Multivariate Probabilistic Method For Precipitation Temporal Downscaling And Bias Correction Under Climate Change

Rainfall is the origin of surface runoff and the main source of groundwater recharge.The uneven spatial distribution of rainfall and the instability of time are also the direct causes of floods and droughts.Therefore,in the research and practical work of hydrology and water resources,it is very important to correctly and reasonably simulate the rainfall process to solve the increasingly frequent disaster phenomenon.As one of the new methods,the Copula function method has extremely high accuracy and wide applicability in solving the hydrological multivariate joint distribution.In this study,the Copula function is applied to the rainfall temporal downscaling and bias correction,and the collected rainfall data is used to establish multivariate probability rainfall time downscaling and bias correction model.The 1-hour extreme rainfall was selected as the model output value,and some megacities(Guangzhou and Shenzhen)in the Pearl River Delta were taken as a research case to verify the availability of the model.Finally,the outputs of multiple regional climate models were combined to reveal the potential changes of 1-hour heavy rainfall data under climate change,which provides a new idea for the acquisition of high-precision rainfall data.The main research contents of this article are as follows:Review and collect relevant research at home and abroad,grasp the background and progress of rainfall time downscaling and bias correction research,and summarize the advantages of Copula function compared to traditional methods.The definition,type and parameter estimation of Copula function are introduced in detail,and the optimal selection of Copula function is discussed.Based on the determination of the marginal distribution of multivariate rainfall,the Copula function is evaluated and optimized.Finally,a multivariate probability rainfall time downscaling model is established using the most appropriate Copula function.Applied the above model to the study of Guangzhou and Shenzhen data.The Vine-Copula structure was used to establish a joint distribution model of daily rainfall,maximum 6-hours rainfall and maximum 1-hour rainfall,and the differences and similarities between the predicted and observed values.were compared and analyzed.The above model was applied to the analysis of rainfall changes in the history(1976-2005)and the future(2070-2100)of Metropolitan in the Pearl River Delta under climate change conditions to help understand the possible changes in the short-term heavy rainfall.The results showed that even if the outputs from different RCMs will lead to different 1-hour heavy rainfall results,the magnitude and frequency of 1-hour heavy rainfall will generally increase to some extent in the future especially under RCP8.5.Finally,the Copula function is applied to the rainfall bias correction of the metropolitan cities in the Pearl River Delta.A three-dimensional copula function model is established by two RCM data and historical data.The output is compared with the result of the quantile mapping to improve historical data.The results show that the Copula function has a better bias correction effect than the quantile map,and also proves the availability of the Copula function in rainfall bias correction.