higher RMSE (0.248) than SMAP (0.213), yet the correlation coefficient is markedly higher for HRSSS-PA (R = 0.53 vs. 0.16 for SMAP), reflecting its superior representation of spatial variability. A different behavior emerges in panel (iii), corresponding to 5 September 2015, during the peak of freshwater discharge from the Ganga-Brahmaputra river system. The in situ profile shows sharp, step-like salinity transitions between ~ 29.4 and 30.5 psu, likely associated with freshwater lenses and dynamic fronts. HRSSS-PA, which relies on passive advection, significantly overestimates salinity throughout the transect and fails to resolve the sharp gradients or structural variability. In contrast, SMAP-despite its coarse resolution-better captures the pattern and magnitude of the observed transitions, albeit with some smoothing. Statistically, HRSSS-PA records a high RMSE of 1.74 psu (R = 0.69), while SMAP achieves a lower RMSE of 0.65 psu and higher correlation (R = 0.79), indicating relatively improved performance under these more complex conditions. Panel (iv), dated 19 September 2015, shows both HRSSS-PA and SMAP generally follow the trend of the in situ salinity field, with HRSSS-PA offering closer agreement, especially across longitudes 84.8° to 85.6°. Minor discrepancies remain in capturing localized dips and peaks. HRSSS-PA achieves an RMSE of 0.2 psu and a moderate correlation of R = 0.44, while SMAP underrepresents the observed variability, with a higher RMSE (0.326) and a weak negative correlation (R = -0.192), indicating reduced reliability in this case. Collectively, Fig. 8 demonstrates that HRSSS-PA generally outperforms SMAP in reproducing both absolute salinity values and spatial gradients, particularly under stable or moderately variable conditions (e.g., 24 and 26 August). The high-resolution SSS fields generated here reveal much finer spatial detail in the Bay of Bengal, underscoring the utility of a Lagrangian advection framework for bridging the gap between coarse satellite products and the finer-scale processes of interest. This approach offers a promising means of exploiting existing satellite observations to improve monitoring of freshwater dynamics and mesoscale variability in the Bay of Bengal, with potential applications for similar downscaling strategies in other oceanic regions. However, its limitations become evident under highly dynamic scenarios such as 5 September, where passive advection fails to resolve the fine-scale features shaped by active frontal displacement and freshwater input. Passive Lagrangian reconstruction assumes surface salinity is strictly conserved over the 14-day integration, neglecting freshwater fluxes, rainfall, evaporation, and mixing. In the Bay of Bengal, these processes are often comparable to horizontal advection, causing rapid divergence from reality. Such episodic fluxes and unresolved mixing set a hard limit on passive reconstructions; active scheme mitigates this by resetting particles when errors grow, but advection alone cannot capture the full salt budget under strong freshwater forcing. 3.6 Evaluation of High-Resolution Sea Surface Salinity Using Along-Track TSG Data from Roger-Revelle 2015 under active advection conditions Continuing from the previous analysis, Figs. 7 and 8 demonstrate that the HRSSS-PA product aligns closely with in situ salinity measurements across several observational days, and typically outperforms SMAP-derived salinity fields. This consistency highlights the dominant role of passive advection during certain periods. However, notable deviations occur, particularly on 5 September 2015 (Fig. 8iii), when HRSSS-PA shows larger discrepancies compared to SMAP. These mismatches point to the limitations of a purely passive advection scheme, which omits additional oceanographic dynamics influencing salinity. To address these observed discrepancies and improve the representation of oceanographic processes, an alternative reconstruction that explicitly incorporates active advection was implemented. In this section, we evaluate the effectiveness and accuracy of this active advection-based reconstruction by comparing its performance against both in situ salinity data collected along a research vessel transect in the northern Bay of Bengal and SMAP salinity observations for specific period (5 Sep, 2015), as depicted in Fig. 9. This region is heavily influenced by the seasonal freshwater discharge from the Ganga-Brahmaputra river system and is known for its sharp salinity gradients and complex surface circulation patterns. Figure 9. Daily along-track profiles of sea surface salinity (SSS) retrieved from HRSSS-AA, SMAP, and in situ observations plotted as a function of longitude along the ship track. As observed in the figure, the in situ salinity profile exhibits strong spatial variability, with values ranging from approximately 29.4 to 30.5 psu, reflecting the influence of riverine freshwater lenses and frontal boundaries. SMAP retrievals (cyan dashed line), limited by a coarse spatial footprint (~ 40 km), capture the broad gradients but consistently overestimate salinity across much of the transect, with a bias of approximately 0.2–0.8 psu. In contrast, the HRSSS-AA retrieval (red dotted line) shows a marked improvement over HRSSS-PA in resolving the structure and magnitude of these variations. On the same day (see Section 3.5), HRSSS-PA exhibited a relatively high RMSE of 1.74 psu and a moderate correlation (R = 0.69), underscoring its limitations under dynamic conditions. The transition to active advection in HRSSS-AA effectively overcomes these issues by treating salinity as a field actively influenced by horizontal flow and sub-mesoscale variability. Quantitatively, HRSSS-AA achieves a reduced RMSE of 0.79 psu, compared to 0.65 psu for SMAP, highlighting the value of dynamic modeling even when absolute errors may appear similar. This emphasizes the value of active dynamics in maintaining realistic salinity patterns, even when absolute error may appear slightly elevated. Beyond these metrics, the active-advection approach enhances the physical realism of reconstructed fields. For instance, along-track SSS on 5 September 2015 reveals sharp, step-like transitions and filaments near 89.56–89.60°E (Fig. 9), characteristic of buoyant freshwater lenses and monsoonal frontal boundaries driven by Ganga-Brahmaputra-Meghna discharge. The interaction of this plume with mesoscale eddies and sub-mesoscale instabilities sharpens these gradients into filaments and narrow fronts (Boccaletti et al., 2007 ; Mahadevan, 2016 ; McWilliams, 2016 ), as commonly observed in the northern Bay of Bengal during the post-monsoon season (Akhil et al., 2014 ; Ramachandran et al., 2018 ; Sengupta et al., 2016 ). The ability of HRSSS-AA to resolve such features demonstrates not only reduced statistical error but also improved representation of physically meaningful processes, plume spreading, frontal sharpening, and eddy-driven filamentation, that are critical for understanding salinity variability, stratification, and air-sea fluxes in river-influenced coastal systems. 4. Summary and Conclusions This study addresses the challenge of reconstructing high-resolution sea surface salinity (SSS) fields from coarse satellite data in the Bay of Bengal, a region shaped by complex circulation, strong monsoonal rainfall, and large freshwater discharge. Using a Lagrangian framework based on passive and active advection, we reconstructed fine-scale salinity structures from SMAP Level-3 SSS and satellite-derived geostrophic currents. This approach provides a new window into upper-ocean processes in regions with strong rainfall and river discharge, offering improvements over conventional satellite products The passive advection method was applied to generate high-resolution salinity fields under stable to moderately dynamic oceanographic conditions. To assess the correctness of our implementation, we compared our results against a reference high-resolution salinity product produced using the same underlying technique by Barceló et al. (2021) over the Gulf Stream. The two independently generated products exhibited strong agreement, with root mean square errors (RMSE) between 0.14–0.18 psu and correlation coefficients consistently above 0.99. Sensitivity analyses indicated an optimal spatial resolution of approximately 4 km, which captured finer oceanographic features effectively, whereas further refinement to 2 km offered minimal improvement. An integration window of 14 days was identified as optimal, balancing detailed sub-mesoscale structure development against numerical drift risks in Bay of Bengal. Validation using in situ thermo-salinograph data from the EKAMSAT-2024 cruise confirmed the improved performance of the passive advection approach. Compared to SMAP, the HRSSS-PA product showed significantly better agreement with in-situ measurements, reducing the cruise-mean RMSE from 2.76 psu (SMAP) to 1.97 psu (HRSSS-PA) and improving the correlation coefficient from 0.49 (SMAP) to 0.60 (HRSSS-PA). Detailed assessments on specific dates highlighted consistent performance, notably an RMSE of 0.89 psu (R = 0.87) during stable conditions. Validation against thermo-salinograph data from the Roger Revelle cruise (24–26 August 2015) showed similar successes, significantly reducing RMSE from 0.466 psu (SMAP) to as low as 0.186 psu (HRSSS-PA). However, during highly dynamic events with intense freshwater input, such as on 5 September 2015, passive advection alone proved inadequate (RMSE = 1.74 psu). Incorporating active advection in such cases significantly improved performance (RMSE = 0.79 psu), enabling sharper frontal resolution and dynamic response. Collectively, these results demonstrate that passive advection is well-suited for relatively stable regimes, while active advection is necessary to capture the complexity of freshwater-driven variability. The combined methodology offers a practical and accurate approach for high-resolution SSS reconstruction in challenging oceanographic settings. Lagrangian advection has proven effective for high-resolution SSS mapping, but its limitations must be acknowledged. The approach is sensitive to errors in the SMAP satellite salinity inputs, particularly in dynamic, freshwater-influenced regions like the Bay of Bengal, where strong salinity gradients and heavy rainfall can exacerbate retrieval uncertainties. Moreover, the passive advection assumption breaks down under intense freshwater forcing events (e.g. monsoonal rain or major river discharge), since new low-salinity inputs and mixing processes are not represented. This has led to notable discrepancies between reconstructed salinity fields and in situ thermo-salinograph measurements during such events. Nonetheless, future high-resolution SSS missions offer reason for optimism. Upcoming sensors like the Copernicus Imaging Microwave Radiometer (CIMR) will provide more precise, finer-scale salinity observations, which should help reduce input errors and significantly improve the accuracy of SSS reconstructions. Declarations Funding Declaration This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. Author Contribution Jai Kumar: Conceptualization, Methodology, Software, Formal Analysis, Investigation, Data Curation, Visualization, Writing – Original Draft, Writing – Review & EditingNeeraj Agarwal: