Variation and attribution of probable maximum precipitation of China using a high-resolution dataset in a changing climate

Abstract

Accurate assessment of the probable maximum precipitation (PMP) is crucial in assessing the resilience of high-risk water infrastructures, water resource management, and hydrological hazard mitigation. Conventionally, PMP is estimated based on a static climate assumption and is constrained by the insufficient spatial resolution of ground observations, thus neglecting the spatial heterogeneity and temporal variability of climate systems. Such assumptions are critical, especially for China, which is highly vulnerable to global warming in 1/4g100g000 existing reservoirs. Here, we use the finest-spatiotemporal-resolution (1gd and 1gkm) precipitation dataset from an ensemble of machine learning algorithms to present the spatial distribution of 1gd PMP based on the improved Hershfield method. Current reservoir design values, a quasi-global satellite-based PMP database, and in situ precipitation are used to benchmark against our results. The 35-year running trend from 1961-1995 to 1980-2014 is quantified and partitioned, followed by future projections using the Coupled Model Inter-comparison Project Phase 6 simulations under two scenarios. We find that the national PMP generally decreases from southeast to northwest and is typically dominated by the high variability of precipitation extremes in northern China and high intensity in southern China. Though consistent with previous project design values, our PMP calculations present underestimations by comparing them with satellite and in situ results due to differences in spatial scales and computation methods. Interannual variability, instead of the intensification of precipitation extremes, dominates the PMP running trends on a national scale. Climate change, mainly attributed to land-Atmosphere coupling effects, leads to a widespread increase (>g20g%) in PMP across the country under the SSP126 scenario, which is projected to be higher along with the intensification of CO2 emissions. Our observation-and modeling-based results can provide valuable implications for water managers under a changing climate.