| The Tibet Plateau,with an average altitude of over 4km,is the highest plateau in the world and is known as the“third pole”of the earth and the“water tower of the world”.Changes in ground water reserves on the Tibet Plateau have an important impact on the ecological restoration,agricultural and livestock development,geological disaster prevention and control,engineering design and geothermal development of the plateau.Effective monitoring and analysis of groundwater storage anomalies is the focus of research in geology,geography,hydrology and sustainability science.The traditional reliance on ground observation wells to monitor groundwater changes is relatively time-consuming and labour-intensive,and the network of stations is not sufficiently dense.Groundwater results based on hydrological modelling calculations are severely constrained by hydrogeological parameters and the completeness of measured data.With the advent of the Gravity Recovery and Climate Experiment(GRACE)satellite,a new way of monitoring groundwater storage changes on a large scale has become available.However,the 1°resolution of GRACE satellite data is coarse,and the ability to monitor groundwater storage variables at relatively fine scales needs to be improved.In addition,the analysis of spatial and temporal characteristics and attribution of groundwater storage variables and mesozoic factors needs to be enhanced.Therefore,it is of great theoretical and practical importance to carry out research on groundwater storage variables and attribution on the Tibetan Plateau from downscaled GRACE data.In this paper,based on the analysis of the change characteristics of groundwater resources on the Tibetan Plateau,we combine GRACE gravity satellite data and Global Land Data Assimilation System(GLDAS)data,construct a downscaling method coupled with B-G segmentation algorithm and multi-stage regression,improve the spatial resolution of GRACE groundwater storage variables and conduct spatial and temporal variation analysis.The spatial resolution and spatial-temporal variation of GRACE groundwater storage variables are analyzed,multi-dimensional analysis of groundwater storage variables on the Qinghai-Tibet Plateau is carried out,attribution analysis of groundwater storage variables in typical watersheds on the Qinghai-Tibet Plateau is carried out using a gray correlation model,and a deep learning-based groundwater storage variables prediction model is constructed.The main work and research results of this paper are as follows.(1)Analyzing the characteristics of changes in groundwater resources on the Qinghai-Tibet Plateau in the past 20 yearsBased on the collection of relevant data from 2002 to 2020,the change characteristics of the underground water resources volume on the Tibetan Plateau in the past 20 years and its correlation with precipitation and temperature are analyzed.The overall trend of underground water resources in Tibet Autonomous Region is decreasing first and then increasing,and the underground water resources in Qinghai Province is increasing,and precipitation is the main influencing factor of the change of underground water resources in Qinghai-Tibet Plateau.(2)Spatial and temporal changes of groundwater storage variables based on the downscaling of GRACE dataA phased statistical downscaling model is proposed in this paper,which can effectively improve the downscaling effect and increase the GRACE resolution from1°to 0.1°.The spatial resolution of the data after downscaling is significantly improved,and the overall correlation(R~2>0.77)is high when spatial sampling is compared,and the correlation of the data before and after downscaling in time series exceeds 0.97,which almost completely maintains the time series variation characteristics of GRACE data.The downscaled groundwater storage variables of GRACE are verified by using the groundwater level observation borehole data of Lhasa station,and the two have good correlation.There are obvious spatial differences in groundwater storage variables on the Tibetan Plateau from 2002 to2020,and there are obvious changes around 2010,with the groundwater storage variables in the western part of the Tibetan Plateau and the eastern part except the Yangtze River basin showing a decreasing trend,among which the Yarlung Tsangpo River basin has the most obvious decreasing trend,reaching-13.84 mm/a,while the groundwater storage variables in the Inner Current,Qaidam and Yangtze River basins show an increasing trend.The groundwater storage variables in the Qinghai-Tibet Plateau show a large intra-annual oscillation,with the lowest values in July and August and the highest values in March and June.(3)Multi-factor and multi-dimensional analysis of groundwater storage variables on the Tibetan PlateauThe linear tendency of rainfall and air temperature is 1.889mm/a and 0.0343℃/a,but the increasing trend is not significant,while the evapotranspiration,NDVI and nighttime light index show a significant increasing trend(0.0953mm/a,0.0008/a and0.0086/a);generally speaking,the snow depth and groundwater storage variables at different scales show fluctuating decreasing trends,with linear tendency of-0.0193cm/a and-0.4823mm/a,respectively;the rainfall in the Yangtze River region and the Yellow River region has a significant increasing trend,while the decreasing trend in the Ganges River region is the most obvious.The groundwater storage variables have quasi-6.3 and 9.5 year cycles on the interannual scale and quasi-3.8and 9.5 year cycles on the NDVI degree.And the joint analysis of groundwater storage variables and related factors based on Copula was conducted.The probabilities of joint occurrence of events of groundwater storage variables and influencing factors such as rainfall,temperature,evapotranspiration,and snow depth were in the range of 0.2-0.5,0.3-0.5,0.2-0.5,0.1-0.5,and 0.2-0.5,respectively.(4)Attribution and prediction of groundwater storage variables in typical basins on the Tibetan PlateauThe monthly-scale variation trends and main drivers of groundwater storage variables in the Yarlung Tsangpo River basin are analyzed,and the main drivers affecting the variation of groundwater storage variables are vegetation cover and evapotranspiration.Then the establishment process of ARIMA model,BP neural network model,and LSTM model was introduced.Combining the models,the LSTM model with multivariate input is optimal,and the NSE coefficients of the training and test sets are 0.96 and 0.75,respectively,which both achieve excellent simulation results. |
PDF Full Text Request