Scaling of precipitation extremes with temperature in China’s mainland: Evaluation of satellite precipitation data

Abstract

This study explores the sensitivity (termed scaling factor, SF) of daily and 30-minute precipitation extremes with several temperature variables, i.e., within-day surface air temperature (SAT) and dew point temperature (DPT), and antecedent SAT and DPT (corresponding to temperatures one day before a precipitation event, denoted as SAT-C and DPT-C) across China’s mainland. To this end, we used observed daily meteorological data from CN05.1 dataset and 30-minute precipitation data from the Integrated Multisatellite Retrievals for the Global Precipitation Measurement (IMERG). Our results reveal a mix of the positive and negative SFs of extreme daily precipitation with SAT across climatic zones, with peak-like structures developing at higher temperatures (between 17 and 24 °C). Although almost all the SFs turn to positive when SAT-C, DPT, and DPT-C are used, a peak structure is observed over some parts of each climate zone, especially in tropical regions. A comparison between the SFs of the full temperature range and the temperature range before peak structure reveals that a single scaling rate is not valid for the entire temperature range. Moreover, the SFs calculated based on the temperature range before the peak structure (for all four types of temperatures) follow better the Clausius-Clapeyron scaling (∼7%/°C) than the SFs of the full temperature range except for the tropical region. Daily SFs based on IMERG data are mostly comparable to CN05.1 results, with discrepancies mainly in tropical and plateau climates (roughly 25% of the study area). However, IMERG’s 30-min precipitation extremes do not rise as much as expected (even decrease in some parts of the country) with increasing temperatures, contrary to common observations reported in previous studies. It suggests that another precipitation dataset is needed for scaling precipitation extremes at a 30-minute scale, at least for China’s mainland.