Datasets used in "Quantifying VIIRS and ABI Contributions to Hourly Dead Fuel Moisture Content Estimation Using Machine Learning"

Data sets used to train, validate, and test XGBoost machine learning models for hourly 10-hour dead fuel moisture content (FMC) estimation across the contiguous United States. The datasets integrate three primary data sources at 375 m spatial resolution: (1) High-Resolution Rapid Refresh (HRRR) numerical weather prediction model outputs including near-surface atmospheric variables and soil state information, (2) Visible Infrared Imaging Radiometer Suite (VIIRS) aboard Suomi-NPP surface reflectances and land surface temperature retrievals, and (3) Advanced Baseline Imager (ABI) aboard GOES-16 reflectances, brightness temperatures, and land surface temperature retrievals. The datasets span the 2020-2021 period and include spatially interpolated predictor variables matched to Meteorological Assimilation Data Ingest System (MADIS) observational sites. Two temporal configurations are provided: (1) co-located observations where VIIRS and ABI are temporally aligned (approximately 35,253 data points), and (2) hourly observations across all available times (approximately 1,004,567 data points) with VIIRS lag-hour tracking to account for temporal offset between polar-orbiting and geostationary observations. The datasets include standardized predictor variables, quality control flags for cloud and snow masking, and geospatial metadata. Processing and model training code is available at https://github.com/NCAR/fmc_viirs.