Data and code from: Characterizing wildfire behavior with ECOSTRESS land surface temperature across four California case studies

Since 2000, wildfires in the western United States have increased in both
frequency and intensity due to climate warming, prolonged drought, and
expanded human activity. Although geostationary systems enable rapid
detection and moderate-resolution sensors offer broad coverage, a gap
persists for high-spatial-resolution thermal observations that can assess
fine-scale fire behavior. The ECOsystem Spaceborne Thermal Radiometer
Experiment on Space Station (ECOSTRESS) provides 70-meter land surface
temperature (LST) observations with multi-day revisit intervals, providing
enhanced spatial detail for active-fire analysis. In this study, we
evaluate the capability of ECOSTRESS Level 2 LST data, which uses 5
thermal bands, to characterize wildfire behavior across four California
fires: Carr (2018), Kincade (2019), August Complex (2020), and Dixie
(2021). We developed a consistent framework to identify hotspots (LST ≥
60°C), estimate a satellite-derived rate-of-spread (ROS) proxy using the
95th percentile radial expansion from ignition, and assess relationships
between mean active-fire temperature and mean post-fire burn severity
(dNBR). Across all fires, median hotspot temperatures ranged from 62 to
77°C, while 95th percentile values ranged from 117 to 179°C, indicating
right-skewed radiometric distributions. The ROS proxy showed directional
variability, with median values typically between ~0.05 to ~3 km/day and
substantial heterogeneity among quadrants. Regression indicates consistent
positive relationships between mean active-fire temperature and mean burn
severity, with R² values from 0.51 to 0.88. These relationships were
stronger in smaller, short-duration fires. Our findings indicate that
ECOSTRESS provides valuable high-spatial-resolution thermal observations
that can resolve fire growth patterns and link active-fire thermal
dynamics to subsequent burn severity.