Conceptualization, Writing – Review & Editing, SupervisionRashmi Sharma: Conceptualization, Writing – Review & Editing, Supervision, Project Administration Acknowledgments The authors would like to express their sincere gratitude to the Director, Space Applications Centre, for motivation. Velocity field data and SMAP sea surface salinity used in the work are taken from https://resources.marine.copernicus.eu/ and https://search.earthdata.nasa.gov/ , respectively. Data Availability Generated Data can be downloaded from the link below and it is free of use-https://www.mosdac.gov.in/high-resolution-sea-surface-salinity References Akhil VP, Durand F, Lengaigne M, Vialard J, Keerthi MG, Gopalakrishna VV, Deltel C, Papa F, de Boyer Montégut C (2014) A modeling study of the processes of surface salinity seasonal cycle in the Bay of Bengal. J Geophys Research: Oceans 119(6):3926–3947 Barceló-Llull B, Drushka K, Gaube P (2021) Lagrangian reconstruction to extract small-scale salinity variability from SMAP observations. J Geophys Research: Oceans, 126, e2020JC016477. Berti S, Lapeyre G (2014) Lagrangian reconstructions of temperature and velocity in a model of surface ocean turbulence. Ocean Model 76:59–71 Boccaletti G, Ferrari R, Fox-Kemper B (2007) Mixed layer instabilities and restratification. J Phys Oceanogr 37(9):2228–2250 Cherian DA, Shroyer EL, Wijesekera HW, Moum JN (2020) The seasonal cycle of upper-ocean mixing at 8° N in the Bay of Bengal. J Phys Oceanogr 50(2):323–342 Chin TM, Vazquez-Cuervo J, Armstrong EM (2017) A multi-scale high-resolution analysis of global sea surface temperature. Remote Sens Environ 200:154–169 Dencausse G, Morrow R, Rogé M, Fleury S (2014) Lateral stirring of large-scale tracer fields by altimetry. Ocean Dyn 64(1):61–78 Desprès A, Reverdin G, d'Ovidio F (2011) Mechanisms and spatial variability of mesoscale frontogenesis in the northwestern subpolar gyre. Ocean Model 39(1):97–113 Drushka K, Asher WE, Sprintall J, Gille ST, Hoang C (2019) Global patterns of submesoscale surface salinity variability. J Phys Oceanogr 49(7):1669–1685 Kumar J, Ratheesh S, Agarwal N, Sharma R (2023) Study of upwelling and mixing process in the Somali coastal region using satellite and numerical model observations: A Lagrangian approach. Deep Sea Research Part II, vol 216. Topical Studies in Oceanography, p 105381 Kumar J, Choudhary RK, Mathur M, Agarwal N, Sharma R (2024) A Study of Mixing and Biological Activity in the North Indian Ocean Using Finite Size Lyapunov Exponents. J Indian Soc Remote Sens 51(2):395–403 Lévy M, Klein P, Treguier AM (2001) Impact of sub-mesoscale physics on production and subduction of phytoplankton in an oligotrophic regime. J Mar Res 59(4):535–565 Liang Z, Bao S, Zhang W, Wang H, Yan H, Dai J, Xiao P (2025) Spatiotemporal Super-Resolution of Satellite Sea Surface Salinity Based on a Progressive Transfer Learning-Enhanced Transformer. Remote Sens 17(15):2735 Lucas AJ, Nash JD, Pinkel R, MacKinnon JA, Tandon A, Mahadevan A, Omand MM, Freilich M, Sengupta D, Ravichandran M, Le Boyer A (2016) Adrift upon a salinity-stratified sea: A view of upper-ocean processes in the Bay of Bengal during the southwest monsoon. Oceanography 29(2):134–145 Mahadevan A (2016) The impact of submesoscale physics on primary productivity of Plankton. Annual Rev Mar Sci 8:161–184 Mahadevan A, Jaeger GS, Freilich M, Omand MM, Shroyer EL, Sengupta D (2016) Freshwater in the Bay of Bengal: Its fate and role in air-sea heat exchange. Oceanography 29(2):72–81 McWilliams JC (2016) Submesoscale currents in the ocean. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 472(2189), 20160117 Meissner T, Wentz FJ, Manaster A, Lindsley R (2019) Remote sensing systems SMAP ocean surface salinities [level 2c, level 3 running 8-day, level 3 monthly]. Remote Sensing Systems, Santa Rosa, CA, USA Ramachandran S, Tandon A, MacKinnon J, Lucas AJ, Pinkel R, Waterhouse AF, Nash J, Shroyer E, Mahadevan A, Weller RA, Farrar JT (2018) Submesoscale processes at shallow salinity fronts in the Bay of Bengal: Observations during the winter monsoon. J Phys Oceanogr 48(3):479–509 Ratheesh S, Agarwal N, Chaudhary A, Lijin J, Sree Lekha J, Mathur M, Sharma R, Kumar R (2020) Response of a high-resolution ocean circulation model to winds from different sources in simulating summer monsoon freshening in the North Bay of Bengal: A case study, Deep Sea Research Part II, vol 172. Topical Studies in Oceanography, p 104727 Reul N, Grodsky SA, Arias M, Boutin J, Catany R, Chapron B, D’Amico F, Dinnat E, Donlon C, Fore A et al (2020) Sea Surface Salinity Estimates from Spaceborne L-Band Radiometers: An Overview of the First Decade of Observation (2010–2019). Remote Sens Environ 242:111769 Rogé M, Morrow RA, Dencausse G (2015) Altimetric Lagrangian advection to reconstruct Pacific Ocean fine-scale surface tracer fields. Ocean Dyn 65(9):1249–1268 Rudnick DL, Ferrari R (1999) Compensation of horizontal temperature and salinity gradients in the ocean mixed layer. Science 283(5401):526–529 Sengupta D, Raj GNB, Ravichandran M, Sree Lekha J (2016) Near-surface salinity and stratification in the north Bay of Bengal from moored observations. Geophys Res Lett 43(9):4448–4456 Vinogradova N, Lee T, Boutin J, Drushka K, Fournier S, Sabia R, Stammer D, Bayler E, Reul N, Gordon A et al (2019) Satellite Salinity Observing System: Recent Discoveries and the Way Forward. Front Mar Sci 6:243 Wang Z, Wang G, Guo X, Hu J, Dai M (2022) Reconstruction of High-Resolution Sea Surface Salinity over 2003–2020 in the South China Sea Using the Machine Learning Algorithm LightGBM Model. Remote Sens 14(23):6147 Weller RA, Farrar JT, Seo H, Prend C, Sengupta D, Sree Lekha J, Ravichandran M, Venkatesan R (2019) Moored observations of the surface meteorology and air–sea fluxes in the northern Bay of Bengal in 2015. J Clim 32(2):549–573 Wijesekera HW, Shroyer E, Tandon A, Ravichandran M, Sengupta D, Jinadasa SUP, Whalen CB (2016) ASIRI: An ocean–atmosphere initiative for Bay of Bengal. Bull Am Meteorol Soc 97(10):1859–1884 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 06 Dec, 2025 Reviewers agreed at journal 03 Dec, 2025 Reviewers agreed at journal 03 Dec, 2025 Reviewers invited by journal 03 Dec, 2025 Editor assigned by journal 02 Dec, 2025 Submission checks completed at journal 02 Dec, 2025 First submitted to journal 01 Dec, 2025 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-8247078","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":554769372,"identity":"e1e77bb6-1ec5-4c57-a8dc-75814dc2cf05","order_by":0,"name":"Jai Kumar","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA70lEQVRIiWNgGAWjYBACCTDJYwOmDsCFEwhrSSNZC8NhEhwmOSM78XOFzHl7/vYDjAd/7rFhkHfvMWB4uAO3FmmJ3M2SZ3huJ844k8BwmOdZGoPhmTMGDIlncGuRk8jdINnAczuB4QbIeQcOMxjOSEtgSGzDq2Xzzwaec/byQC0HfxCjBeiwbUBbDjBuAGo5wAPUIi+RfACvFsmet9ssG3iSEzeeSWw4zHMgjceA5/CBA/i0SBzP3XyzscfOXu744cMffxywkZNvb2x8+BOPFjBg7AGTDSCSx+AAcpziBD+Q2PINhNWPglEwCkbByAIAYP9TXBy6EZIAAAAASUVORK5CYII=","orcid":"","institution":"Space Applications Centre, ISRO","correspondingAuthor":true,"prefix":"","firstName":"Jai","middleName":"","lastName":"Kumar","suffix":""},{"id":554769373,"identity":"c8ac821c-b162-46d0-b83b-8c9e9a1036db","order_by":1,"name":"Neeraj Agarwal","email":"","orcid":"","institution":"Space Applications Centre, ISRO","correspondingAuthor":false,"prefix":"","firstName":"Neeraj","middleName":"","lastName":"Agarwal","suffix":""},{"id":554769374,"identity":"1e754999-314b-4e42-9725-c9e91da25f4e","order_by":2,"name":"Rashmi Sharma","email":"","orcid":"","institution":"Space Applications Centre, ISRO","correspondingAuthor":false,"prefix":"","firstName":"Rashmi","middleName":"","lastName":"Sharma","suffix":""}],"badges":[],"createdAt":"2025-12-01 07:08:41","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8247078/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8247078/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":97687175,"identity":"d0fa878d-a543-4929-ae6d-10a979bfbe45","added_by":"auto","created_at":"2025-12-08 10:25:06","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2022023,"visible":true,"origin":"","legend":"","description":"","filename":"ManuscriptFinalOceanDynamics.docx","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/4f58e0ade4f7577acf4faf0b.docx"},{"id":97687233,"identity":"96104967-06b9-474f-abf8-bcd2731e7997","added_by":"auto","created_at":"2025-12-08 10:25:09","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5673,"visible":true,"origin":"","legend":"","description":"","filename":"25fcb34f640b4de59455d41f00ee9180.json","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/dcd1233246dcc26b7e62c782.json"},{"id":97687180,"identity":"e3990f4d-426b-4e3d-94c6-7c9feab23fda","added_by":"auto","created_at":"2025-12-08 10:25:07","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":99113,"visible":true,"origin":"","legend":"","description":"","filename":"25fcb34f640b4de59455d41f00ee91801enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/06e25278515e428e2aff8791.xml"},{"id":97687237,"identity":"bbb7586d-555e-4bab-bc3d-461df10dff4e","added_by":"auto","created_at":"2025-12-08 10:25:09","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":82915,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/61ac82cd233f21d323987f0f.png"},{"id":97892887,"identity":"fec835b4-2837-40e2-8ffe-7f012de9dfef","added_by":"auto","created_at":"2025-12-10 15:23:54","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":86477,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/ca90a5712b3caca07f5a2526.png"},{"id":97892531,"identity":"39e03742-6cbf-4dec-bc9c-dd65047a0e41","added_by":"auto","created_at":"2025-12-10 15:13:22","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":23616,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/7ea2b4460bfca0eb07ba558b.png"},{"id":97687221,"identity":"0efe667b-584b-4482-aa85-97cff431931e","added_by":"auto","created_at":"2025-12-08 10:25:09","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":26414,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/fcc5cdb615c816e123016762.png"},{"id":97687135,"identity":"e547e194-ef39-48bc-ae14-6d3acf82fe4a","added_by":"auto","created_at":"2025-12-08 10:25:04","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19817,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/0b73c32d3b53489d59668993.png"},{"id":97687171,"identity":"4e857e27-e54b-4f5f-9048-cc25776993e7","added_by":"auto","created_at":"2025-12-08 10:25:05","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":53412,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/7be65022e6deb618c8b0e1e9.png"},{"id":97687234,"identity":"267f1452-8669-4bf0-98aa-374493bb13c5","added_by":"auto","created_at":"2025-12-08 10:25:09","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":34500,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/92bb5cf4ebcbc734f53a3467.png"},{"id":97687179,"identity":"4830dd22-45fd-424d-86dd-279d115a1cb8","added_by":"auto","created_at":"2025-12-08 10:25:07","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":49875,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/5aead278acb5461e2248d79f.png"},{"id":97687178,"identity":"eca7bbd8-3b20-4196-969a-95c6d50599aa","added_by":"auto","created_at":"2025-12-08 10:25:07","extension":"png","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":15858,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage9.png","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/324e02f68106361a79ba2e73.png"},{"id":97687173,"identity":"29ce57e9-97fd-46e2-b073-386479d734f2","added_by":"auto","created_at":"2025-12-08 10:25:05","extension":"xml","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":98478,"visible":true,"origin":"","legend":"","description":"","filename":"25fcb34f640b4de59455d41f00ee91801structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/236b28794a79526d0f3dbf1d.xml"},{"id":97687215,"identity":"f84bf9b6-5774-402b-bed4-ea622f378c99","added_by":"auto","created_at":"2025-12-08 10:25:08","extension":"html","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":105188,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/3c5af105a894bcb093541afa.html"},{"id":97687241,"identity":"a3ee9843-8dc9-432c-a179-1a94e77741c4","added_by":"auto","created_at":"2025-12-08 10:25:10","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":316304,"visible":true,"origin":"","legend":"\u003cp\u003eFlow Diagram for passive and active advection\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/cff3b67a74a6a1ab726d892a.jpeg"},{"id":97687174,"identity":"c0b29d48-7c11-4a64-b3e6-69fde0c8292f","added_by":"auto","created_at":"2025-12-08 10:25:06","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":308896,"visible":true,"origin":"","legend":"\u003cp\u003eReconstructed sea surface salinity fields using passive advection for (a) 7 February 2017 and (b) 14 March 2017. Panels show results from (i) Barceló et al., 2021, (ii) the proposed method, and (iii) the original SMAP salinity observations.