GRACE And GLDAS Data-based Estimation Of Spatial Variations In Terrestrial Water Variations Over Xinjiang

In recent years, along with the high development and utilization of water resources, the unbalance distribution of it has aggravate. It has induced a series of eco-environmental problems, such as the reduce of biodiversity and the deteriorate of salinization, desertification, water and salt migration imbalances. In the 1970 s, the rise of gravity satellite provided a new method to abtain the global climate change and water cycle information. It not only solved the limitation of traditional hydrologic method in acquiring the large scale hydrological data, but also offered a new thinking for monitor the whole hydrological cycle effect and its partial response parameters.In this study, years of global GRACE data were used, assisted by TRMM precipitation data, DEM, river network distribution data, NDVI, evapotranspiration, runoff data, terrestrial water storages were inversed. Thus, we analyzed the time and space dynamic change trend of terrestrial water storages in 11 years and the different changes in seasons, discussed the amplitudes and phase variations of terrestrial water storages. Take human factors and natural factors into consideration, the driving factors of terrestrial water storage changes were analyzed.The distributions of NDVI in 11 years and their change trend were discussed with the changes trend of terrestrial water storages and precipitation, the correlations between terrestrial water storages, precipitation and NDVI were investigated. We simulated the total surface water changes（including the soil moisture content changes, snow water equivalent, canopy water storage and surface runoff changes） using GLDAS hydrological model NOAH, compared the soil moisture content changes with terrestrial water storage changes in seasons and their temporal and spatial variations, figured out the groundwater quantity changes, discussed the groundwater quantity change trends in seasons. The groundwater level mathematical model of middle reaches of the keriya river basin was roughly established, with the measured data of the groundwater level as the dependent variable, the soil moisture contents and their changes simulated by GLDAS, the terrestrial water storage and groundwater level changes as the independent variables, selected the optimal mathematical manipulation. The soil salinity, soil moisture and groundwater level spatial characteristics in salt deposition period（spring） and salt recurrent period（autumn） were investigated. The main conclusions are as follows:（1）The terrestrial water storages of 2003, 2007 and 2013 in Xinjiang were obvious different in seasons, which had the spring>summer>winter>autumn trend. The equivalent water heights of the terrestrial water storages in 2008 reduced sharply compared with it in 2003. The equivalent water heights of the terrestrial water storages in the spring of 2003 increased gradually from south to north. Those in the spring of 2013 increased from the geographic center of Xinjiang to the exterior, serious losses transformed into earnings. The terrestrial water storages changed more fiercely in the latter 6 years than those in the former 5 years, and the area that changed fiercely was increasing.（2） The main factors cased the disease of terrestrial water storage: ① the reduced precipitation and the evapotranspiration which was apparently higher than the precipitation; ②the unbalanced River network distribution with a lack in the west made the surface runoff decreased; ③the drying up rivers because of the water overuse on both sides of it; ④the waste and loss of water resources caused by the industrial water; ⑤the reducing groundwater storages that was induced by the use of domestic water pumped from the underground.（3）TRMM precipitation showed an increasing trend in large area and river network distributed intensively, they were the main natural factors which made the terrestrial water storage increased apparently. Meanwhile, the main artificial factors were optimizing water channels, decrease in the quantity of water resources loss, reservoirs which stored water to irrigate, making full use of river water and avoiding the waste of terrestrial water.（4）Annual period values and GRACE terrestrial water storages had the consistent fluctuation tendency, verified the calculation accuracy of annual period values. The maximum of terrestrial water storages presented a hysteretic nature compared with the maximum rainfall, the lag time was 4 months.（5） NDVI showed a summer > autumn > spring > winter trend, it increased as the distance to the river decreased. There was a positive correlation between precipitation and NDVI, the correlation coefficients had a winter < spring < summer < autumn, precipitation had a more obvious influence on NDVI than the terrestrial water storages.（6） GLDAS oil moistures had a strong consistency with GRACE terrestrial water storages inversed, showed that the soil moisture was one of the key factors which could influence and represent the terrestrial water storage. Among the 11 years, although the soil moisture and terrestrial water storage values fluctuated sharply, it was obvious that they both risen and declined promptly and then decreased slowly.（7） Changes of surface runoff, canopy water storage and snow water equivalent were minimal, which proved the change of groundwater could be calculated by terrestrial water storage and soil moisture. The sequencing of correlations with the measured groundwater level was GLDAS soil moisture> groundwater storage change> GRACE terrestrial water storage change > TRMM precipitation. There was a negative correlation between the groundwater storage change and groundwater level, the TRMM precipitation and groundwater level were almost uncorrelated, others were positively correlated with the groundwater level.（8） The fitting equations of groundwater level with GLDAS soil moisture and its change, GRACE terrestrial water storage change, groundwater storage change were based on linear model, the multiple stepwise regression model Y=0.107X₁+0.003X₂+0.126X₃-0.109X₄+0.54 was slightly superior to the linear function.（9） There are differences between the most estimated and the measured value, but the overall trend is higher consistency. There are some outliers, the model can simulate groundwater level data roughly, only for middle reaches of Keriya river field.In conclusion, data from GRACE gravity satellite and GLDAS hydrological model provide a new thinking for water resources migration study. Although, the groundwater level estimation model established in this study has boundedness, cannot be pervasive in the whole basin, it can present the trend apparently: groundwater levels are shallow in the keriya river basin and the area closed to the oases, they are much deeper in the area which has a big distance to the river and oases. This study provides a scientific basis for water resources rational allocation and utilization, lays the foundation of further study on the water and salt transport.