Spatio-temporal Evolution Of Drought And Future Projections Under Climate Change

Drought is a kind of disaster with strong destructive power and serious economic loss.In order to better prevent and control drought disaster and effectively manage water resources,it is of great significance to study the spatial and temporal distribution and predict drought in different climatic regions.In this study,the meteorological observation data of precipitation(P),minimum temperature(Tmin),maximum temperature(Tmax),average temperature(Tave),wind speed at 2m high(U2),relative humidity(RH),sunshine hours(n)etc.at the 552 selected weather stations(which belonged to different climatic regions in mainland China)since 1961 were collected.Based on the improved Mann-Kendall method,the annual trend of each meteorological element was tested.The FAO56 Penman-Monteith formula was used to calculate the of reference crop evapotranspiration(ET0),and the change characteristics of ET0 and its sensitivity to the meteorological elements were analyzed when the single,double and multiple meteorological elements were changing at the same time.Based on above analysis,ET0 was estimated by different methods(including the Irmak et al.(2003),Berti et al.(2014),Priestley-Taylor(1972),Doorenbos-Pruitt,and Valiantzas formulae,respectively)and further used to calculate the standardized precipitation evapotranspiration index(SPEI).The periodic variation characteristics of SPEI was analyzed by combining the spectral characteristics and the wavelet variance obtained by the wavelet analysis,and the drought evolution characteristics of SPEI in different climatic zones in China was further evaluated.In addition,the observed daily precipitation data were bias-corrected and the differences in SPEI calculated based on observed precipitation(Pm)and bias-corrected precipitation(Pc)were compared.Different drought indexes(including precipitation anomaly percentage,standardized precipitation index-SPI,standardized precipitation evapotranspiration index-SPEI and evaporation demand index-EDDI)were used to study the drought severity and spatiotemporal variations at different timescales from 1 to 12 months.Finally,the statistical downscaled data from 28 global climate models(GCM)in CMIP5 during the period 2020-2100 were used to predict SPEI at the 12-month timescale with 2 representative concentration paths(RCP4.5 and RCP8.5 scenarios),and the future drought evolution characteristics in 7 climate zones and mainland China was analyzed.The main conclusions based on the above analysis are as follows:(1)Under the condition that single meteorological element changed,the reference crop evapotranspiration(ET0)was more sensitive to Tmax,U2 and n.As the two meteorological elements changed simultaneously,the sensitivity of ET0 to the simultaneous changes of Tmin and Tmax was greater than that of the combination of other elements.However,with the simultaneous change of(-U2)&(-n),ET0 decreased more than other conditions.When three or more meteorological elements changed at the same time,the increase of Tmin,Tmax,Tave and RH and the decrease of U2 and n led to the actual decrease of ET0.The reduced ET0 induced by the decrease of wind speed U2 and sunshine hour n compensated the influence of increased Tmax on ET0.(2)Corresponding to the seasonality and regionalization of P,ET0 and D(=P-ET0),the trends in SPEI increased in spring and winter for northwestern and northeastern China but decreased in the four seasons for middle China,indicating seasonality and regionality of drought severity.Drought frequencies and cycles monitored by seasonal and interannual SPEIIRA,SPEIPT,SPEIMHS,SPEIVal and SPEIPM varied with seasonal and regional characteristics.Large deviations were found in SPEIIRA,SPEIPT,SPEIMHS and SPEIVal in the drought of arid and semi-arid regions and the Tibetan plateau region(i.e.,sub-regions 1,2 and 6),indicating that the drought in arid and semi-arid regions where less precipitation was received were more affected by the ET0 estimation methods than the humid and semi-humid regions when using SPEI for drought assessment.However,for most subregions,the differences in the extreme drought events which were indicated by SPEI(SPEI <-2.0)estimated from different ET0 estimation methods were not significant.The results of wavelet analysis showed that although the fluctuations and spectra of different SPEIs in the northwestern China(sub-region 1)were generally similar,the periodic signals were different.(3)In the past 53 years,the P increased and ET0 decreased at most stations in China,and drought severity in China was generally alleviated.The trends of P and ET0 also affected the spatial and temporal distribution patterns of Pa,SPI,SPEI and EDDI.When the timescale decreased from 24-month to 1-month,the temporal fluctuations of SPI,SPEI and EDDI became more and more intense.SPI and SPEI at the 12-month timescale performed better than EDDI when compared to the severe or extreme drought events in history,but all performed poorly under the conditions of sudden turn from drought to flood.In different sub-regions,Pa,SPI,and SPEI performed well in representing historically severe or extreme droughts,but EDDI performed poorly in subregions 1,4,5,and 7.Overall,SPI,SPEI and EDDI could better identify drought-threaten areas,but Pa was less effective.The rank of correlations between pairs of SPI,SPEI and EDDI was SPI vs.SPEI >SPEI vs.EDDI>SPI vs.EDDI.In addition,EDDI at the times from 1-week to 12-week effectively revealed the onset and end of flash floods,which implied the potential applications for flash warnings that SPI and SPEI have not been able to successfully do.Each drought index has its advantages and limitations and should be used carefully.(4)During the period of 1961-2015,the corrected precipitation Pc was generally larger than the observed precipitation Pm,and it gradually decreased as the sub-region rank of 7,5,3,4,6,2 and 1.Therefore,the drought index Im calculated using Pc was greater than Im calculated using Pm.When using Im(based on Pm and Pc)for climate division,the climate types of the different zones remained the same or shifted to a humid climate type.The effect of the precipitation bias correction on SPEI was small,but in the case of severe or extreme droughts,the Pc-based SPEI value was larger,which also meant a wetter condition.Therefore,both the drought index Im and SPEI varied with changes in Pc values,with significant changes of Im in each sub-region and small changes in SPEI.?(5)The temporal and spatial variations of percentage of stations suffering from drought,drought frequency,drought event occurrence time,drought durations,drought severity,and drought peak consistently showed more severe and more frequent droughts and larger extreme drought peaks of droughts in China,especially in northwestern China.Southeastern China,northeastern China,central and eastern China(sub-regions 4 to 7)would experience less severe drought in the coming future.Although the annual precipitation was predicted to increase in most regions of China,it would not mask the aggravating influence of increased minimum and maximum air temperatures on drought,especially in northwestern China.The prevention of drought in northwestern China would remain grim at the end of the 21 st century.The contribution of GCMs to the uncertainty of drought projection was larger than the contributions of stations and RCPs,with a variance percentage of >40%.The drought index SPEI performed better than the other indices that only took into account single variable influences,such as the SPI and the EDDI.The relationships between crop yields and drought indices and the months that had the highest correlations varied with the different crops,drought indices and timescales.Substantial differences were found when assessing the influences of drought evolution on maize and cotton yields(considering the 6 months of the crop-growing season)using different drought indices.Further studies are needed to investigate the impacts of drought on the yields of other crops.