Exploring The Applicability of Radar-Based Quantitative Precipitation Estimates For Emergency Assessment of Post-Wildfire Debris Flow Hazards In Colorado

Debris flows in a post wildfire landscape are often initiated by short duration, high intensity rainfall. The U. S. Geological Survey relies on in-situ rain gages co-located with debris flow events to characterize the rainfall triggering conditions. Spatial coverage of rain gages lack continuity and gages are often scarce or nonexistent, especially in mountainous regions where they are difficult to install and maintain. Several remotely sensed products provide gridded estimates of precipitation that could improve the spatial and temporal characterization of rainfall which initiates post-wildfire debris flows. If gridded precipitation products are adequate, they can be used to improve rainfall thresholds, and to understand debris flow initiation in areas with sparse gage coverage or no gages. This study explores the potential for remotely sensed gridded precipitation products to enhance emergency assessments of post-wildfire debris-flow hazards. The Multi-Radar Multi-Sensor (MRMS) system provides a mosaic of Quantitative Precipitation Estimates (QPE) for the contiguous United States with a spatial resolution of 1 km. These products are radar-based, with multi sensor corrections from local gages, satellite data, atmospheric environmental data, and controls for orographic effects. We compare MRMS estimates of storm-total accumulation and peak storm intensity to local tipping-bucket rain gage measurements in several recently burned areas in Colorado. Preliminary results suggest that the multi-sensor 1-hour QPE storm total accumulation performs well, with some underestimation. A simple multiplicative bias correction, using the average total accumulation of the local gages, reduces the bias. Interestingly, an evaluation of the gridded product shows that the grid cell containing a rain gage often shows more bias than one of the eight nearest neighbor cells. Because radar beams are blocked by complex terrain in our area of interest, the surface precipitation is estimated from moisture observed above ground level. The raindrop travel distances, from the observation to the ground, may preclude direct mapping between ground measurements and moisture observed at higher elevations. Furthermore, the spatial heterogeneity of storm accumulation and intensity, observed in the gridded product, suggests a more limited spatial accuracy of rain gages than is often assumed. MRMS estimates of peak 15-minute rainfall intensity were generally much lower than gage measurements. This underestimation is potentially because the two-minute MRMS rainfall intensity products do not reflect adjustment from the Hydrometeorological Automated Data System network of hourly rain gages. We plan to investigate whether a comparison of MRMS radar-only and MRMS multi-sensor one-hour QPEs may provide an avenue to adjust the two-minute MRMS precipitation intensity data in a way that improves debris-flow hazard assessment for locations with sparse or no local gages.