Characterization Of Precipitation Trend Changes In China Based On Quantile Regression

Precipitation data are the result of coupled non-linear complex systems at multiple spatial and temporal scales,and the effects of climate change are not only reflected in changes in the mean or tail extremes of precipitation,but also in the various components of precipitation.This paper takes full advantage of the fact that quantile regression does not have strict requirements on data distribution,but can provide trend analysis characteristics of data at each quantile,and investigates the trend characteristics of the model data at different quantile levels under different seasonal precipitation observations and different warming scenarios in China,demonstrating the heteroskedasticity of precipitation data and presenting a complete view of precipitation trends in China.First,quantile regressions were applied to analyse the regression trends in the q=0.1,0.2,…,0.9 quantile for the four seasons 1961-2020,respectively.The trend characteristics of low,normal,high and flooding years for each season are characterized respectively.The spatial distribution of the four seasonal trend coefficients of observed precipitation in China is basically similar at different quartiles,but as the number of quartiles gradually increases,the spatial extent of enhancement(weakening)becomes larger and more and more areas are enhanced(weakened)through significant enhancement(weakening),i.e.the annual precipitation trend of observed precipitation in China is bipolar in flood years.Annual precipitation in spring shows a weak increasing trend in the lower and middle divisions of the middle and lower reaches of the Yangtze River and a gradual decrease in the very high divisions;annual precipitation in summer shows a weak decreasing region in the lower divisions of the northeast and a significant strengthening in the high and very high divisions,characterized by a complete phase shift;annual precipitation in autumn shows a significant increasing trend in the very high divisions of the middle and lower reaches of the Yangtze River;annual precipitation in winter shows a significant The annual precipitation in autumn shows a significant increase in the very high partition in the middle and lower reaches of the Yangtze River,and in winter shows a significant increase in the very high partition in southern China.The variability of annual precipitation trends in these regions is notably inconsistent at different sub-gradients,and the heteroscedasticity of the data is relatively obvious,which deserves more attention in the study of precipitation characteristics and mechanisms.Second,in order to explore the trend characteristics of Chinese precipitation at different percentile levels in each season under future warming scenarios,the NCC model BCC-CSM2MR was applied to investigate the trends of daily precipitation at q=0.3,0.5,0.7,0.9,0.5,0.5,0.6,0.7,and 0.9 percentile levels for 2015-2100 under SSP2-4.5 and SSP5-8.5 scenarios with moderate and high radiative forcing,respectively,0.9 quantile.The spatial characteristics of the regression trends at different quartiles are basically similar for the four seasons,and the annual precipitation tends to increase in most regions of China under the future climate change scenario,with the higher the quantile,the more obvious the increasing trend.However,in the SSP2-4.5 scenario,the annual autumn precipitation in northwest China tends to decrease more significantly as the quantile rises,and the likelihood of flooding in autumn increases in northeast China,while it gradually decreases in low rainfall and normal years,making these regions key areas for precipitation analysis under moderate radiative forcing.Under the SSP58.5 scenario,annual precipitation in the northwest decreases gradually at all deciles,not only in autumn but also in summer.The ecological vulnerability of the Northwest,located in an arid region of the country,makes this region worthy of more attention.Compared to the precipitation observations,the warming scenario shows a more consistent trend in precipitation at all deciles,and the heteroscedasticity is less pronounced.Finally,to address the anomalous feature that annual summer precipitation in Northeast China has opposite trends at high and low quartiles,and to take advantage of the fact that quantile regression can detect quantitative relationships between variables and influencing factors at different quartiles,the main key circulation factors affecting summer precipitation in Northeast China,EAP remotely correlated type,EAP high pressure and EAP high pressure,are selected as the response variables,taking into account the different characteristics of summer precipitation in Northeast China in early summer and high summer,The results show that EAP telecorrelation has a significant impact on summer precipitation in Northeast China.The results show that EAP remote correlation has a positive effect on summer precipitation in southeastern Northeast China and a negative effect on summer precipitation in western Northeast China,both of which are more significant in flood years.The intensity of high pressure in South Asia has a consistent positive influence on summer precipitation in Northeast China across the region.The influence of the Northeast Cold Eddy on early summer precipitation in Northeast China is relatively stable,but the influence of peak summer precipitation is more sensitive.The quantitative effects of the four key circulation factors on summer precipitation in Northeast China from 1961 to 2020 are considered as regression models,and the simulated results of early and peak summer precipitation in Northeast China in 2021 are given.