Feasibility of satellite-based rainfall and soil moisture data in determining the triggering conditions of debris flow: The Jiangjia Gully (China) case study

Abstract

Heavy precipitation is the main trigger for debris-flow hazards. The antecedent moisture condition, which is usually represented by antecedent precipitation, is another hydrometeorological contributing factor to debris flows that originate from shallow landslides. Satellite techniques are an economical and effective way to access rainfall and soil wetness information to determine the triggering conditions of debris flow. However, satellite-based thresholds need to be compared with ground-based thresholds and adjusted to the data source prior to their application. In this study, rainfall intensity-duration (I-D) thresholds were derived from the early and final run products of the Integrated Multi-satellitE Retrievals for Global Precipitation Measurement (IMERG-E and IMERG-F) and gauge measurements using logistic regression for debris flows in the Jiangjia Gully, Yunnan Province, southern China, which has a dense rain gauge network. Data from both the IMERG-E and IMERG-F covered the entire study period. Our evaluation revealed that the I-D threshold derived from the IMERG-E deviated from the gauge-based thresholds. Although the threshold determined by the IMERG-F was comparable to the ground-based ones, the presence of substantial false positives and false negatives indicated that its performance was weaker than that of the gauge-based ones. Furthermore, the IMERG-F was suitable if the nearest available gauge was farther than 10 km from the debris-flow initiation zone. We evaluated the feasibility of the surface soil moisture product provided by the Climate Change Initiative program of the European Space Agency (CCI-SM), root-zone soil moisture derived from the CCI-SM using an exponential filter (SM-RZ), and antecedent precipitation for improving the performance of the thresholds by separately using them as the third explanatory variable in addition to rainfall intensity and duration in logistic regression. Satellite soil moisture data were available for 54% of the study period. The results suggested that including antecedent precipitation effectively improved the performance of the thresholds. In contrast, the performance of the thresholds increased only slightly when the CCI-SM or SM-RZ was included. Although these findings are valid only for this study area and need to be assessed for other regions, they present new insights for using satellite rainfall and soil moisture estimates to define thresholds for debris flow.