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/b087dc263d7521417ba4dfcd.png"},{"id":97687218,"identity":"b13e153f-8222-4d7b-a5c4-1b0942d33302","added_by":"auto","created_at":"2025-12-08 10:25:08","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":85927,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of root mean square error between HRSSS products generated at different spatial resolutions and in situ data, alongside the comparison between SMAP sea surface salinity and in situ data from EKAMSAT 2024 ship cruise.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/e2e3b922a7c20ecfdbcbe8a1.png"},{"id":97687217,"identity":"95e6a62a-508f-4477-b49a-6005da7be99b","added_by":"auto","created_at":"2025-12-08 10:25:08","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":308517,"visible":true,"origin":"","legend":"\u003cp\u003eReconstructed HRSSS-PA field overlay with TSG salinity along ship track (i)-(v) and original SMAP sea surface salinity on 24th August, 2015\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/0490ee70bb477a7804fad68c.png"},{"id":97687169,"identity":"12f33ce4-d43e-4b42-bc9f-6ba7fc9cba40","added_by":"auto","created_at":"2025-12-08 10:25:05","extension":"jpeg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":102049,"visible":true,"origin":"","legend":"\u003cp\u003eSea surface salinity generated from HRSSS (Red Line), SMAP (Orange Line) and TSG (blue line) along ship track on 24th August 2015.\u003c/p\u003e","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/030bff47edbfdee0a535e402.jpeg"},{"id":97687172,"identity":"505fc72b-fb99-443c-b156-c1673a3d7760","added_by":"auto","created_at":"2025-12-08 10:25:05","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":475349,"visible":true,"origin":"","legend":"\u003cp\u003eSea surface salinity along the ship track for Apr-May 2024 from (i) EKAMSAT TSG, (ii) SMAP SSS, and (iii) HRSSS-PA for the complete ship-track.\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/f7b986e0f3c9c970d24e4f0a.png"},{"id":97687214,"identity":"17d46c08-9afb-464c-90d2-76b184089042","added_by":"auto","created_at":"2025-12-08 10:25:08","extension":"jpeg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":171317,"visible":true,"origin":"","legend":"\u003cp\u003eDaily along-track profiles of sea surface salinity (SSS) retrieved from HRSSS-PA, SMAP, and in situ observations for four representative periods in April-May 2024. Each panel corresponds to a specific date range: (i) 28 \u0026amp; 29 April 2024, (ii) 30 April 2024, (iii) 3 \u0026amp; 4 May 2024, and (iv) 9 \u0026amp; 10 May 2024. Salinity measurements are plotted as a function of longitude along the ship track.\u003c/p\u003e","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/9b1092d9e469338cae91e6e6.jpeg"},{"id":97687216,"identity":"597ed806-8130-4455-aa88-383361defede","added_by":"auto","created_at":"2025-12-08 10:25:08","extension":"jpeg","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":307615,"visible":true,"origin":"","legend":"\u003cp\u003eDaily along-track profiles of sea surface salinity (SSS) retrieved from HRSSS-PA, SMAP, and in situ observations for four representative periods in Aug–Sept 2015. Each panel corresponds to a specific date range: (i) 24 August 2015, (ii) 26 August 2015, (iii) 5 September 2015, and (iv) 19 September 2015. Salinity measurements are plotted as a function of longitude along the ship track.\u003c/p\u003e","description":"","filename":"floatimage8.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/9e71e9588c701635dbfdfc77.jpeg"},{"id":97687238,"identity":"0a328cb6-6151-4456-be83-81e0f11658cf","added_by":"auto","created_at":"2025-12-08 10:25:09","extension":"jpeg","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":96907,"visible":true,"origin":"","legend":"\u003cp\u003eDaily along-track profiles of sea surface salinity (SSS) retrieved from HRSSS-AA, SMAP, and in situ observations plotted as a function of longitude along the ship track.\u003c/p\u003e","description":"","filename":"floatimage9.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/95079da92ed654bdbce8be5d.jpeg"},{"id":97902362,"identity":"8f5bbcb1-38b4-481a-b4c3-f66e5719b051","added_by":"auto","created_at":"2025-12-10 15:51:41","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2744191,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8247078/v1/54c67374-41e6-4235-955f-320bfbe57205.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"A Lagrangian-Based Technique to Generate High-Resolution Sea Surface Salinity Fields from Low-Resolution Satellite Observations: A Study in the Bay of Bengal","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eSea surface salinity (SSS) plays a critical role in regulating upper-ocean stratification, mixed-layer dynamics, and air-sea fluxes, with downstream impacts on monsoon systems, cyclone intensities, and climate variability. This is especially evident in the Bay of Bengal (BoB), where intense monsoonal rainfall and large river discharges from systems such as the Ganga-Brahmaputra-Meghna and Irrawaddy create strong near-surface salinity gradients and highly stratified upper-ocean conditions (Mahadevan et al., \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Ratheesh et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). The resulting barrier layers suppress convective overturning and entrainment cooling, altering the diurnal cycle of sea surface temperature (SST) and regional air-sea interactions.\u003c/p\u003e\u003cp\u003eDespite the dynamic significance of SSS, current satellite-derived products (e.g., from SMOS, Aquarius, and SMAP) suffer from coarse effective resolutions of ~\u0026thinsp;50\u0026ndash;100 km due to radiometric noise, land-sea contamination, and multi-day compositing windows (Reul et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Vinogradova et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). These limitations hinder the detection of mesoscale (10\u0026ndash;100 km) and sub-mesoscale (\u0026lt;\u0026thinsp;10 km) features such as freshwater plumes, fronts, and eddies that critically shape surface salinity distributions. In contrast, satellite SST and SSH routinely achieve finer than 10 km resolution (Chin et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2017\u003c/span\u003e), enabling high-fidelity studies of oceanic density fronts and mixing-processes that require co-located gradients in temperature and salinity (L\u0026eacute;vy et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2001\u003c/span\u003e; Rudnick \u0026amp; Ferrari, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e1999\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThe Bay of Bengal offers a compelling testbed for improving SSS resolution. Observations from initiatives like ASIRI\u0026ndash;OMM reveal widespread salinity-driven stratification and highlight the need for integrated measurements using ships, moorings, and autonomous platforms (Wijesekera et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Shipboard process cruises during the southwest monsoon have uncovered sub-mesoscale frontal networks embedded in the stratified surface ocean (Sharma et. al., 2016; Lucas et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Ramachandran et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), while long-term mooring records show persistent near-surface stratification maintained by runoff and mesoscale stirring (Sengupta et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). These dynamics modulate nutrient availability, upper-ocean mixing, and freshwater redistribution, underscoring the need for high-resolution SSS fields to resolve key biophysical and climate-relevant processes (Weller et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Cherian et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eTo address the coarse resolution of satellite-derived SSS products, several reconstruction techniques have emerged. Among them, Lagrangian advection methods leverage high-resolution surface current fields (from altimetry or models) to transport coarse salinity data along fluid trajectories, thereby reconstructing fine-scale tracer patterns consistent with ocean dynamics. This passive-tracer approach was first demonstrated in the northeastern Irminger Sea, where lateral stirring by mesoscale currents sharpens filamentary structures and fronts through backward‐forward advection of salinity particles (Despr\u0026egrave;s et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). The method was applied in the Southern Ocean south of Tasmania, where a\u0026thinsp;~\u0026thinsp;2-week optimal advection period balanced fine-scale detail and unmodeled processes (Dencausse et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2014\u003c/span\u003e), and later extended to the Pacific, confirming its robustness and consistent performance across dynamic regimes (Rog\u0026eacute; et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Regional studies in the Indian Ocean support the Lagrangian framework, highlighting upwelling and mixing in the Somali coast (Kumar et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) and using FSLEs to assess mixing and biological activity across the North Indian Ocean (Kumar et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). These studies underscore the capacity of Lagrangian diagnostics to resolve mesoscale-to-submesoscale features and their biological implications. Idealized turbulence studies further validated that Lagrangian reconstructions faithfully recover sub-mesoscale features in passive tracers (Berti \u0026amp; Lapeyre, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2014\u003c/span\u003e), and global implementations have since mapped sub-mesoscale salinity variability at unprecedented resolution (Drushka et al., \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Lagrangian advection, which conserves tracers and follows surface flow kinematics, provides a consistent framework for linking coarse-resolution SSS with finer SST and SSH data, enabling integrated analyses of frontogenesis and upper ocean mixing. Alongside this, recent advances in artificial intelligence have introduced data-driven methods to enhance SSS resolution (Wang et al., \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2022\u003c/span\u003e, Liang et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). While these AI-based methods offer flexible and accurate alternatives, they may trade off physical interpretability and tracer conservation for improved statistical performance.