利用多元线性回归模型重构中国九大流域陆地水储量变化北大核心CSCD

GRACE与GRACE-FO任务间的数据空缺导致无法连续监测陆地水储量变化。基于此,本文采用多元线性回归模型,以GRACE/GRACE-FO陆地水储量变化数据为参考值,以降水、气温和模型模拟的陆地水储量数据为预测参数,采用3种不同策略重构中国九大流域2002-04~2021-12连续的陆地水储量变化。结果表明,基于去趋势项和去季节项信号重构策略的重构结果略优于去趋势项信号重构策略,且两者结果均优于整体信号重构策略,在人类活动或冰川融化频繁的流域(如海滦河、长江、西南诸河和内陆河流域)这种优势更为明显。此外,重构结果的性能也受GRACE/GRACE-FO数据信噪比和预测参数与GRACE/GRCAE-FO数据的相关性影响。

Reconstruction of Terrestrial Water Storage Changes of Nine Major River Basins in China Using a Multiple Linear Regression Model

Data gaps between the GRACE and GRACE-FO missions have prevented continuous monitoring of the terrestrial water storage changes (TWSC). We use the multiple linear regression model, take the GRACE/GRACE-FO solutions as reference values, the precipitation, air temperature, and model-simulated terrestrial water storage data as prediction parameters, and consider three different strategies to reconstruct the continuous TWSC of nine major river basins in China from April 2002 to December 2021. The results show that the reconstructed results based on the de-trended and de-seasonalized signal reconstruction strategy are slightly better than those dependent on the de-trended signal reconstruction strategy. These two reconstructed results are better than those using the overall signal reconstruction strategy, and this advantage is more evident in the basins with intense human activities or glaciers melt (such as Haihe and Luanhe rivers, Yangtze river, southeast rivers, and inland river basins). In addition, the performance of reconstructed result is also affected by the signal-to-noise ratio of GRACE/GRACE-FO data and the correlation of prediction parameters with GRACE/GRACE-FO data.