Alice Fassoni-Andrade (University of Brasilia (UnB), Brazil); Evlyn Márcia Leão de Moraes Novo (National Institute for space research, Brazil)

The Pantanal, the world's largest tropical wetland, is characterized by its diverse hydrological dynamics across its 11 sub-regions and its very flat topography. These sub-regions are defined by their flooding patterns, topography, soil types, and vegetation. Accurate data on water surface elevation (WSE) are crucial for managing this ecosystem, particularly given increasing threats from extreme hydrological events. However, in situ monitoring is limited. This study assesses the potential of the Surface Water and Ocean Topography (SWOT) mission by evaluating the accuracy of the Level 2 KaRIn high-rate river single-pass vector product (L2_HR_RiverSP), specifically the reach subproduct, for WSE monitoring in the Pantanal. For the assessment, we utilized SWOT-derived WSE time series (2022–2024), which were compared with data from 24 in situ stations of the Brazilian National Water Agency (ANA), covering ten sub-regions and eight rivers. To ensure data quality, both datasets underwent outlier removal using Isolation Forest and min-max normalization. Performance was evaluated using R², RMSE, MAE, and BIAS, calculated using only coincident SWOT and in situ measurements. The results revealed significant spatial variability in the performance of the SWOT L2_HR_RiverSP reach subproduct across the Pantanal. The São Lourenço River (Barão de Melgaço sub-region) showed the best agreement with in situ data (RMSE = 0.065 m, R² = 0.94, bias = 0.01 m), while the Cuiabá River (also in Barão de Melgaço) exhibited the poorest performance (RMSE = 0.61 m, R² = 0.32). At the sub-regional level, Nabileque and Paiaguás displayed the highest average correlations (R² > 0.49) and lower errors. In comparison, Aquidauana and Abrobral showed larger discrepancies (RMSE > 0.37 m). These findings highlight the importance of considering sub-regional and river-specific factors when using SWOT data in the Pantanal. The observed variability suggests that local factors, such as river morphology, vegetation cover, and complex hydrological processes, can impact the accuracy of SWOT-derived WSE. Further research should explore the potential of alternative SWOT subproducts (e.g., the node product) and on developing calibration or bias-correction strategies tailored to specific sub-regions. A more comprehensive analysis of the characteristics of these sub-regions could help explain the observed errors. This study provides a critical baseline for future SWOT data applications in the Pantanal, contributing to improved water resource management and conservation efforts in this ecologically significant region.