Supplementary archive for Does machine-learning fusion improve on calibrated single-sensor aerosol optical depth? A paired, dependence-aware evaluation of a frozen VIIRS–ABI procedure

Supporting data, frozen models, analysis code and figures for a paired evaluation of a frozen VIIRS–ABI aerosol optical depth fusion procedure. The archive includes the complete input table of 259 April–June 2022 development pairs; archived inputs and predictions for 152 observations on additional dates in 2024, including 90 records with qualified AERONET references; stored evaluation and sensitivity-analysis outputs; the frozen model bundle; prediction and model-fitting modules; and figure sources. The primary observations originate from publicly available NOAA VIIRS and GOES-16 ABI products and NASA AERONET Version 3 Level 2.0. This archive starts from derived inputs and does not contain a complete raw-satellite-granule collection. Earlier-panel summaries are included, but complete earlier-panel record-level inputs are outside its scope. Numerical reproduction reports are archived evidence; raw-granule processing and model refitting were not performed as part of this packaging step. Some original fitting modules retain project-specific paths that require adaptation. File sizes and SHA-256 checksums are provided in ARCHIVE_MANIFEST.json. Funding: This research was funded by the Jing-Jin-Ji Regional Integrated Environmental Improvement National Science and Technology Major Project of the Ministry of Ecology and Environment of China (No. 2025ZD1200900). License scope: CC BY 4.0 applies to author-owned data derivatives, figures and documentation; MIT applies to author-owned source code. NOAA, NASA AERONET and other upstream materials and third-party components retain their respective source terms and attribution requirements. This record does not grant additional rights over third-party content. The deposit-time license choices are recorded here; the preserved archive README describes license assignment at deposit.