.png){kind=icon}
.png){kind=icon}
Return
Developing an analysis-ready data processing framework for SWOT-derived inland surface water products
.png){kind=icon}
OA
AI
DOI：10.3389/fenvs.2026.1872674.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The Surface Water and Ocean Topography (SWOT) mission provides new opportunities for inland water monitoring by observing surface water extent and water surface elevation (WSE). However, the direct use of SWOT Level-2 HR Raster products for surface water mapping remains limited by fragmented water patterns, random noise, mixed pixels, and systematic stripe artifacts. To improve product usability, this study develops an analysis-ready data (ARD) processing framework for SWOT-derived inland surface water products, with emphasis on surface water extent and water-mask structure rather than independent WSE accuracy validation. The framework integrates wse_qual, wse_uncert, water_frac, water_area, sig0, dark_frac, and layover_impact for multi-parameter filtering and spatial structure optimization, where WSE-related variables are used as product-native quality indicators. Dongting Lake was selected as the study area, and Sentinel-1 SAR-derived water masks were used only as reference data for water-extent consistency and spatial-structure assessment. Results show that ARD processing moderately improves the water-extent agreement between SWOT-derived water masks and the Sentinel-1 reference. IoU increases from 0.2926 to 0.4081, Precision from 0.3203 to 0.5154, and F1-score from 0.4527 to 0.5796, while Recall decreases from 0.7718 to 0.6621, indicating a trade-off between reducing false positives and preserving reference water pixels. Spatial fragmentation is also substantially reduced: connected components decrease from 18,187 to 557, the small-patch area ratio drops from 5.63% to 0, and boundary density declines from 0.0048 to 0.0034. The persistence of along-track stripe artifacts represents a central constraint of the current ARD framework, indicating that geometry-related systematic errors require additional correction beyond pixel-level filtering and spatial optimization. Overall, the proposed framework improves the analysis readiness of SWOT-derived inland surface water products for water mapping and flood monitoring.
Keywords:
spatial structure,SWOT,sentinel-1,analysis-ready data,surface water mapping
Journal
IF:
3.7
Papers:
8.0K
Citations:
2.3W
View more organizations