Study on multiple rainfall data applied to debris flow warning in Taiwan

Due to climate change, precipitation characteristics have been significantly variation and rainfall patterns are presented more concentrated, high-intensity and long-duration trend in the past two decades. Catastrophic debris-flow disaster threaten lives and property of residents. For mitigation impact of debris-flow, SWCB (Soil and Water Conservation Bureau, Taiwan) has had a leading role in sponsoring debris-flow research and developing a rainfall-based debris-flow warning model. Early warning criteria for debris-flow triggered are also determined depending on the historical rainfall data, and the observational data of rain-gauge are adopted to issue debris-flow warning. However, application of rain-gauge rainfall data has some disadvantages such as low density in mountain area, observation failure to properly represent actual rainfall condition, and data transmission likely interrupted during heavy rainfall or Typhoon. In order to improve the efficiency of debris-flow warning system, two types of gridded precipitation are analyzed and discussed in this study, which are the spatial interpolation rainfall of rain-gauge and the radar-estimated rainfall (QPESUMS). For comparison the differents of multiple rainfall data mentioned above with rain-guage, the third quartile is firstly applied to calculate the regional representative rainfall from grid cells within warning issued area. The results show that the spatial interpolation rainfall underestimates the rainfall intensity and cumulative rainfall owing to the influence of complex topography. By contrast, the radar-estimated rainfall has the advantage in comprehension of the rainfall spatial variability and provide a more complete spatial coverage. Besides, for assessing the appropriate and feasibility of multiple rainfall data applied to debris flow warning, the disaster–capture ratio has been proposed which is defined as the number of debris-flow hazards after issuing warning divided by total number of debris- flow hazards. According to analyis results of historical disaster records from 2012 to 2016, the disaster–capture ratio are 47.6%, 38.1% and 61.9% as warning issued refer to rain gauge, the spatial interpolation rainfall and the radar-estimated rainfall respectively. By the aforementioned process, we realize that the application of radar-estimated rainfall to debris flow warning is obviously increasing efficiency of debris-flow warning ,and gives assistance for reducing uncertainty of rainfall observational data, especially in mountain area.