摘要:The intensity of extreme precipitation has been projected to increase with increasing air temperature according to the thermodynamic Clausius–Clapeyron (C-C) relation. Over the last decade, observational studies have succeeded in demonstrating the scaling relationship between extreme precipitation and temperature to understand the projected changes. In mid-latitude coastal regions, intense precipitation is strongly influenced by synoptic patterns and a particular characteristic is the long-lasting heavy precipitation driven by abundant moisture transport. However, the effect of synoptic patterns on the scaling relationship remains unclear. Here we conduct an event-based analysis using long-term historical records in Japan, to distinguish extreme precipitation arising from different synoptic patterns. We find that event peak intensity increases more sharply in persistent precipitation events, which lasted more than 10 h, sustained by atmospheric river-like circulation patterns. The long duration-accumulated precipitation extremes also increase with temperature at a rate considerably above the C-C rate at higher temperatures. Our result suggests that long-lasting precipitation events respond more to warming compared with short-duration events. This greatly increases the risks of future floods and landslides in the mid-latitude coastal regions.