Aggregate data and reproducibility code for spatial-support diagnostics in LuTan-1 and Sentinel-1 comparisons with leveling

This dataset contains 57 deidentified aggregate CSV tables and Python standard-library code supporting a fixed-cohort comparison of delivered LuTan-1 and Sentinel-1 time-series products with leveling. The primary comparison retains 20 targets in five survey groups and five 2024 windows at 25, 50 and 100 m extraction radii in eastern Beijing. The tables document spatial-support diagnostics, within-group contrasts, loss decomposition, reference-energy subsets, calendar and projection sensitivity, and anonymous trajectory-deletion summaries. All adverse outcomes and existing aggregate values are retained. Ten replay runners verify released aggregate arithmetic; an additional temporary-copy test confirms that a deliberately corrupted value is rejected. The software reproduces aggregate calculations, not extraction from raw radar products, survey adjustment, calibration independence or physical object correspondence. Conventional per-target RMS is an algebraic rescaling of the contrast-dimension-weighted statistic, not independent evidence. Raw radar exports, survey workbooks, target coordinates, point-level links, source imagery, manuscripts, reviews and credentials are excluded. Group labels and anonymous count/trial identifiers are retained for reproducibility; this is a deidentified release, not a claim that the study region is anonymous. The data and documentation are licensed under CC BY 4.0; Python source code is licensed under MIT. See LICENSE.md for the file-level scope. AI assistance in coding and manuscript preparation is disclosed in AI_Tool_Scope.md. This record supports the manuscript “Spatial Support in InSAR–Leveling Validation: A Fixed-Cohort Diagnostic Framework Applied to LuTan-1 and Sentinel-1”. Funding: the National Natural Science Foundation of China (Grant No. 42476170) and Common Application Support Platform for National Civil Space Infrastructure “13th Five-Year Plan” Land Observation Satellites (Grant No. 2017-000052-73-01-001735) Version v3 update (3 November 2026): The original LT1_Aggregate_Data_Code_v1.0_SixAuthors.zip from record v2 (DOI 10.5281/zenodo.23614410) is retained unchanged. A separate Supplementary_Code_JSTARS.zip adds the diagnostic interface, fixed-case adapter, support-summary replay and group-composition resampling implementation accompanying the current manuscript. It uses Python 3.8 or later with the standard library only. The diagnostic suite passes 27 interface, numerical and key-integrity checks and replays 150 released group-window energy rows; the support and group-composition scripts reproduce the distributed JSON/CSV outputs byte for byte. The example inputs are copies of released aggregate tables; no new field observations are introduced. This verification concerns declared interface behavior and aggregate arithmetic only. It does not independently validate source extraction, physical accuracy, object identity or calibration independence. See Zenodo_v3_update_notes.txt and the diagnostic archive README.md for changes, execution instructions, licensing and scope.