What SWOT KaRIn can and cannot resolve over first-year pack ice: a backscatter continuum and the geometric detection of leads and ridges in Barrow Strait

The scripts process SWOT Level 2 High-Rate Pixel Cloud (L2_HR_PIXC) granules over a fixed footprint in Barrow Strait, Canadian Arctic Archipelago, across eleven cycles from December 2024 to July 2025. They apply quality control, detrend backscatter across the swath within incidence-angle bins, grid to 200 m, and then do two things: test whether the detrended backscatter separates into distinct populations, using Ashman D, the bimodality coefficient, kernel-density mode counting, a Silverman critical-bandwidth bootstrap and the Hartigan dip test; and detect linear fracture features (leads and pressure ridges) by their shape with a Frangi vesselness filter rather than by their brightness. The detections are validated against RADARSAT Constellation Mission ScanSAR imagery and against fractures digitized by hand, with block-bootstrap confidence intervals, label-permutation and random-offset null tests, a detection-threshold sweep and a co-registration tolerance check. A further script measures how the agreement between the SWOT and RCM fracture networks decays with the time difference between acquisitions. Contents: five numbered scripts that run in order (SWOT processing and continuum tests; RCM validation; manual-lead comparison; fracture-network decorrelation; study-area map), a shared module of the processing and statistical functions, a configuration file holding every path and parameter, and a conda environment file. Together they reproduce every figure and table in the paper and its Supporting Information. Written in Python, organized in cells for Spyder but runnable from the command line. The code is archived here; the SWOT and RCM data are not redistributed. SWOT L2_HR_PIXC products are distributed by NASA PO.DAAC and CNES AVISO+. RADARSAT Constellation Mission imagery is available through the Government of Canada EODMS portal under its data-access terms.