This dataset presents a high-resolution, multimodal spatiotemporal collection formatted for machine learning analysis of post-fire hydrologic recovery, burn severity, and radar backscatter dynamics associated with the 2025 Monroe Canyon Fire in Utah, USA. The dataset aggregates 10-meter resolution spatial data structured into 621 patches (128 x 128 pixels each) stored as compressed TFRecords (.tfrecord.gz), enabling memory-efficient, zero-RAM-overhead streaming for deep learning pipelines. The collection couples multi-temporal Sentinel-1 Synthetic Aperture Radar (SAR) time series (descending VV, VH, incidence angle, and VH/VV cross-polarization ratio) with Sentinel-2 optical burn indicators (dNBR, pre-fire peak NDVI) and daily GRIDMET precipitation grids across a multi-year monitoring window (July 2024 – September 2026). These dynamic time-series features are paired with static environmental covariates, including USGS topographic variables (elevation, slope, aspect), OpenLandMap soil characteristics (clay, sand, soil organic carbon), LANDFIRE vegetation classifications, and official U.S. Geological Survey / U.S. Forest Service Burned Area Emergency Response (BAER) perimeters. The full spatial extent encompasses 10.17 million valid pixels, sampling both the interior BAER burn scar (3.77 million pixels; 37.01%) and surrounding unburned control terrain (6.41 million pixels; 62.99%) to establish baseline environmental conditions. To prevent signal distortion from prior ecological disturbances, historical MODIS fire footprints (2010 to pre-fire baseline) are masked, leaving 9.52 million "clean" pixels for rigorous model training and comparative hydrologic analysis. This benchmark resource provides the data necessary to decouple pre-fire fuel load biases from post-fire canopy volume scattering loss, advance SAR-based vegetation degradation monitoring, and evaluate post-wildfire watershed recovery.

The following plot depicts radar and hydrologic time-series assessment for the Monroe Canyon Fire (July 2024 – September 2026):