Spatiotemporal Dynamics Of Large-scale Drought Events In The Yellow River Basin And Driving Mechanism Study

Under the influence of global climate change,the inter-annual variability and uncertainty of hydro-meteorological processes such as precipitation,evapotranspiration,snow accumulation,snowmelt,runoff yield and concentration are gradually increasing,the contradiction between regional water balance is becoming more and more prominent,and the frequency and pattern of droughts have accelerated.Arid and semi-arid regions with complex moisture sources and diverse topography are sensitive areas to climate change and vulnerable to drought intrusion.Most of the Yellow River Basin（YRB）is located in the arid and semi-arid regions of China,with fragile ecological environment and huge carrying economic volume and population.Analysing the occurrence and development pattern of drought in the YRB under changing environment and exploring the driving relationship of climate anomalies on drought in the basin are of great importance to strengthen regional drought monitoring and management and promote ecological protection and high-quality development in the YRB.This study takes the YRB as the study area,constructs the multi-timescale standardized moisture anomaly index incorporating snow dynamics(SZI_snow),the adaptability of SZI_snow in the YRB was assessed by comparing it with the traditional drought index;The SZI_snow was then combined with a three-dimensional dynamic identification algorithm for drought events to track large-scale drought events occurring in the YRB between 1948 and 2021,and the Severity-Area-Duration（SAD）method was used to analyse differences in the intensity of large-scale drought events;Finally,pearson correlation analysis and cross wavelet were used to investigate the teleconnection between SST anomalies,air pressure anomalies and drought in the YRB,and to clarify the role of climate anomalies in driving drought in the basin.The following main results were obtained:（1）Applicability of the drought index SZI_snow in the YRBIn terms of time,the drought dates identified by SPI,SPEI and SZI_snow were closer to each other,and all of them considered that severe droughts occurred in the whole basin around 1992 and 1998.In terms of spatial distribution,the spatial pattern of droughts identified by SZI_snow was closer to that of SPEI,while it was more different from that of SPI.From June 1997 to May 1998,88%of the drought areas identified by SZI_snow were consistent with SPEI results,and 62%were consistent with SPI.In the second half of 1996,SPI and SPEI identified a larger area of drought in the upper part of the basin compared to SZI_snow due to the neglect of the pre-snow process.Therefore,compared with the traditional drought indices,the SZI_snow,which introduces the snow accumulation process,has better performance in drought identification in the YRB and is more capable of meeting the drought identification requirements in moisture deficit areas.（2）Dynamic characteristics of large scale drought events in the YRBDuring 1948-2021,16 large-scale drought events occurred in the YRB.Among them,Drought Events No.928,No.962 and No.1133 are typical representatives of high-intensity,large-area and multicenter drought events in the basin.Drought event No.928,which occurred in October 1998,originated in the east and west of the basin at the same time,and then was spatially linked through the central basin drought zone,and the drought intensity expanded sharply after 1999,with the maximum drought intensity of-3.99,-3.13,-3.13 and-4.11 in the first four months,respectively,all much smaller than the extreme drought threshold;Drought event No.962 occurred in December 2000,and the impact on the basin was relatively small in the first three months,but the drought expanded from the fourth month,controlling the upper and middle reaches of the basin and continued to advance downstream until July 2001,when it triggered a basin-wide drought,controlling 92%of the basin;Drought event No.1133 occurred in December 2012,from the third month,there were extreme arid areas that were not affiliated with each other.As time went on,the number of extreme arid areas further increased and distributed in a band.In addition,among the 16 large-scale drought events,Drought Event No.962,which occurred in December 2000,was stronger than the other drought events in 3-month,6-month,9-month and 12-month time intervals.（3）Climate anomalies driving drought in the YRBFrom 1950 to 2021,SZI_snow in the YRB was significantly positively correlated with the Arctic Oscillation（AO）index,and in the northern and eastern regions of the basin,the correlation coefficient exceeded 0.25;SZI_snow in the YRB was significantly negatively correlated with the Ni（?）o 3.4 index,and in the middle of the basin,the negative correlation coefficient exceeds 0.15.In addition to finding the above correlations,the cross wavelet method also indicates the resonance periods of the remote correlation factors with the basin drought indices in several years.the resonance period of the AO index with the SZI_snow of the YRB is 2-3 years in 1956-1968 and 1year in 2010-2020;the resonance period of the Ni（?）o 3.4 index with the SZI_snow of the YRB is 2-4 years in 1954-1970,1-2 years in 1970-1975,1996-2000 and 2010-2018,and 4-6 years in 1980-2008.In summary,this study used the GLDAS-2 product to construct the drought index SZI_snow,and demonstrated the applicability of SZI_snow to drought assessment in the YRB using the mainstream drought index.During the past 74 years,16 high-impact drought events occurred in the YRB,and the drought event that occurred in December 2000 had the highest drought intensity.In addition,the results show that the drought in the YRB is aggravated when the Arctic Oscillation is in negative phase or when El Ni（?）o occurs,and vice versa.As the drought situation in the YRB is severe,there is a need to construct higher resolution drought indices to accurately analyse the historical droughts in the basin,or to explore the future drought situation in the context of climate change by combining climate model outputs.