Satellite derived rainfall threshold for flash flood in urbanized tropics

Abstract Determining the threshold for antecedent rainfall that led to flash flood in sparse, unavailable rain gauge, or conflict hydrological data retrieval in tropical cities is vital to support the flash flood adaptation framework. The satellite precipitation of Global Precipitation Mission (GPM) is promising as alternative due to its fine grid (~10km) and high temporal resolution. Utility of GPM data as sole input for triggering threshold in urbanized tropics is less reported. Consequently, this paper attempt to determine the threshold for antecedent rainfall for flash flood occurrences using GPM satellite data in Klang Valley area as experimental site. The results showed that the quantitative cumulative rainfall threshold derived from the percentile plot showed that the value was not vary much between the zone from 5 - 7mm for 3 hourly rainfall period. The IDF method obtained the non-linear fashion of exponential and logarithmic relationship. Accuracy assessment results showed that the performance overall accuracy for all indicators were good (>85% accuracy). The percentile method slightly scored better than IDF method. The outcome of this study is useful in providing alternative for data conflict areas to determine reliable triggering threshold for flash flood as the satellite data is can be publicly accessed.