Quantitative Analysis And Application Of Regional Drought Based On Copula Function And Bayesian Network

Drought is a complexand destructive natural disaster,which seriously threatens the development of agriculture,ecology and social economy.At present,the research results on drought are scattered,lack of systematization and operability,especially the research on drought transmission is relatively weak.The quantitative analysis of regional drought,which focuses on the analysis of spatio-temporal evolution characteristics,the mining of drought periodicity and the quantification of drought trigger threshold,is helpful to further deepen the understanding of regional drought,and is also the basis of regional drought assessment and prevention.In this study,taking the Hunhe River Basin(HRB)as an example,Standardized Precipitation Index(SPI)and Standardized Runoff Index(SRI)were applied to reveal the spatio-temporal evolution characteristics and regional differences of drought,the characteristics of drought cycle,the driving factors of drought and the trigger threshold of drought.Specifically,the Copula function and Bayesian network were introduced to establish the meteorological trigger threshold model of hydrological drought.The proposed quantitative analysis approach of the meteorological trigger threshold model of hydrological drought was verified in HRB.The research results can provide scientific support for describing regional drought evolution situation,assessing drought risk,developing drought resistance measures and groundwater development and protection.The primaryfindings of thestudy canbe concluded as below:(1)Based on SPI and SRI sequence,the modified Mann-kendall trend test method and inverse distance weight spatial interpolation method were adopted to systematically analyze the temporal evolution characteristics and spatial distribution of drought in HRB.The results showed that,the meteorological drought showed a weakening trend in the middle and lower part and the eastern part of HRB,while a strengthening trend in the middle and upper part and the western part of HRB.Hydrological drought showed an increasing trend in the whole HRB,and the enhancement effect of hydrological drought gradually increasedfrom thelowerreaches to theupperreaches.(2)The drought return period of HRB were calculated by Copula function based on the drought characteristic variables extracted from the modified run theory.It can be seen from the calculation results that the return periods of moderate and severe meteorological drought in BKQ and DHF were basically the same as those of hydrological drought at the same level.In SY and XJWP,the return periods of hydrological drought at different levels were greater than that of meteorological drought at correspondinglevels.(3)The driving factors of meteorological drought in HRB were analyzed using cross wavelet transform.The results showed that Sunspot,El Nino Southern Oscillation(ENSO)and Arctic Oscillation(AO)were strongly correlated with the annual standardized rainfall index(SPI-12)of HRB,which indicated that they had great influence on meteorological drought in the HRB.Specifically,SPI-12 showed negative correlation with sunspots on 8-13 year cycles,positive correlation with ENSO on 3-5 year and 9-14 year cycles,and negative correlation with AO on 1-4year and8-12 yearcycles.(4)The propagation duration and influencing factors of meteorological drought to hydrological drought in four sub-basins of HRB were revealed by Pearson correlation coefficient and its uncertainty analysis.It can be seen from the calculation results that,the propagation time from meteorological drought to hydrological drought in BKQ,DHF,SY and XJWP basins were 4.5,4.5,17.2 and15.8 months,respectively.Meanwhile,due to the operationof Dahuofang reservoir(DHFR),thedrought propagationduration inthedownstream area ofthe reservoir wassignificantlyhigherthan that in the upstream area of the reservoir,and the effect of this improvement was weakened with the increase ofthedistance from DHFR.(5)Basing on the determination of drought propagation duration,the meteorological trigger threshold model of hydrological drought was established according to Copula function and Bayesian network,and the meteorological trigger threshold of hydrological drought and drought resistance capabilityin HRB were quantitatively analyzed.The application results showedthat DHF showedthe most vulnerable resistance to hydrological drought,while SY showed the strongest resistance.When moderate hydrological drought was reached,the BKQ and XJWP showed more sensitivity to increases in Cumulative precipitation deficit(CPD),that is,a slight increase in CPD had the potential totrigger moresevere drought.