In general,China is short of water resources and some regions even experience a shortage of daily water supply.This could threaten the stability and economic development of the nation.A study on the water storage variations is especially important for the water management and storage prediction in three largest river basins of China,namely,Yangtze,Yellow,and Zhujiang,where the most dense population and leading economic regions are located.The satellite gravity mission GRACE(Gravity Recovery and Climate Experiment) provides an opportunity to macroscopically identify water(or mass) variations in the Earth's system with a spatial resolution of 300-400 km and a temporal resolution of about one month.We use the first release of the DEOS(Delft Institute of Earth Observation and Space Systems) Mass Transport(DMT-1) model based on GRACE data to analyze water storage changes in the three river basins.The DMT-1 model consists of monthly solutions,which are computed using an innovative methodology.The methodology includes,in particular,the application of a statistically optimal Wiener-type filter based on full variance-covariance matrices of noise and signal.This results in particularly sharp mass variation maps.Taking one monthly solution as an example,we compare the results derived from the DMT-1 model with ones produced with the standard post-processing scheme based on a combination of the de-striping and Gaussian filtering.The comparison shows that the DMT-1 model outperforms the other models and is suitable for the analysis of the mass changes in river basins.A subset of the DMT-1 solutions in the interval between February 2003 and May 2008 is used to estimate the secular trends and seasonal variations for the three river basins.The estimated trends show that the water storage of the Yellow River basin does not have significant changes,while the Zhujiang and Yangtze river basins have a large and statistically significant water storage increase.The estimation of seasonal variations demonstrates that the water storage va
