公开免费的地球重力卫星数据已在全球范围内区域地下水储量评估方面获得了大量成功应用，然而，其较低的空间分辨率为该反演技术深入推广应用带来了困难。针对该问题，本项目拟采用多源遥感数据比较、多元统计、神经网络、地下水数值模型、陆面水文模型和参数自动估计与优化方法，研究多源数据的有效性和可靠性，建立地球重力卫星反演数据与地下水储量变化关键因子之间的关系，再利用以地下水储量为主变量的地下水模型和陆面水文模型研究地球重力卫星反演数据的同化方法，重点解决地球重力卫星反演数据从区域到含水层或更小尺度转换的问题并探讨其适用性，从而捕获小尺度地下水储量变化规律，最终形成一套地球重力卫星反演数据的统计和动力的降尺度方法，结果将更能辅助管理者准确获知不同尺度地下水资源的问题，从而为有效合理的地下水资源管理提供科学依据和决策支持。

Pubic and free GRACE data have been successfully applied in the evaluation of regional groundwater storage changes all over the world, however, low resolution of the data have been hindered the wider application. Based on the fundamental problems, the project will integrate a series of methods, such as multi-source data, multi-dimensional analysis, artificial neural networks, groundwater models, land hydrology models, parameter automatic estimation and optimization. The research will try to carry out studies on the effectiveness and reliability of multi-source data, the establishment of the relationship among GRACE data and the key factors affecting the changes of groundwater storages, and the assimilation of GRACE-derived data using groundwater models with the primary variable of groundwater storage and improved land hydrology models. Finally the downscaling method of GRACE data will be developed from regional scales to small scales or aquifer scales, and thus the groundwater storage changes at small scales will be captured. A comprehensive method about the statistical and dynamical downscaling method will be established. The findings and methods will aid water managers in better understanding the changes of groundwater storages, and provide technical support of effective and reliable groundwater resources management.