Consecutive Dry Days as a Scale-Dependent Predictor of Tropical Peatland Fire Occurrence in Indonesia

Rusmawan Suwarman, Rafly Azaria, Sandy Hardian Susanto Herho , Edvin Aldrian

Tropical peatland fires in Indonesia generate severe environmental, health, and economic impacts, yet current fire prediction systems exhibit scale-dependent limitations. This study investigates the relationship between Consecutive Dry Days (CDD) indices and fire occurrence across multiple spatial scales in South Sumatra and West Kalimantan provinces (2015-2019). Using hierarchical buffer analysis (25, 50, 100, 150 km radii) around meteorological stations, we analyzed MODIS hotspot data with $>$80\% confidence against CDD classifications. Maximum CDD values reached 41 days (South Sumatra) and 27 days (West Kalimantan) during the 2015 El Ni\~{n}o event. Correlation analysis revealed pronounced scale dependency, with optimal meteorological station representativeness at 50 km radius ($r = 0.776$--$0.821$, $p < 0.01$). Weak negative correlations at 25 km radii reflect urban bias in station placement, while correlations degraded beyond 100 km due to atmospheric boundary layer constraints. Hotspot frequencies increased exponentially with CDD duration, particularly on peatlands where very long droughts ($>$30 days) generated $156.2 \pm 34.7$ hotspots per event. These findings indicate current meteorological networks inadequately sample fire-prone landscapes, suggesting strategic station deployment at 50 km intervals could substantially improve early warning systems across Southeast Asia's vulnerable peatland regions.