The Use of Rain Gauge Measurements and Radar Data for the Model‐Based Prediction of Runoff‐Generated Debris‐Flow Occurrence in Early Warning Systems

Abstract

AbstractHigh‐intensity and short‐duration rainfalls can generate sudden and abundant runoff at the base of rocky cliffs that, entraining sediments, may originate debris flows. Two gauge networks have been set up in headwater sites of Dolomites (Northeastern Italian Alps) to monitor rainfall corresponding to the debris‐flow activity occurring there. The rain gauges are positioned both upstream and downstream the initiation areas of debris flows. Other five rain gauges sparse in the area integrate the two networks. In the years 2009–2020, rain gauges recorded rainfalls that triggered 41 debris flows. In most cases, rainfalls show a higher spatial variability along with both distance and altitude. Precipitation data are then compared with rainfalls estimated through a weather radar far about 70 km from there, to verify the possible interchangeability of the two measurement systems for the prediction of debris‐flow occurrence through suitable modeling of triggering discharges. The following results are obtained: (1) raw‐radar images mostly tend to underestimate precipitations recorded by rain gauges; (2) such underestimation entails, on average, a larger one on the simulated discharges and the prediction of debris‐flow occurrences (missed in 65% of the cases). Some methods for the correction on ground truth of raw‐radar images are applied to assess their use for evaluating the triggering discharges. Results show that once corrected using rain gauge data, radar‐derived rainfall estimates produce debris‐flow initiation predictions that more frequently match observations. Therefore, the presence of rain gauges close to the watershed centroids results essential for early warning systems based on triggering discharge modeling.