基于Forward-Modeling方法的黑河流域水储量变化特征研究

黑河流域陆地水储量变化对流域下游等周边区域水资源的合理利用以及经济和社会发展等有着重要的意义.本文利用2003年1月至2013年12月的GRACE RL05数据反演了黑河流域陆地水储量长时间序列的变化,并针对重力场模型和数据处理中产生的信号泄漏问题,采用Forward-Modeling方法进行了改正并恢复泄漏信号;将GRACE获得的泄漏信号恢复前后的黑河流域水储量变化结果与全球水文模型GLDAS和CPC进行比较分析,结果表明泄漏信号改正后的结果与水文模型结果的时间序列相关性均有明显提高,从其空间分布结果可以看出Forward-Modeling方法有效地恢复初始信号、增强被湮没的信号,泄漏信号误差减小;通过分析黑河流域水储量变化的长时间序列结果,发现其具有明显的阶段性变化特征,即2003—2006年呈明显下降趋势,约为-0.86cm·a-1,在2007—2010年趋于平衡状态,而2011—2013年则呈现缓慢上升趋势约为0.14cm·a-1;联合GRACE数据和GLDAS数据反演了黑河流域地下水储量变化,并与全球降雨数据GPCC进行了比较分析,两者相关性可达到0.88以上.

Investigation of water storage variation in the Heihe River using the Forward-Modeling method

The Heihe drainage is located in the Gansu Corridor and the middle Qilian Mountains. It is a typical inland river basin and water-lack area in Northwest China. Study of storage variation in the Heihe drainage is essential to the rational utilization of downstream basins and the development of economy in this region. With the appearance of GRACE (Gravity Recovery and Climate Experiment) gravity satellite mission, it is possible to quantitatively study the terrestrial water storage (TWS) at large or medium scales using satellite data. Simulation research indicates the original signal leakage when the spherical harmonic order truncation and filter methods are employed in TWS computation on the GRACE model. The signal distortion is inevitable due to the signal annihilation, which limits the accuracy of GRACE model application. The latest Forward-Modeling method can restore the original signal, enhance the annihilation signal and decrease the error. This work inverts the time series variation of the water storage capacity of the Heihe River using the GRACE RL05’s model between January and December 2003. The Forward-Modeling method is mainly used to improve the inverse results. And then a precise analysis and its geophysical interpretation are conducted to reveal variation characteristics of spatial and temporal distribution, compared with GLDAS (Global Land Data Assimilation System) and CPC (Climate Prediction Center). The results draw three major conclusions. (1) The average magnitude of corrected leakage signal is close to 1 cm (8 mm), which is the major considered correction in TWS computation using the GRACE model. It can effectively restore the original signal, enhance the annihilation signal and improve the spatial resolution on water storage variation using the Forward-Modeling method. (2) Analyzing the long-term variation of water storage capacity in the Heihe basin region during 2003—2013 shows a significant characteristic of phase change. The water storage declined significantly between 2003 and 2006 at a rate of 0.86 cm·a^-1, while tended to an equilibrium state between 2007 and 2010, and raised slowly at about 0.14 cm·a^-1 between 2011 and 2013. The annual seasonal change of the Heihe basin region is close to the hydrologic model value inversed from GLDAS data and CPC, with the correlation coefficient 0.64. It shows that the water storage rises to the maximum surplus around August or September, and decreases to the maximum loss around April or May. (3) Rainfall is the major source of groundwater recharge in the Heihe basin region, with a consistent correlation up to 0.88, by comparison between the water storage variation from GRACE and monthly rainfall variation from GPCC (Global Precipitation Climatology Centre). The results show consistency in annual seasonal change and close correlation in amplitude and phase from the Forward-Modeling method and GPCC model value. It also implies that the rainfall decreases in autumn and winter with the maximum loss in February, and increases gradually in spring and summer with the maximum surplus in July.