Non-stationarities in the occurrence rate of heavy precipitation across China and its relationship to climate teleconnection patterns

The variability of occurrence rate of heavy precipitation (HPOR) across China was analysed to examine if the change in heavy precipitation satisfied the stationarity assumption. HPOR obtained by peaks-over-threshold (POT) sampling can be interpreted as a realisation of a point process, which can be represented as a Poisson process. A number of methods, including kernel density estimation method, Cox regression model, Poisson regression model and Generalized Additive Models for Location, Scale, and Shape (GAMLSS), were used to evaluate the nonstationarity in HPOR. Nonstationarity was also assessed for HPOR defined by different thresholds which indicate different levels of severity of heavy precipitation. Our results indicated that: (1) HPOR in northwestern China, a typical arid region, is subject to the most uneven temporal distribution and a significantly increasing trend in terms of inter-annual variability. Moreover, this uneven temporal distribution is increasing for greater heavy precipitation events defined by higher percentiles, and the changing pattern is opposite in eastern, central and southwestern China; (2) the seasonal variability of HPOR is significantly related to climate indices during the year, which shows that the HPOR is not independent and exhibits nonstationary with seasonality. However, as the thresholds increase, the impacts of climate indices on HPOR are found to be weakening; (3) the annual number of greater heavy precipitation is subject to over-dispersion, however, under-dispersion can be identified for the annual number of heavy precipitation events defined by lower thresholds. Higher Southern Oscillation Index (SOI) and Pacific Decadal Oscillation (PDO) may trigger an increase in the frequency of heavy precipitation in northeastern and eastern China. These findings show the challenges of flood preventions and water resources management in a changing climate.