\u003c/p\u003e\u003cp\u003eThis study presents a Lagrangian-based approach to reconstruct high-resolution sea surface salinity (SSS) fields from coarse satellite data, using the Bay of Bengal as a test case. By advecting SMAP SSS with altimetry-derived currents and treating salinity as a passive tracer, the method aims to enhance the effective resolution for capturing finer-scale variability. We evaluate its performance through comparisons with existing high-resolution products and in situ observations, and further explore an adaptive advection scheme to address regional complexities. To the best of our knowledge, this is the first application of such techniques in a region with intense freshwater forcing.\u003c/p\u003e\u003cp\u003eThe paper is organized as follows: Section 2 describes the data used and the methodology; Section 3 presents the results and discussion; and Section 4 concludes with the study\u0026rsquo;s findings.\u003c/p\u003e"},{"header":"2. Data sets","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1 Data sets used in reconstruction of high resolution sea surface salinity\u003c/h2\u003e\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003e\u003cb\u003eSMAP sea surface salinity\u003c/b\u003e:\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eVersion 4.0 of the Level-3 SMAP SSS product from NASA\u0026rsquo;s SMAP mission, which is an 8-day running mean gridded dataset (Meissner et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2019\u003c/span\u003e), has been used in this study. This product is distributed via NASA\u0026rsquo;s Physical Oceanography Distributed Active Archive Center (PO.DAAC). It is accessible from the corresponding dataset page on the PO.DAAC portal (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://search.earthdata.nasa.gov/\u003c/span\u003e\u003cspan address=\"https://search.earthdata.nasa.gov/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). The data are on a 25 km equal-area grid, and the effective spatial resolution of SMAP SSS is on the order of ~\u0026thinsp;70 km for feature detection.\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003e\u003cb\u003eGeostrophic currents from Altimetry\u003c/b\u003e:\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eSurface ocean currents were obtained from multi-mission satellite altimeter data, distributed as a gridded absolute dynamic topography by the Copernicus Marine Environment Monitoring Service (CMEMS). We use the daily, global 1/4\u0026deg; (approximately 25 km) resolution geostrophic current product (a Level-4 merged altimetry product) from CMEMS (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://data.marine.copernicus.eu/\u003c/span\u003e\u003cspan address=\"https://data.marine.copernicus.eu/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). These data provide the eastward (u) and northward (v) components of surface velocity, derived from the gradients of sea surface height under geostrophic approximation.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2 Validation data set\u003c/h2\u003e\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003e\u003cb\u003eHigh-Resolution SSS for Gulf Stream\u003c/b\u003e:\u003c/p\u003e\u003cp\u003eTo validate the algorithm, we use a high-resolution SSS product for the Gulf Stream region produced by Barcel\u0026oacute; et al., 2021. In their study, a Lagrangian advection technique was applied to SMAP SSS and altimetry currents to reconstruct SSS at finer scale, using a 7-day backward advection. The resulting high-resolution SSS product covers the Gulf Stream region for the years 2017, 2018, and 2019. This dataset serves as an independent benchmark; since it was produced with a similar methodology by an independent group, we can directly compare reconstructed fields against theirs to assess the performance of our in-house algorithm.\u003c/p\u003e\u003cp\u003e\u003cb\u003eThermo-Salinograph in situ SSS\u003c/b\u003e:\u003c/p\u003e\u003cp\u003eWe also use in situ SSS measurements from ship-based thermo-salinographs (TSG) for validation in the Bay of Bengal. Thermo-salinographs continuously measure sea surface salinity (and temperature) from water pumped through a ship\u0026rsquo;s intake, providing high temporal and spatial resolution data along the ship\u0026rsquo;s track with high accuracy. We obtained the quality-controlled TSG data from ship cruises conducted in the Bay of Bengal during September, 2015 (under the Ocean Monsoon Mixing (OMM) program (Wijesekera et. al. \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2016\u003c/span\u003e) and May, 2024 (Ekamsat cruise).\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e2.3 Methodology\u003c/h2\u003e\u003cdiv id=\"Sec6\" class=\"Section3\"\u003e\u003ch2\u003e2.3.1 Passive Lagrangian Reconstruction Technique\u003c/h2\u003e\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eIn this approach, sea surface salinity is treated as a passive tracer transported solely by horizontal advection of surface currents, with no local sources or sinks and neglecting mixing over short timescales. Based on this assumption, finer-scale salinity fields can be reconstructed by advecting observed surface salinity field with surface currents. Two approaches are commonly employed in such Lagrangian reconstructions: (i) backward advection (Barcel\u0026oacute; et al., 2021) and (ii) forward advection (Dencausse et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2014\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eIn one-way backward advection, we begin by seeding a dense grid of virtual particles at the target time T. We then integrate each particle\u0026rsquo;s trajectory backward over the chosen advection window T to \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:T-{\\Delta\\:}T\\)\u003c/span\u003e\u003c/span\u003e. At that origin point, we sample salinity from the coarse SMAP field and carry that value forward with the particle to time T. Because the particles trace out fine‐scale deformations by the altimeter‐derived flow, this process effectively interpolates the 25 km SMAP salinity onto a 4 km grid, yielding a high‐resolution salinity map at T without ever \u0026ldquo;inventing\u0026rdquo; new tracer\u0026ndash;each particle simply brings its assigned SMAP salinity forward along its trajectory. The \u003cb\u003eforward advection\u003c/b\u003e approach, conversely, would start at \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{0}\\)\u003c/span\u003e\u003c/span\u003e and advect particles forward to \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{f}\\)\u003c/span\u003e\u003c/span\u003e, carrying the salinity from \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{0}\\)\u003c/span\u003e\u003c/span\u003e to later times. In this study, we primarily use backward advection for passive advection, but forward advection in active\u0026ndash;advection case.\u003c/p\u003e\u003cp\u003eThe equations of motion for the passive advection of fluid particles on the sphere (longitude\u0026ndash;latitude coordinates) under the geostrophic approximation are given by:\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Equ1\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ1\" name=\"EquationSource\"\u003e\n$$\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\frac{\\partial\\:\\varphi\\:\\left(\\varphi\\:,\\:\\theta\\:,\\:t\\right)}{\\partial\\:t}\\:\\:=\\frac{u}{{R}_{e}cos\\left(\\theta\\:\\right)}$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e1\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Equ2\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ2\" name=\"EquationSource\"\u003e\n$$\\:\\frac{\\partial\\:\\theta\\:\\left(\\varphi\\:,\\:\\theta\\:,\\:t\\right)}{\\partial\\:t}\\:\\:=\\frac{v}{{R}_{e}}$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e2\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\upvarphi\\:}\\:\\)\u003c/span\u003e\u003c/span\u003e is longitude, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\theta\\:\\)\u003c/span\u003e\u003c/span\u003e is latitude, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:R\\)\u003c/span\u003e\u003c/span\u003e is the mean radius of the Earth, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:u(\\varphi\\:,\\theta\\:,t)\\)\u003c/span\u003e\u003c/span\u003e and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:v(\\varphi\\:,\\theta\\:,t)\\)\u003c/span\u003e\u003c/span\u003e are the eastward and northward components of the surface velocity field, respectively. We integrate these equations using a fourth\u0026ndash;order Runge\u0026ndash;Kutta numerical scheme with a time step of \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\Delta\\:}t\\)\u003c/span\u003e\u003c/span\u003e = 0.1 day (approximately 2.4 hours). Bilinear interpolation in space and linear interpolation in time are used to estimate the velocity field at the particle\u0026rsquo;s position and time during each step.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section3\"\u003e\u003ch2\u003e2.3.2 Active Advection and Algorithm Modification\u003c/h2\u003e\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eIn the standard passive scheme, salinity remains constant for each particle throughout the integration, fixed at the initial SMAP value, which might result in large deviations. In order to take into account this problem, the above method was modified such that the intermediate reconstructed salinity field is periodically compared with the concurrent SMAP satellite observations. Whenever deviations exceed a predefined threshold (1 psu in this study), particle salinity values are reset to the observed SMAP values. In essence, this approach limits the deviation of the reconstructed field from the satellite-observed field, preventing the algorithm from drifting too far due to neglected processes. This active advection method effectively uses a variable advection time for different water parcels: portions of the field that start to diverge significantly are re-initialized (a shorter advection), whereas other areas continue to be advected for the full period. By doing so, the algorithm can maintain more realistic salinity levels, especially in the presence of strong freshwater fluxes, at the cost of potentially smoothing out some advection-driven small scales. We applied this modified algorithm (hereafter referred to as active advection (AA)) in the Bay of Bengal to assess whether it improves the accuracy of the high-resolution SSS reconstruction relative to the original fixed-time algorithm. The chosen threshold, crucial for balancing the preservation of small-scale features against limiting error growth, directly influences the method's effectiveness. Flowchart for passive as well as active advection case is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"3. Results and discussion","content":"\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eHaving implemented both the passive (HRSSS-PA) and active (HRSSS-AA) reconstruction frameworks, we now present a tiered assessment of their skill and robustness. Before applying the developed technique using passive advection in the Bay of Bengal, the algorithm is verified by comparing the outputs with the high-resolution Gulf Stream product of Barcel\u0026oacute; et al., 2021,\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003e3.1 Validation with an Existing Product (Gulf Stream)\u003c/b\u003e\u003c/p\u003e\u003cp\u003eFigure 2. Reconstructed sea surface salinity fields using passive advection for (a) 7 February 2017 and (b) 14 March 2017. Panels show results from (i) Barcel\u0026oacute; et al., 2021, (ii) the proposed method, and (iii) the original SMAP salinity observations.