| Water is an irreplaceable resource for the survival of all living things,and it is an important controlling factor of ecological environment cycle.Monitoring and quantifying terrestrial water change by various methods is of great significance for local governments and country to prevent drought and flood,formulate water resources management policies,and analyze climate change.The Slepian basis function is the approach that is capable of transforming the geophysical signals into spatial spectrum information and obtaining the optimal spectral smoothing solution on the spherical surface,which is suitable to study the geophysical signal at a regional spatial scale.The goal of this paper is to apply the 60-order Slepian basis function,combine with the continuous GPS observation data at 72 Crustal Movement Observing Network of China(CMONOC)sites to analyze the characteristics of water load deformation and the temporal and spatial distribution patterns of terrestrial water storage changes in the Sichuan,Yunnan and Chongqing from 2011 to 2015.Before the inversion,In order to ensure that there are enough GPS sites participating in the inversion at each time,we selects the Kriging Kalman filter model to recover the missing data.For testing the effect of different expanded boundary on the quality of terrestrial water storage,the gridded GLDAS data were employed to calculate the vertical loading displacement time series at each site,then they were used to recover the water storage change.Results show that the terrestrial water storage change can be well reproduced by GLDAS data when the boundary is expanded up to 3 degrees.The first 31 terms of Slepian basis functions with the maximum order of 60 and γ>0.1are used to transform the vertical displacement time series of discrete GPS sites into Slepian displacement spatial spectrum.The amplitude of the converted displacement time series is smaller than that of the original time series,and the converted time series is smoother.The root mean square value of the difference between the two is less than 5mm.Comparing the displacement time series converted by the Slepian function with the load displacement predicted by GRACE and GLDAS,it is found that they all have similar characteristics of seasonal and interannual variation,which indicates that the Slepian displacement time series reflects the seasonal hydrological load variation.The results show that the spatial spectrum of Slepian displacement retains the seasonal characteristics of the original time series,and can be used for terrestrial water storage inversion.In this paper,we calculated the annual amplitude of terrestrial water change in each year from 2011 to 2015 in the region,and found that the characteristics of terrestrial water changes in Sichuan,Yunnan,and Chongqing are roughly the same in the past five years.The change is larger in 2012,with the average regional amplitude of 206 mm,and the minimum in 2015,with the average regional amplitude of 164 mm.Comparison of the equivalent water heights(EWH)determined by GPS,GRACE,GLDAS and Precipitation data,result indicates that seasonal precipitation is one of the major driving factors for terrestrial water storage change.The annual variation of terrestrial water storage is large in Yunnan in comparison with that in Sichuan and Chongqing,which peaks at the amplitude of about 30 cm in Southwest Yunnan.Whereas it presents small fluctuations of 7 cm in northern Sichuan and Chongqing.Compared with the GPS-derived values,the GRACE and GLDAS significantly underestimate the seasonal amplitude of terrestrial water storage change by 24% and 47 %,respectively.Analyzing the phases in the region,it is found that the phase is generally larger in Yunnan,and the phase in western Sichuan is relatively small.The first to reach the peak is the precipitation data,with the phase of 180-220 days.Compared with the precipitation data,the GPS results are about one month later,with the phase of 220-270 days.The GRACE and GLDAS results are the latest,with the phase of 220-300 days and 200-320 days respectively.Similar results can be found by comparing the phases of the mean equivalent water height time series in the region.The average EWH time series inferred by GPS data show a good correlation with precipitation data,while temporal variations between precipitation and GRACE-and GLDAS-derived EWH present a time delay of one or two months.In contrast to the GRACE and GLDAS,GPS-derived terrestrial water storage product has a good capability of tracing the regional-scale extreme rainfall as illustrated in the strong rainfall event occurred in South China on January 2015.In this paper,four drought indices,GPS-DSI,GRACE-DSI,GLDAS-DSI and PRECI-DSI are calculated by using the results of terrestrial water storage and precipitation data,and combined with the existing drought indexes SPEI and CI,the drought characteristics of Yunnan were analyzed.The results show that,compared with the meteorological drought index mainly driven by precipitation data,GPS-DSI reflects more diverse hydrological elements and can reflect more real surface drought characteristicsOur results imply that the GPS inversion for water changes based on Slepian basis function succeeds in tracking the high spatial-temporal variations of terrestrial water storage,and it is a useful tool to fill the data gap when GRACE measurements are not available. |
