气象产品的广义联合偏差修正方法及其应用

卫星气象产品、气候模式预测数据通常与地面观测数据存在偏差，为保证数据的可靠性和合理性，需要对其进行偏差校正，但偏差校正过程往往受具体区域气象特征、方法本身假定等因素的影响，导致修正效果不佳。为此，本文提出一种广义联合偏差修正方法，相较于现有研究最常用的单变量QM修正方法以及固定多变量修正顺序的JBC修正方法，该方法充分考虑到流域尺度降水和气温双变量的时空相关性，并结合其对径流的主导作用对变量修正顺序进行动态调整，实现了QM法和JBC法的优势互补。在澜沧江-湄公河流域的应用表明：考虑降水、气温相关性可显著改善降水和温度极值的修正效果，尤其是5、6月份，修正后气象与实测数据的纳什系数提升了0.5以上；考虑气象要素的修正次序显著降低了修正后的降水和温度频率分布及均值偏差；利用修正后的气象数据驱动分布式水文模型时，部分月份的径流模拟精度提升了54.3%。

Generalized joint deviation correction method of regional climate model and its application

There are still deviations between meteorological products and climate model prediction data and ground observation data, which usually need to be corrected to ensure data reliability; The correction effect of commonly used deviation correction methods is affected by regional characteristics and meteorological elements. For areas covering multiple climate zones, large spatial heterogeneity and comprehensive influence of multiple meteorological elements, the effect of single deviation correction method is not ideal. Therefore, a generalized joint deviation correction method is proposed in this paper. According to the temporal and spatial correlation between rainfall and temperature and their effect on regional hydrological process, The quantitative mapping (QM) method and joint bias correction (JBC) method are coupled for joint correction. The method is applied to the Lancang Mekong River Basin, and the results show that Compared with the (QM) method, the generalized joint deviation correction takes into account the correlation between precipitation and temperature, improves the correction effect of precipitation and temperature extreme values, and the Nash coefficient increases significantly, especially in May and June, the Nash coefficient increases by more than 0.5; compared with joint bias correction Compared with (JBC) method, this method considers the dynamic relationship between meteorology and hydrology, reduces the deviation of precipitation and temperature frequency distribution and mean value, and makes the modified data distribution closer to the measured data distribution; when the modified meteorological data is used to drive the distributed hydrological model, the accuracy of runoff simulation is improved by 91.2%; on this basis, the modified meteorological data of this method is applied to the future meteorology and hydrology of the basin forecast. The generalized joint deviation correction method realizes the complementary advantages between traditional methods, extends the scope of application, improves the accuracy of meteorological data correction, and provides a reliable basis for future runoff prediction and water resources planning and management of subsequent basins.