\u003c/p\u003e\u003cp\u003eBarcel\u0026oacute; et al., 2021 used a 7-day backward passive advection (Lagrangian technique) of SMAP data to reconstruct SSS in the Gulf Stream region. Several example periods were simulated using developed methodology for the Gulf Stream region and the high resolution salinity outputs were compared with the Barcel\u0026oacute; et al., 2021 product. Figure\u0026nbsp;2 compare these two datasets on two different dates in 2017: 7 February, and 14 March. The figures show a strong correspondence between our high-resolution SSS fields and those obtained by Barcel\u0026oacute; et al., 2021, both qualitatively in terms of spatial patterns and quantitatively in terms of range of salinity values. In particular, small-scale frontal features in the salinity field (for example, those highlighted by the black ellipse in Fig.\u0026nbsp;2) are reconstructed in almost the same manner by both methods, indicating that present implementation of the Lagrangian advection is reliable and correctly captures the salinity gradients. Furthermore, the characteristics formed as a result of passive advection within the black elliptical box strongly mirror the products produced by Barcel\u0026oacute; et al., 2021. The Root Mean Square Error (RMSE) values between the reconstructed field from the in-house developed algorithm and the reference field, evaluated at weekly intervals, are presented in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The RMSE values ranged from a minimum of 0.138 psu on 28 February 2017, indicating the closest match, to a maximum of 0.181 psu on 14th March 2017, representing the largest deviation. Intermediate RMSE values were observed on 7 February (0.146 psu), 14 February (0.151 psu), 21 February (0.169 psu), and 7 March (0.162 psu), demonstrating consistent accuracy across the evaluated period. These low RMSE values, coupled with correlation coefficients consistently above 0.99, highlight the high reliability and excellent agreement of the in-house algorithm with established methodologies. Thereafter, this algorithm has been used to generate high resolution SSS for considered domain of interest, which is Bay of Bengal.\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\u003eCorrelation between reconstructed field by in-house developed algorithm and by Barcel\u0026oacute; et al., 2021 for integration time of 7 days using backward passive advection in Gulf Stream.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDate\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eCorrelation\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eRoot Mean Square Error (in psu)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e7 Feb, 2017\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.992\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.146\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e14 Feb, 2017\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.990\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.151\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e21 Feb, 2017\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.991\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.169\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e28 Feb, 2017\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.992\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.138\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e7 Mar, 2017\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.991\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.162\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e14 Mar, 2017\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.991\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.181\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\u003ch2\u003e3.2 Sensitivity study of HRSSS product with respect to spatial resolution of reconstructed product\u003c/h2\u003e\u003cp\u003eTo assess the impact of spatial resolution on reconstruction accuracy, a sensitivity analysis was conducted by generating HRSSS products at four different spatial scales\u0026mdash;25 km, 10 km, 4 km, and 2 km\u0026mdash;through interpolation of the velocity fields. These HRSSS products, along with SMAP sea surface salinity, were compared against in situ data collected along a ship track on various days in 2024, as illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e3\u003c/span\u003e. The root mean square error (RMSE) was calculated between each HRSSS product and the in situ data, as well as between SMAP salinity and the in situ data. For each validation day, the TSG provided approximately 86,400 along-track measurements, enabling a robust point-by-point comparison with the HRSSS reconstructions.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFrom Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e3\u003c/span\u003e, it is evident that the sea surface salinity product generated using velocity fields at a spatial resolution of 25 km shows the lowest accuracy compared to in situ data across all reconstruction cases. As the spatial resolution is refined from 25 km to 10 km, and subsequently to 4 km, there is a clear and consistent improvement in the accuracy of the HRSSS product. The reduction in root mean square error (RMSE) with increasing resolution demonstrates that finer resolution allows for a more precise capture of the underlying physical processes influencing sea surface salinity. However, when the resolution is further increased from 4 km to 2 km, the improvement in accuracy becomes negligible. The lack of significant reduction in RMSE between the 4 km and 2 km resolutions suggests that the benefits of increasing resolution may plateau at some point, beyond which additional refinement does not yield proportionate gains in accuracy. This finding is critical for determining the optimal resolution for HRSSS product generation, balancing computational resources with the need for accurate data representation. Furthermore, in most cases, the RMSE between SMAP-derived sea surface salinity and in situ observations is higher than that of the reconstructed high-resolution salinity fields, highlighting better performance of the reconstruction approach.\u003c/p\u003e\u003cp\u003eAn important limitation of the current reconstruction method is its reliance on CMEMS geostrophic currents at a quarter-degree (~\u0026thinsp;25 km) resolution. This coarse velocity field constrains the smallest flow-driven variability that can be resolved, meaning that while present algorithm produces salinity fields on a 4 km grid, the evolution of sub-mesoscale features is still governed by mesoscale currents, which miss variability below ~\u0026thinsp;20\u0026ndash;25 km. Consequently, the sharp gradients in reconstruction largely arise from the stretching of salinity fronts by mesoscale flows rather than explicitly resolved sub-mesoscale currents, explaining why accuracy gains diminish beyond 4 km resolution. Future improvements may come from incorporating higher-resolution currents, for instance from numerical models or upcoming missions such as SWOT, which is expected to provide ocean currents at ~\u0026thinsp;10 km scales, thereby enhancing the fidelity of reconstructed salinity fields.\u003c/p\u003e\u003cp\u003e\u003cb\u003e3.3 Selection of optimal advection time for reconstruction of high resolution sea surface salinity in the Bay of Bengal\u003c/b\u003e\u003c/p\u003e\u003cp\u003ePrevious applications of the Lagrangian advection technique in different regions have revealed that the optimal integration period varies depending on local ocean dynamics, underscoring the need to determine a region-specific value for the Bay of Bengal. For instance, in the Gulf Stream, a 7-day advection period was found to produce the best results (Barcel\u0026oacute; et al., 2021), whereas south of Tasmania an optimal period of roughly 14 days was reported (Dencausse et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). These variations suggest that a universal advection time cannot be assumed and must be tailored to each region's physical processes. The primary objective of reconstructing HRSSS product is to obtain tracer fields that resolve sub-mesoscale structures while remaining consistent with in situ observations. An advection period that is too short would not allow sufficient lateral stirring to develop small-scale salinity fronts, whereas an excessively long period could introduce errors as processes neglected by pure advection (e.g. air-sea fluxes, vertical and horizontal mixing, frontogenesis) begin to significantly modify the salinity field. In the latter case, sub-mesoscale features might become over-emphasized and large-scale patterns could drift away from the observed state (Dencausse et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). With these considerations in mind, a series of tests using integration times of 7, 14, 21, and 28 days were conducted to identify the optimal advection time for high-resolution sea surface salinity (HRSSS) reconstruction in the Bay of Bengal for passive advection of the tracer field.\u003c/p\u003e\u003cp\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003e presents the reconstructed high-resolution SSS fields obtained using each of these advection periods for 24 August 2015, with panels (i)\u0026ndash;(iv) showing the 7-, 14-, 21-, and 28-day results respectively, and panel (v) showing the original SMAP SSS field for comparison. All fields in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003e are overlaid with coincident TSG measurements collected along a ship track, providing\u003c/p\u003e\u003cp\u003ea reference for real-world salinity gradients. This visual comparison reveals that a short advection period (7 days) fails to fully develop the finer-scale fronts, as the small-scale salinity variations along the ship track are not well represented. In contrast, an overly long integration (28 days) yields salinity fronts that are displaced spatially relative to the in situ data, indicating a bias in the\u003c/p\u003e\u003cp\u003eposition of features. The intermediate durations show a more balanced outcome: notably, the 14-day advection produces fronts that align closely with the TSG observations, suggesting that this period allows sufficient development of sub-mesoscale features without incurring large errors.