Assessing SWOT PIXC surface classification using imagery-based products

The Surface Water and Ocean Topography (SWOT) mission provides unprecedented spatial detail for observing inland surface waters. Yet, the behavior and reliability of surface-type classification in the Level-2 KaRIn High-Rate Pixel Cloud (L2_HR_PIXC) product remain insufficiently characterized, due to the complexity of inland waterbodies, particularly across diverse hydromorphological and climatic settings. Because PIXC classification forms the foundation of all higher-level SWOT inland water products, a systematic evaluation of its performance is essential.

In this study, we systematically evaluate the SWOT PIXC surface classification and associated water fraction estimates using multiple river case studies spanning different climate regimes defined by the Köppen–Geiger classification. For each case study, PIXC surface classes and water fraction values are grouped by classification type and analyzed against independent water occurrence information derived from  DSWx-HLS (Harmonized Landsat& Sentinel-2), DSWx-S1 (Sentinel-1), and complementary long-term water occurrence datasets from Global Surface Water (Landsat).

Beyond inter-product comparison, we investigate how discrepancies between PIXC classification and imagery-based water occurrence depend on key KaRIn observables and geometric variables, including interferometric coherence, radar backscatter (σ⁰), phase noise standard deviation, incidence angle, etc. This analysis enables a process-oriented interpretation of classification behavior across surface classes, environments, and viewing conditions. The results provide a structured assessment of the strengths and limitations of SWOT PIXC classification, supporting informed use of SWOT inland water products and contributing to ongoing processing algorithm evaluation and future refinement efforts.