Automatic Crest Line Extraction Algorithm for Internal Solitary Waves Based on SWOT

Sea surface height anomaly (SSHA) observations from Surface Water and Ocean Topography (SWOT) provide a new opportunity for identifying crest lines of internal solitary waves (ISWs). However, L3 LR Unsmoothed SSHA is often affected by residual large-scale trends, rainfall contamination, and stripe noise, which limit segmentation performance. To address this issue, we propose an automatic segmentation workflow for SWOT SSHA. The workflow first applies Gaussian low-pass filtering for scale separation to extract high-frequency SSHA, then uses Otsu adaptive thresholding to segment ISW signals, and finally removes false targets using morphological geometric constraints. Validation based on 230 SWOT images from the northern South China Sea shows that, compared with the conventional method based on subtracting reanalysis fields, the proposed method increases the contrast-to-noise ratio (CNR) of high-frequency SSHA by 1.35 on average (Std = 0.99) and improves signal gain by 13.65 dB on average (Std = 7.71 dB). The method remains robust under complex conditions, including strong typhoons, severe stripe noise, weak shelf signals, and multi-wave interference. In some cases, quasi-synchronous optical imagery further confirms the authenticity of the extracted crest lines.