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eBesides this, SMAP sea surface salinity shows a very weak front formation. In summary, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003e (panels i\u0026ndash;iv) indicates that an advection time of approximately 14 days captures the salinity front development most faithfully in this region, whereas shorter or longer periods either under-develop or misplace these fronts.\u003c/p\u003e\u003cp\u003eFigure 5. Sea surface salinity generated from HRSSS (Red Line), SMAP (Orange Line) and TSG (blue line) along ship track on 24th August 2015.\u003c/p\u003e\u003cp\u003eA quantitative comparison with in situ data further supports the choice of a 14-day advection period. Figure\u0026nbsp;5 compares the along-track salinity profiles from the reconstructed 14-day HRSSS product (red line) and the original SMAP product (orange line) against the shipboard TSG measurements (blue line) on 24 August 2015. The 14-day HRSSS reconstruction shows excellent agreement with the in situ data, closely tracing the observed salinity variations along the ship track. In contrast, there is very less variation in the SMAP salinity field and it misses some of the finer-scale fluctuations captured by the HRSSS-PA. This is reflected in the root-mean-square (RMS) differences: the HRSSS-PA product\u0026rsquo;s error relative (RMSE: 0.19) to the in situ measurements is substantially lower than that of the SMAP product (RMSE: 0.47). In effect, the reconstructed high-resolution salinity field not only matches the ship-based observations more closely, but also reproduces the variability of sea surface salinity with higher fidelity. These results confirm that an advection period on the order of two weeks is the optimal choice for the Bay of Bengal, as it yields a high-resolution salinity field that best reconciles the need for sub-mesoscale detail with consistency to direct observations.\u003c/p\u003e\u003cp\u003e\u003cb\u003e3.4 Evaluation of High-Resolution Sea Surface Salinity Using Along-Track TSG Data from EKAMSAT 2024 under Passive Advection Conditions\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThis section presents a detailed assessment of the reconstructed high-resolution sea surface salinity (HRSSS) fields derived from passive advection techniques using in situ salinity measurements collected during the EKAMSAT research cruise during May 2024.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFigure \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e6\u003c/span\u003e displays a three-panel visualization of sea surface salinity (SSS) along the ship track. Panel (i) shows the in situ SSS obtained from the EKAMSAT TSG, serving as the reference for evaluating satellite products. Panel (ii) illustrates the coinciding SMAP Level 3 gridded salinity, a widely used satellite product. Panel (iii) presents the HRSSS-PA estimates reconstructed using passive advection. Qualitatively, HRSSS-PA more closely follows the finer-scale structures and frontal gradients observed in the in situ data, whereas SMAP appears spatially smoother and less responsive to small-scale salinity variability. Quantitative evaluation using root mean square error (RMSE) and Pearson correlation coefficient (R) further supports these visual observations. Compared to in situ data, SMAP yields an RMSE of 2.76 psu with a correlation of 0.49, suggesting moderate agreement but limited accuracy in reproducing spatial detail. In contrast, HRSSS-PA demonstrates improved performance, achieving an RMSE of 1.97 psu and a stronger correlation of 0.60. These results suggest that the HRSSS-PA product offers a more reliable representation of SSS in coastal and dynamically active regions.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003e\u003cb\u003eFigure 7.\u003c/b\u003e Daily along-track profiles of sea surface salinity (SSS) retrieved from HRSSS-PA, SMAP, and in situ observations for four representative periods in April-May 2024. Each panel corresponds to a specific date range: (i) 28 \u0026amp; 29 April 2024, (ii) 30 April 2024, (iii) 3 \u0026amp; 4 May 2024, and (iv) 9 \u0026amp; 10 May 2024. Salinity measurements are plotted as a function of longitude along the ship track.\u003c/p\u003e\u003cp\u003eFigure 7 extends this comparison to four selected periods during April-May 2024 to assess temporal variability in product performance. For 28\u0026ndash;29 April 2024 [panel (i)], HRSSS-PA achieves an RMSE of 2.48 and R\u0026thinsp;=\u0026thinsp;0.52, compared to SMAP's RMSE of 3.52 and a weak negative correlation (R = \u0026minus;\u0026thinsp;0.11), confirming HRSSS-PA's stronger consistency in both magnitude and structure. On 30 April 2024 [panel (ii)], HRSSS-PA yields an RMSE of 3.19 (R\u0026thinsp;=\u0026thinsp;0.44) while SMAP records a higher RMSE of 4.78 (R\u0026thinsp;=\u0026thinsp;0.32), indicating better performance by HRSSS-PA despite challenging conditions. For 3\u0026ndash;4 May 2024 [panel (iii)], HRSSS-PA and SMAP both exhibit lower RMSEs (2.08 and 2.43, respectively), though correlation values remain weak (R\u0026thinsp;=\u0026thinsp;0.12 and 0.05), suggesting limitations in capturing finer-scale salinity variability. The best agreement is observed during 9\u0026ndash;10 May 2024 [panel (iv)], where HRSSS-PA shows an RMSE of 0.89 and a strong correlation of 0.87, while SMAP performs comparably well with an RMSE of 1.36 and a higher correlation of 0.91.\u003c/p\u003e\u003cp\u003eBecause the SMAP Level-3 product is an 8-day running mean, it smooths day-to-day variability; relative to instantaneous ship measurements, this temporal averaging can lower correlations and inflate apparent errors during rapidly evolving frontal events (Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e6\u003c/span\u003e\u0026ndash;7). Collectively, these results highlight the superior accuracy and effective resolution of HRSSS-PA during relatively stable periods, while performance degrades for both products under more dynamic conditions\u0026mdash;underscoring the continuing need for high-resolution in situ observations to validate and enhance satellite SSS reconstruction algorithms. In brief, the baseline passive reconstruction (HRSSS-PA) sharpens mesoscale fronts by treating SSS as a strictly conserved tracer over the integration window and advecting SMAP with altimetry-derived currents, yielding high-resolution SSS for the Bay of Bengal.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003e3.5 Evaluation of High-Resolution Sea Surface Salinity Using Along-Track TSG Data from R/V Roger-Revelle 2015 under Passive Advection Conditions\u003c/b\u003e\u003c/p\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eTo further evaluate the generalizability of the HRSSS-PA reconstruction method, an independent validation was performed using historical salinity observations collected aboard the R/V Roger Revelle in August\u0026ndash;September 2015. This complementary analysis provides a contrasting testbed featuring a different oceanographic regime and time period, thereby allowing assessment of the HRSSS-PA product's robustness under varying conditions. Figure\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e8\u003c/span\u003e presents a side-by-side comparison of sea surface salinity (SSS) profiles retrieved from HRSSS-PA, SMAP, and in situ observations along the ship track on four selected days during the cruise. In panel (i), for 24 August 2015, the HRSSS-PA reconstruction aligns closely with the in situ salinity profile throughout the observed track. It effectively captures both the general trend and sharper gradients, particularly between longitudes 84.1\u0026deg; and 84.4\u0026deg;, where a distinct salinity drop occurs. The SMAP estimate, in contrast, displays a notably smoother gradient and underestimates the sharp salinity drop, indicating loss of spatial detail. This visual agreement with in situ data corresponds well with the reported low RMSE (0.19) and high correlation (R\u0026thinsp;=\u0026thinsp;0.93) for HRSSS-PA. Although SMAP achieves a slightly higher correlation (R\u0026thinsp;=\u0026thinsp;0.96), its RMSE is more than double that of HRSSS-PA (0.47), emphasizing its reduced accuracy in magnitude.\u003c/p\u003e\u003cp\u003eIn panel (ii), corresponding to 26 August 2015, HRSSS-PA again tracks the in situ salinity profile more closely, particularly in the central portion of the transect (84.2\u0026deg;\u0026ndash;84.5\u0026deg;), where small-scale fluctuations are better represented than in the SMAP product. While SMAP maintains a consistently smooth profile, it fails to capture the local salinity troughs and peaks observed in in situ. Numerically, HRSSS-PA exhibits a slightly higher RMSE (0.248) than SMAP (0.213), yet the correlation coefficient is markedly higher for HRSSS-PA (R\u0026thinsp;=\u0026thinsp;0.53 vs. 0.16 for SMAP), reflecting its superior representation of spatial variability.\u003c/p\u003e\u003cp\u003eA different behavior emerges in panel (iii), corresponding to 5 September 2015, during the peak of freshwater discharge from the Ganga-Brahmaputra river system. The in situ profile shows sharp, step-like salinity transitions between ~\u0026thinsp;29.4 and 30.5 psu, likely associated with freshwater lenses and dynamic fronts. HRSSS-PA, which relies on passive advection, significantly overestimates salinity throughout the transect and fails to resolve the sharp gradients or structural variability. In contrast, SMAP-despite its coarse resolution-better captures the pattern and magnitude of the observed transitions, albeit with some smoothing. Statistically, HRSSS-PA records a high RMSE of 1.74 psu (R\u0026thinsp;=\u0026thinsp;0.69), while SMAP achieves a lower RMSE of 0.65 psu and higher correlation (R\u0026thinsp;=\u0026thinsp;0.79), indicating relatively improved performance under these more complex conditions.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003ePanel (iv), dated 19 September 2015, shows both HRSSS-PA and SMAP generally follow the trend of the in situ salinity field, with HRSSS-PA offering closer agreement, especially across longitudes 84.8\u0026deg; to 85.6\u0026deg;. Minor discrepancies remain in capturing localized dips and peaks. HRSSS-PA achieves an RMSE of 0.2 psu and a moderate correlation of R\u0026thinsp;=\u0026thinsp;0.44, while SMAP underrepresents the observed variability, with a higher RMSE (0.326) and a weak negative correlation (R = -0.192), indicating reduced reliability in this case.\u003c/p\u003e\u003cp\u003eCollectively, Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e8\u003c/span\u003e demonstrates that HRSSS-PA generally outperforms SMAP in reproducing both absolute salinity values and spatial gradients, particularly under stable or moderately variable conditions (e.g., 24 and 26 August). The high-resolution SSS fields generated here reveal much finer spatial detail in the Bay of Bengal, underscoring the utility of a Lagrangian advection framework for bridging the gap between coarse satellite products and the finer-scale processes of interest. This approach offers a promising means of exploiting existing satellite observations to improve monitoring of freshwater dynamics and mesoscale variability in the Bay of Bengal, with potential applications for similar downscaling strategies in other oceanic regions.