Sunil Thapa, Tek Maraseni, Armando Apan, Bikram Banerjee
Abstract Fire regimes are intensifying globally under climate change, yet comprehensive multi-seasonal assessments integrating fire activity with environmental drivers remain scarce for tropical and subtropical landscapes. This study quantifies climate-driven fire susceptibility trends across Queensland, Australia (2001–2024), a region spanning monsoonal tropical rainforests (>2500 mm yr −1 ) to semi-arid grasslands (<300 mm yr −1 ). We developed two complementary indices: a trend-based weighted combined index (WCI) and a component-based seasonal fire index (SFI), synthesizing temporal trajectories and mean-state conditions across seven environmental variables (i.e. fire frequency, burned area, temperature, vapor pressure deficit, soil moisture, precipitation, and normalized difference vegetation index). Analysis of 1.52 million moderate resolution imaging spectroradiometer and VIIRS active fire detections revealed pronounced spatiotemporal heterogeneity, with spring accounting for 57.4% of activity and annual burned area varying sixfold (61 788–252 179 km 2 ). Precipitation emerged as the dominant bioregional control ( ρ = 0.74, p = 0.006), reflecting the productivity paradox where wetter regions support greater fuel accumulation. Northern tropical savannas exhibited persistent high susceptibility (WCI > 0.50), while southern coastal regions remained low-risk (WCI < 0.25). Three distinct temporal phases emerged: extensive burning (early 2000s), suppression (2013–2020), and recent resurgence (2023–2024) driven by fuel-climate feedbacks from the 2020 to 2022 La Niña. Seasonally, winter displayed maximum fire susceptibility values (mean WCI = 0.402), establishing critical pre-fire season conditions that preceded peak spring fire activity. Model validation demonstrated good predictive performance for autumn (AUC = 0.763) and acceptable discrimination for winter-spring seasons (AUC = 0.63–0.69). Comparative analysis revealed that the trend-based WCI model exhibited superior temporal stability and long-term predictive capability, while the component-based SFI model demonstrated greater sensitivity to seasonal extremes and interannual climate variability. These findings provide critical empirical evidence for adaptive, bioregion-specific fire management strategies that integrate Indigenous fire knowledge, seasonal climate forecasts, and emerging remote sensing technologies to enhance landscape resilience under accelerating climate change.