Spatial Effect Of Annual Runoff Response To Changing Environment In The Yangtze River Basin Based On The Bayesian Model

In recent decades,global warming and intensified human activities have brought a great disturbance to the global hydrological cycle.In essence,runoff is the product of the comprehensive effect of climate change and human activities.Because precipitation factors affected by climate change and human activities,as well as topographic and geomorphic factors forced by climate change and human activities,there is no spatial correlation.That is,the natural elements of one region will affect the natural elements of another region,which is also the essence of the first principle of geography proposed by Tobler in 1970: "Everything is spatially related,and the spatial correlation of things near is greater than that of things far away." Therefore,it has become a focus of scholars at home and abroad to understand runoff and its response to climate change and human activities from the direct water vapor transport source of runoff-precipitation and the topographic and geomorphic factors that affect the shape of precipitation in surface space.Among them,the spatial correlation of precipitation,vegetation cover and other factors plays a non-negligible role in runoff change.Therefore,studying the response characteristics and spatial effects of runoff to climate change and human activities has good scientific significance and practical value not only for understanding the physical mechanism of runoff formation but also for water-related socio-economic activities such as water resources planning and management.Although some researches on these issues have been carried out by predecessors,the physical mechanism and spatial effects of annual runoff response are still lacking.In order to solve this problem,the improved Gini coefficient method is used to analyze the spatial matching relationship between annual runoff and other factors,and then the spatial correlation between them is analyzed.In addition,an analysis model of annual runoff response was constructed by coupling the Budyko framework with the Bayesian hierarchical spatio-temporal model,and the influences of meteorological factors and underlying surface factors on parameter n and annual runoff in the Budyko equation were analyzed.The main research work of this paper is as follows:(1)Based on the spatio-temporal evolution law of hydrometeorological elements,Pearson correlation coefficient,improved Gini coefficient and Copula entropy method were used to study the correlation and spatial matching relationship between runoff and precipitation,actual evapotranspiration,potential evapotranspiration and NDVI,as well as among multiple factors.The results show that the precipitation,actual evapotranspiration,potential evapotranspiration and runoff in the Yangtze River Basin have a certain increasing trend in time,and gradually increase from upstream to downstream in space.However,NDVI is spatially large in the upstream and small in the downstream,which is just the opposite of other variables.In addition,Moran index,cluster analysis,spatial regression model and other methods showed that annual runoff in the Yangtze River Basin had strong spatial autocorrelation,and the spatial response equation between annual runoff and precipitation,actual evapotranspiration,potential evapotranspiration and NDVI had a better fitting effect,and the spatial error model had the best performance.(2)In order to further study the spatial effects of runoff response,this paper combined Budyko equation,water balance equation and Bayesian hierarchical spatio-temporal model to build a runoff spatio-temporal response model,and analyzed the simulation effects of four models on runoff and Budyko parameter n.The results showed that,The Bayesian hierarchical spatio-temporal model considering spatio-temporal interaction terms has better simulation effect,and the R2 of annual runoff simulation can reach 0.91.In addition,by comparing the simulated value and the actual value of annual runoff,it is found that there is little difference between the two on the whole,but the simulated value is lower than the actual value at the extreme value.(3)In order to analyze the root cause of spatio-temporal variation of annual runoff,based on the Bayesian hierarchical spatio-temporal analysis model,this paper studied the effects of NDVI,nighttime light index(NTLI),precipitation seasonal index and actual evapotranspiration seasonal index on parameter n,considering the climate factors and underlying surface factors affecting parameter n in Budyko equation.And the geographical weighted regression model is used to analyze the spatial range of four influencing factors on parameter n.The results show that NDVI and NTLI have great influence on parameter n,but their spatial distribution is different.However,seasonal indexes of precipitation and actual evapotranspiration have little influence on parameter n.Both NDVI and NTLI have a significant influence on parameter n in most basins,while the actual evapotranspiration seasonal index has a small spatial range of influence,and the precipitation seasonal index has no significant influence on runoff change.