Short-Term Hydroclimatic Precursors Before the First Satellite Detection of Peat-Fire Events in South Sumatra: Evidence from Sentinel-1 Reconstruction

Repository description This repository contains the data products, quality-control records, analytical inputs, protocols, and supporting materials associated with the study “Short-term hydroclimatic precursors before first satellite detection of reconstructed peat-fire events in South Sumatra: complementary evidence from Sentinel-1.” The study examines short-term environmental conditions preceding the first retained satellite active-fire detection of reconstructed tropical peat-fire events in the Tanjung Lago–Banyuasin II peatland landscape, South Sumatra, Indonesia. The underlying event inventory contains 218 reconstructed candidate fire events, from which 79 non-singleton events were retained for the primary time-stratified case-crossover analysis, corresponding to 79 case observations and 234 same-weekday referent observations (313 observations in total). The repository includes materials related to event reconstruction and case–referent construction, observation-opportunity and prior-event quality control, Sentinel-1 antecedent-window extraction, surface-wetness proxy construction, model-ready datasets, inferential outputs, sensitivity analyses, and analytical protocol documentation. Environmental information used in the study includes CHIRPS rainfall, ERA5-Land vapor pressure deficit (VPD) and soil-water indicators, Sentinel-1 C-band backscatter, and MODIS/VIIRS active-fire observation information. The primary analysis used a time-stratified case-crossover design in which each event location was compared with itself on same-weekday referent dates within the same month and year. Hydroclimatic exposures were evaluated across prespecified non-overlapping lag windows before the first retained active-fire detection. Sentinel-1 observations were evaluated as complementary moisture-sensitive information using prespecified antecedent windows and baseline periods. Conditional logistic regression and stratum-level bootstrap resampling were used for statistical inference and uncertainty assessment. An important feature of the study is the explicit treatment of satellite observation opportunity. The outcome time (t0) represents the first retained satellite active-fire detection and should not be interpreted as the actual time of fire ignition. Sentinel-1 backscatter metrics are likewise interpreted as moisture-sensitive indicators of surface conditions rather than direct measurements of peat soil moisture or groundwater-table depth. The repository is intended to support transparency, reproducibility, methodological verification, and reuse of the derived analytical products. Users should consult the README, file manifest, protocol documentation, and accompanying manuscript before reusing individual files because some datasets represent intermediate, quality-controlled, or analysis-specific populations. Study region: South Sumatra, IndonesiaStudy period: 2015–2025Primary analytical population: 79 reconstructed non-singleton peat-fire eventsPrimary design: Time-stratified case-crossoverMain data sources: MODIS, VIIRS, Sentinel-1, CHIRPS, and ERA5-LandPrimary research themes: peat fire, hydroclimate, remote sensing, Sentinel-1, vapor pressure deficit, satellite active-fire detection, and environmental monitoring.