\u003c/p\u003e\u003cp\u003eHowever, its limitations become evident under highly dynamic scenarios such as 5 September, where passive advection fails to resolve the fine-scale features shaped by active frontal displacement and freshwater input. Passive Lagrangian reconstruction assumes surface salinity is strictly conserved over the 14-day integration, neglecting freshwater fluxes, rainfall, evaporation, and mixing. In the Bay of Bengal, these processes are often comparable to horizontal advection, causing rapid divergence from reality. Such episodic fluxes and unresolved mixing set a hard limit on passive reconstructions; active scheme mitigates this by resetting particles when errors grow, but advection alone cannot capture the full salt budget under strong freshwater forcing.\u003c/p\u003e\u003cp\u003e\u003cb\u003e3.6 Evaluation of High-Resolution Sea Surface Salinity Using Along-Track TSG Data from Roger-Revelle 2015 under active advection conditions\u003c/b\u003e\u003c/p\u003e\u003cp\u003eContinuing from the previous analysis, Figs.\u0026nbsp;7 and \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e8\u003c/span\u003e demonstrate that the HRSSS-PA product aligns closely with in situ salinity measurements across several observational days, and typically outperforms SMAP-derived salinity fields. This consistency highlights the dominant role of passive advection during certain periods. However, notable deviations occur, particularly on 5 September 2015 (Fig.\u0026nbsp;8iii), when HRSSS-PA shows larger discrepancies compared to SMAP. These mismatches point to the limitations of a purely passive advection scheme, which omits additional oceanographic dynamics influencing salinity.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eTo address these observed discrepancies and improve the representation of oceanographic processes, an alternative reconstruction that explicitly incorporates active advection was implemented. In this section, we evaluate the effectiveness and accuracy of this active advection-based reconstruction by comparing its performance against both in situ salinity data collected along a research vessel transect in the northern Bay of Bengal and SMAP salinity observations for specific period (5 Sep, 2015), as depicted in Fig.\u0026nbsp;9. This region is heavily influenced by the seasonal freshwater discharge from the Ganga-Brahmaputra river system and is known for its sharp salinity gradients and complex surface circulation patterns.\u003c/p\u003e\u003cp\u003eFigure 9. Daily along-track profiles of sea surface salinity (SSS) retrieved from HRSSS-AA, SMAP, and in situ observations plotted as a function of longitude along the ship track.\u003c/p\u003e\u003cp\u003eAs observed in the figure, the in situ salinity profile exhibits strong spatial variability, with values ranging from approximately 29.4 to 30.5 psu, reflecting the influence of riverine freshwater lenses and frontal boundaries. SMAP retrievals (cyan dashed line), limited by a coarse spatial footprint (~\u0026thinsp;40 km), capture the broad gradients but consistently overestimate salinity across much of the transect, with a bias of approximately 0.2\u0026ndash;0.8 psu. In contrast, the HRSSS-AA retrieval (red dotted line) shows a marked improvement over HRSSS-PA in resolving the structure and magnitude of these variations. On the same day (see Section 3.5), HRSSS-PA exhibited a relatively high RMSE of 1.74 psu and a moderate correlation (R\u0026thinsp;=\u0026thinsp;0.69), underscoring its limitations under dynamic conditions. The transition to active advection in HRSSS-AA effectively overcomes these issues by treating salinity as a field actively influenced by horizontal flow and sub-mesoscale variability. Quantitatively, HRSSS-AA achieves a reduced RMSE of 0.79 psu, compared to 0.65 psu for SMAP, highlighting the value of dynamic modeling even when absolute errors may appear similar. This emphasizes the value of active dynamics in maintaining realistic salinity patterns, even when absolute error may appear slightly elevated.\u003c/p\u003e\u003cp\u003eBeyond these metrics, the active-advection approach enhances the physical realism of reconstructed fields. For instance, along-track SSS on 5 September 2015 reveals sharp, step-like transitions and filaments near 89.56\u0026ndash;89.60\u0026deg;E (Fig.\u0026nbsp;9), characteristic of buoyant freshwater lenses and monsoonal frontal boundaries driven by Ganga-Brahmaputra-Meghna discharge. The interaction of this plume with mesoscale eddies and sub-mesoscale instabilities sharpens these gradients into filaments and narrow fronts (Boccaletti et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2007\u003c/span\u003e; Mahadevan, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; McWilliams, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2016\u003c/span\u003e), as commonly observed in the northern Bay of Bengal during the post-monsoon season (Akhil et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Ramachandran et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Sengupta et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). The ability of HRSSS-AA to resolve such features demonstrates not only reduced statistical error but also improved representation of physically meaningful processes, plume spreading, frontal sharpening, and eddy-driven filamentation, that are critical for understanding salinity variability, stratification, and air-sea fluxes in river-influenced coastal systems.\u003c/p\u003e\u003c/div\u003e"},{"header":"4. Summary and Conclusions","content":"\u003cp\u003eThis study addresses the challenge of reconstructing high-resolution sea surface salinity (SSS) fields from coarse satellite data in the Bay of Bengal, a region shaped by complex circulation, strong monsoonal rainfall, and large freshwater discharge. Using a Lagrangian framework based on passive and active advection, we reconstructed fine-scale salinity structures from SMAP Level-3 SSS and satellite-derived geostrophic currents. This approach provides a new window into upper-ocean processes in regions with strong rainfall and river discharge, offering improvements over conventional satellite products\u003c/p\u003e\u003cp\u003eThe passive advection method was applied to generate high-resolution salinity fields under stable to moderately dynamic oceanographic conditions. To assess the correctness of our implementation, we compared our results against a reference high-resolution salinity product produced using the same underlying technique by Barcel\u0026oacute; et al. (2021) over the Gulf Stream. The two independently generated products exhibited strong agreement, with root mean square errors (RMSE) between 0.14\u0026ndash;0.18 psu and correlation coefficients consistently above 0.99. Sensitivity analyses indicated an optimal spatial resolution of approximately 4 km, which captured finer oceanographic features effectively, whereas further refinement to 2 km offered minimal improvement. An integration window of 14 days was identified as optimal, balancing detailed sub-mesoscale structure development against numerical drift risks in Bay of Bengal. Validation using in situ thermo-salinograph data from the EKAMSAT-2024 cruise confirmed the improved performance of the passive advection approach. Compared to SMAP, the HRSSS-PA product showed significantly better agreement with in-situ measurements, reducing the cruise-mean RMSE from 2.76 psu (SMAP) to 1.97 psu (HRSSS-PA) and improving the correlation coefficient from 0.49 (SMAP) to 0.60 (HRSSS-PA). Detailed assessments on specific dates highlighted consistent performance, notably an RMSE of 0.89 psu (R\u0026thinsp;=\u0026thinsp;0.87) during stable conditions. Validation against thermo-salinograph data from the Roger Revelle cruise (24\u0026ndash;26 August 2015) showed similar successes, significantly reducing RMSE from 0.466 psu (SMAP) to as low as 0.186 psu (HRSSS-PA). However, during highly dynamic events with intense freshwater input, such as on 5 September 2015, passive advection alone proved inadequate (RMSE\u0026thinsp;=\u0026thinsp;1.74 psu). Incorporating active advection in such cases significantly improved performance (RMSE\u0026thinsp;=\u0026thinsp;0.79 psu), enabling sharper frontal resolution and dynamic response. Collectively, these results demonstrate that passive advection is well-suited for relatively stable regimes, while active advection is necessary to capture the complexity of freshwater-driven variability. The combined methodology offers a practical and accurate approach for high-resolution SSS reconstruction in challenging oceanographic settings.\u003c/p\u003e\u003cp\u003eLagrangian advection has proven effective for high-resolution SSS mapping, but its limitations must be acknowledged. The approach is sensitive to errors in the SMAP satellite salinity inputs, particularly in dynamic, freshwater-influenced regions like the Bay of Bengal, where strong salinity gradients and heavy rainfall can exacerbate retrieval uncertainties. Moreover, the passive advection assumption breaks down under intense freshwater forcing events (e.g. monsoonal rain or major river discharge), since new low-salinity inputs and mixing processes are not represented. This has led to notable discrepancies between reconstructed salinity fields and in situ thermo-salinograph measurements during such events. Nonetheless, future high-resolution SSS missions offer reason for optimism. Upcoming sensors like the Copernicus Imaging Microwave Radiometer (CIMR) will provide more precise, finer-scale salinity observations, which should help reduce input errors and significantly improve the accuracy of SSS reconstructions.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eFunding Declaration\u003c/h2\u003e\u003cp\u003eThis research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eJai Kumar: Conceptualization, Methodology, Software, Formal Analysis, Investigation, Data Curation, Visualization, Writing \u0026ndash; Original Draft, Writing \u0026ndash; Review \u0026amp; EditingNeeraj Agarwal: Conceptualization, Writing \u0026ndash; Review \u0026amp; Editing, SupervisionRashmi Sharma: Conceptualization, Writing \u0026ndash; Review \u0026amp; Editing, Supervision, Project Administration\u003c/p\u003e\u003ch2\u003eAcknowledgments\u003c/h2\u003e\u003cp\u003eThe authors would like to express their sincere gratitude to the Director, Space Applications Centre, for motivation. Velocity field data and SMAP sea surface salinity used in the work are taken from \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://resources.marine.copernicus.eu/\u003c/span\u003e\u003cspan address=\"https://resources.marine.copernicus.eu/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e and \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://search.earthdata.nasa.gov/\u003c/span\u003e\u003cspan address=\"https://search.earthdata.nasa.gov/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e, respectively.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eGenerated Data can be downloaded from the link below and it is free of use-https://www.mosdac.gov.in/high-resolution-sea-surface-salinity\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAkhil VP, Durand F, Lengaigne M, Vialard J, Keerthi MG, Gopalakrishna VV, Deltel C, Papa F, de Boyer Mont\u0026eacute;gut C (2014) A modeling study of the processes of surface salinity seasonal cycle in the Bay of Bengal. J Geophys Research: Oceans 119(6):3926\u0026ndash;3947\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBarcel\u0026oacute;-Llull B, Drushka K, Gaube P (2021) Lagrangian reconstruction to extract small-scale salinity variability from SMAP observations. J Geophys Research: Oceans, 126, e2020JC016477.