Evaluating Post-Wildfire Debris Flow Rainfall Thresholds and Volume Models at the 2020 Grizzly Creek Fire in Glenwood Canyon, Colorado, USA
Abstract. As wildfire increases in the western United States, so do postfire debris-flow hazards. The U.S. Geological Survey (USGS) has developed two separate models to estimate (1) rainfall intensity thresholds for postfire debris flow initiation and (2) debris-flow volumes. However, the information necessary to test the accuracy of these models is seldom available. Here, we studied how well these models performed over a two-year period in the 2020 Grizzly Creek Fire burn perimeter in Glenwood Canyon, Colorado, USA, through the development of a debris flow response inventory. The study area had the advantage of a network of 11 rain gauges for rainfall intensity measurements and repeat lidar data for volume estimates. Our observations showed that 89 % of observed debris flows in the first year postfire were triggered by rainfall rates higher than the fire-wide rainfall threshold produced by the current USGS operational model (M1). No debris flows were observed in the second year postfire, despite eight rainstorms with intensities higher than the modeled rainfall threshold. We found that the operational model for debris flow initiation rainfall thresholds works well in this region during the first year but may be too conservative in year 2 due to vegetation recovery and sediment exhaustion. However, rainfall thresholds in the second year can be improved by using updated remote sensing imagery to recalculate the debris-flow initiation probability with the M1 model. The current volume model overpredicts for this region by a median value of 4.4 times. However, the offset between the predictions and observations is linear, and the volumes from the Grizzly Creek debris flows had a similar magnitude to historic postfire debris flows in the region. Consequently, the current volume model could be adjusted with a regional correction factor.
Status: final response (author comments only)
Viewed (geographical distribution)