Western U.S. reference basins had more runoff in the 1-6 water years after large wildfires, and the 2020-2022 fires confirmed the 1985-2019 increase out of sample (CAMELS, MTBS, annual data)

Williams et al. (2022) reported that western U.S. streamflow increased by approximately 30 % for 6 years after large forest fires of 1984-2019. We used the large 2020-2022 fires as an independent test. Data: 317 western CAMELS reference basins, water-year runoff and basin precipitation and temperature for 1952-2025, and MTBS burned-area perimeters. For each basin, we fitted a climate model of log runoff on years without fire. We compared each burned basin with unburned basins in the same HUC-02 region and years (residual difference-in-differences: post-fire years 1-6 minus pre-fire years -5 to -1). Events had 20 % or more of the basin burned. For 33 events of 1985-2019, runoff was +21 % above the matched null (null 95 % range -8 % to +9 %). For 17 out-of-sample events of 2020-2022, runoff was +37 % (null -10 % to +12 %), and 16 of 17 events were positive. In 12 fire clusters, 11 means were positive (p = 0.006). A time-shift placebo on the same basins also rejected zero (one-sided p = 0.04). All 10 variants (10 % threshold, forest basins, no pre-fire subtraction, trend, other transforms) gave positive excess outside the null range (+9 % to +28 % for 1985-2019, +16 % to +53 % for 2020-2022). The excess increased with burned fraction and was near zero after post-fire year 6. For 2020-2022 fires it was large in dry water years 2021, 2022 and 2025 and near zero in wet 2023-2024; a dependence on wetness was not resolved (pooled p = 0.07). Limitations: perimeter-based burned area without severity, a small out-of-sample set concentrated in California, annual data only, and possible post-fire gauge rating changes.

wildfire · streamflow · runoff · post-fire hydrology · water supply · CAMELS · MTBS · out-of-sample test · difference-in-differences · western United States · replication