基于GRAPES的气象-水文模式在淮河流域的一次试验

Application of GRAPES Meteorological-hydrological Coupled Model in Huaihe River

Global-regional assimilation and prediction system(GRAPES)模式为中国气象局于2000年开始组织研究开发的数值预报系统,GRAPES_Meso模式是其区域中尺度数值预报系统版。GRAPES_Meso中的陆面模式选取的是Noah-Land Surface Model(NOAH-LSM)模式。NOAH-LSM陆面模式选取Simple Water Balance(SWB)对陆面水文过程进行描述。SWB是一个简单水量平衡模型,不能完整地描述陆面水文过程,特别是对径流的模拟存在不足。随着GRAPES_Meso模式不断的发展,对预报能力的要求逐渐提高,对其陆面模式中产流过程的描述也需要进一步的研究。本文中所应用的改进陆面水文过程的NOAH-LSM陆面模式,借鉴了水文模型的思想,利用蓄水容量曲线描述单元网格内产流面积的变化,并在模式中增加Muskingum汇流模块,完整陆面模式水文循环。将原GRAPES模式和改进陆面水文过程的GRAPES模式分别与新安江模型进行单向耦合,选取模式TS评分相差不大的试验流域——淮河流域上游控制站王家坝站,进行流量模拟对比。从试验结果可以看出,改进后的模式在洪量相对误差、洪峰相对误差、确定性系数上均优于原模式,并为未来利用水文模型对降水落区检验进行了探索性研究。

Global-regional assimilation and prediction system （GRAPES） model is developed by China Meteorological Administration from 2000. GRAPES_Meso model is used in regional mesoscale nu- merical forecast system. Noah-Land Surface Model （NOAH-LSM） is chosen by GRAPES_Meso for land surface model, in which Simple Water Balance （SWB） is used for land surface hydrological process. There is obviously insufficient in the description of the hydrological process, which is not a complete de- scription of the hydrological cycle, especially in the simulation of runoff. The reason is that SWB is a simple water balance model. With the development of GRAPES_Meso model, the requirement to the ability of the prediction is increasing. So the land surface model also need the further research. In this paper, NOAH-LSM model has been improved by the storage capacity curve and Muskingum module, in order to consider the change of runoff-yield and complete the land surface water cycle. The improved NO- AH-LSM of GRAPES model and GRAPES model are all coupled with Xin＇ anjiang model. From the experiment result, we can see that though the TS score is almost same, the improved one performs better than the origin in relative error of flood volume, relative error of flood peak and Nash coefficient. It provides the exploratory research on the precipitation verification by hydrological model.