Specialized SWOT SSH Data Assimilation Improves the Representation of Multiscale Variability in the Ocean Model

Abstract The Surface Water and Ocean Topography (SWOT) mission provides wide‐swath sea surface height (SSH) observations at scales poorly sampled by conventional altimetry. Assimilating SWOT data remains challenging because part of the short‐wavelength variance may not be dynamically representable by a given model configuration. We therefore develop a two‐step multiscale 4DVAR framework to assimilate real SWOT SSH observations. The first step constrains broad‐scale SSH mismatches, whereas the second assimilates the remaining SWOT information using a shorter decorrelation scale and the first‐step analysis as background. In the northern South China Sea, the multiscale framework reduces SWOT SSH residuals, yields more organized scale‐dependent increments, and improves the representation and forecast persistence of model‐resolved multiscale SSH variability. Independent SST, Argo, and drifter observations further indicate reduced upper‐ocean thermal biases, mainly within the mixed layer. These results demonstrate that scale‐dependent assimilation can make more effective use of SWOT's multiscale information within a finite‐resolution regional model.