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBerti S, Lapeyre G (2014) Lagrangian reconstructions of temperature and velocity in a model of surface ocean turbulence. Ocean Model 76:59\u0026ndash;71\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBoccaletti G, Ferrari R, Fox-Kemper B (2007) Mixed layer instabilities and restratification. J Phys Oceanogr 37(9):2228\u0026ndash;2250\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCherian DA, Shroyer EL, Wijesekera HW, Moum JN (2020) The seasonal cycle of upper-ocean mixing at 8\u0026deg; N in the Bay of Bengal. J Phys Oceanogr 50(2):323\u0026ndash;342\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChin TM, Vazquez-Cuervo J, Armstrong EM (2017) A multi-scale high-resolution analysis of global sea surface temperature. Remote Sens Environ 200:154\u0026ndash;169\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDencausse G, Morrow R, Rog\u0026eacute; M, Fleury S (2014) Lateral stirring of large-scale tracer fields by altimetry. Ocean Dyn 64(1):61\u0026ndash;78\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDespr\u0026egrave;s A, Reverdin G, d'Ovidio F (2011) Mechanisms and spatial variability of mesoscale frontogenesis in the northwestern subpolar gyre. Ocean Model 39(1):97\u0026ndash;113\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDrushka K, Asher WE, Sprintall J, Gille ST, Hoang C (2019) Global patterns of submesoscale surface salinity variability. J Phys Oceanogr 49(7):1669\u0026ndash;1685\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKumar J, Ratheesh S, Agarwal N, Sharma R (2023) Study of upwelling and mixing process in the Somali coastal region using satellite and numerical model observations: A Lagrangian approach. Deep Sea Research Part II, vol 216. Topical Studies in Oceanography, p 105381\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKumar J, Choudhary RK, Mathur M, Agarwal N, Sharma R (2024) A Study of Mixing and Biological Activity in the North Indian Ocean Using Finite Size Lyapunov Exponents. J Indian Soc Remote Sens 51(2):395\u0026ndash;403\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eL\u0026eacute;vy M, Klein P, Treguier AM (2001) Impact of sub-mesoscale physics on production and subduction of phytoplankton in an oligotrophic regime. J Mar Res 59(4):535\u0026ndash;565\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiang Z, Bao S, Zhang W, Wang H, Yan H, Dai J, Xiao P (2025) Spatiotemporal Super-Resolution of Satellite Sea Surface Salinity Based on a Progressive Transfer Learning-Enhanced Transformer. Remote Sens 17(15):2735\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLucas AJ, Nash JD, Pinkel R, MacKinnon JA, Tandon A, Mahadevan A, Omand MM, Freilich M, Sengupta D, Ravichandran M, Le Boyer A (2016) Adrift upon a salinity-stratified sea: A view of upper-ocean processes in the Bay of Bengal during the southwest monsoon. Oceanography 29(2):134\u0026ndash;145\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMahadevan A (2016) The impact of submesoscale physics on primary productivity of Plankton. Annual Rev Mar Sci 8:161\u0026ndash;184\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMahadevan A, Jaeger GS, Freilich M, Omand MM, Shroyer EL, Sengupta D (2016) Freshwater in the Bay of Bengal: Its fate and role in air-sea heat exchange. Oceanography 29(2):72\u0026ndash;81\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMcWilliams JC (2016) Submesoscale currents in the ocean. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 472(2189), 20160117\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMeissner T, Wentz FJ, Manaster A, Lindsley R (2019) Remote sensing systems SMAP ocean surface salinities [level 2c, level 3 running 8-day, level 3 monthly]. Remote Sensing Systems, Santa Rosa, CA, USA\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRamachandran S, Tandon A, MacKinnon J, Lucas AJ, Pinkel R, Waterhouse AF, Nash J, Shroyer E, Mahadevan A, Weller RA, Farrar JT (2018) Submesoscale processes at shallow salinity fronts in the Bay of Bengal: Observations during the winter monsoon. J Phys Oceanogr 48(3):479\u0026ndash;509\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRatheesh S, Agarwal N, Chaudhary A, Lijin J, Sree Lekha J, Mathur M, Sharma R, Kumar R (2020) Response of a high-resolution ocean circulation model to winds from different sources in simulating summer monsoon freshening in the North Bay of Bengal: A case study, Deep Sea Research Part II, vol 172. Topical Studies in Oceanography, p 104727\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eReul N, Grodsky SA, Arias M, Boutin J, Catany R, Chapron B, D\u0026rsquo;Amico F, Dinnat E, Donlon C, Fore A et al (2020) Sea Surface Salinity Estimates from Spaceborne L-Band Radiometers: An Overview of the First Decade of Observation (2010\u0026ndash;2019). Remote Sens Environ 242:111769\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRog\u0026eacute; M, Morrow RA, Dencausse G (2015) Altimetric Lagrangian advection to reconstruct Pacific Ocean fine-scale surface tracer fields. Ocean Dyn 65(9):1249\u0026ndash;1268\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRudnick DL, Ferrari R (1999) Compensation of horizontal temperature and salinity gradients in the ocean mixed layer. Science 283(5401):526\u0026ndash;529\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSengupta D, Raj GNB, Ravichandran M, Sree Lekha J (2016) Near-surface salinity and stratification in the north Bay of Bengal from moored observations. Geophys Res Lett 43(9):4448\u0026ndash;4456\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVinogradova N, Lee T, Boutin J, Drushka K, Fournier S, Sabia R, Stammer D, Bayler E, Reul N, Gordon A et al (2019) Satellite Salinity Observing System: Recent Discoveries and the Way Forward. Front Mar Sci 6:243\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang Z, Wang G, Guo X, Hu J, Dai M (2022) Reconstruction of High-Resolution Sea Surface Salinity over 2003\u0026ndash;2020 in the South China Sea Using the Machine Learning Algorithm LightGBM Model. Remote Sens 14(23):6147\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWeller RA, Farrar JT, Seo H, Prend C, Sengupta D, Sree Lekha J, Ravichandran M, Venkatesan R (2019) Moored observations of the surface meteorology and air\u0026ndash;sea fluxes in the northern Bay of Bengal in 2015. J Clim 32(2):549\u0026ndash;573\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWijesekera HW, Shroyer E, Tandon A, Ravichandran M, Sengupta D, Jinadasa SUP, Whalen CB (2016) ASIRI: An ocean\u0026ndash;atmosphere initiative for Bay of Bengal. Bull Am Meteorol Soc 97(10):1859\u0026ndash;1884\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"ocean-dynamics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"odyn","sideBox":"Learn more about [Ocean Dynamics](https://link.springer.com/journal/10236)","snPcode":"10236","submissionUrl":"https://submission.springernature.com/new-submission/10236/3","title":"Ocean Dynamics","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Sea Surface Salinity, Lagrangian Advection, High-Resolution Reconstruction, SMAP, Geostrophic Currents, Bay of Bengal","lastPublishedDoi":"10.21203/rs.3.rs-8247078/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8247078/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eIn this study, we examined salinity fields as both passive and active tracers to generate high-resolution sea surface salinity fields using a Lagrangian reconstruction technique, integrating satellite-derived data and numerical advection schemes. Specifically, we employed Version 4.0 of NASA\u0026rsquo;s Soil Moisture Active Passive (SMAP) Level-3 SSS product, which features an 8-day running mean and approximately 25 km spatial resolution. Geostrophic surface currents at comparable spatial resolution, sourced from satellite altimetry data provided by the Copernicus Marine Environment Monitoring Service (CMEMS), were utilized in this work. By applying backward and forward numerical advection schemes to the SMAP SSS fields using these altimetry-derived currents, we captured smaller-scale salinity features, enhancing spatial resolution from around 25 km down to 4 km. Utilizing salinity as a passive tracer allowed us to focus exclusively on horizontal advection without accounting for sources, sinks, or mixing. A sensitivity analysis was performed, which determined that the highest feasible resolution using this approach is 4 km, with an optimal advection integration period of 14 days Preliminary validation against ship-based thermo-salinograph observations from 2015 and 2024 demonstrates that HRSSS-PA achieves RMSEs of 0.19 to 3.09 psu with correlations of 0.12 to 0.93, generally outperforming SMAP, which shows higher errors (0.47 to 4.78 psu) and weaker or inconsistent correlations (-0.11 to 0.96). This highlights the ability of HRSSS-PA to capture both magnitude and fine-scale salinity variability. To address cases where the assumptions of passive advection break down, the framework was further extended to an active advection approach, in which salinity values were dynamically adjusted during transport to account for freshwater input and precipitation, enabling improved representation under strongly forced oceanographic conditions.\u003c/p\u003e","manuscriptTitle":"A Lagrangian-Based Technique to Generate High-Resolution Sea Surface Salinity Fields from Low-Resolution Satellite Observations: A Study in the Bay of Bengal","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-08 10:24:35","doi":"10.21203/rs.3.rs-8247078/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"142223796775450942985242920794271882740","date":"2025-12-06T19:12:49+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"155477184063066202652353411595500128822","date":"2025-12-03T15:20:42+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"172252916281264635536780047550049231047","date":"2025-12-03T15:12:48+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-12-03T13:58:13+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-12-02T15:19:00+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-12-02T05:46:37+00:00","index":"","fulltext":""},{"type":"submitted","content":"Ocean Dynamics","date":"2025-12-01T06:50:38+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"ocean-dynamics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"odyn","sideBox":"Learn more about [Ocean Dynamics](https://link.springer.com/journal/10236)","snPcode":"10236","submissionUrl":"https://submission.springernature.com/new-submission/10236/3","title":"Ocean Dynamics","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"dd1cc233-838d-4ee9-aeb9-703f0642771c","owner":[],"postedDate":"December 8th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-12-08T10:24:35+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-08 10:24:35","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8247078","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8247078","identity":"rs-8247078","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}