利用地形高度差对ERA5气温降尺度的方法研究

A methodological study of Downscaling of ERA5 Temperature Using Terrain Height Difference

基于ERA52m气温数据(T_(ERA)),考虑地形高度对气温的影响,结合温度垂直递减率,开展了气温降尺度研究。利用广西91个国家观测站2021年1月—12月逐小时气温数据(T_(sit))来验证该方法降尺度后气温(T_(air))精度,对比探讨降尺度效果,并讨论T_(air)误差时空分布特征。结果表明:(1)全年T_(sit)与T_(air)有效匹配数据点的相关系数(R)为0.978,均方根误差(RMSE)为1.59℃,平均偏差(MB)为0.441℃。比T_(ERA)与T_(sit)的相关性高(R=0.974)、均方根误差小(RMSE=1.71℃)。(2)T_(air)细化地形对气温的影响,展示了很好的空间分布细节纹理,特别是海拔较高区域。(3)T_(air)误差有明显的时空变化特征,桂南精度高(RMSE<1.5℃,|MB|<1℃),桂东北、桂西北精度相对较低(RMSE>1.8℃,|MB|>1℃),秋季的精度(RMSE=1.54℃,MB=0.35℃)优于其他季节(RMSE>1.58℃,MB>0.38℃)。

Based on the 2m temperature data(T_(ERA))of ERA5,the temperature downscaling research was carried out by considering the influence of terrain height on temperature and combining it with the temperature lapse rate.The hourly temperature data(T_(sit))of 91 national observation stations in Guangxi from January to December 2021 were used to verify the accuracy of the method after temperature downscaling(T_(air)),to compare and explore the effects of downscaling,and to discuss the spatiotemporal distribution characteristics of T_(air)errors.The results show that:(1)The correlation coefficient(R)of annual between T_(sit) and T_(air)effectively matched data points was 0.978,the root mean square error(RMSE)was 1.59℃,and the average deviation(MB)was 0.441℃.The correlation was higher(R=0.974)and the root mean square error was smaller(RMSE=1.71°C)than those of T_(ERA)with T_(sit).(2)T_(air)refined the impact of terrain on temperature,showing a good texture of spatial distribution details,especially in higher altitude regions.(3)T_(air)error showed obvious spatiotemporal variation characteristics,with high accuracy in southern Guangxi(RMSE<1.5℃,|MB|<1℃),relatively low accuracy in northeastern and northwestern Guangxi(RMSE>1.8℃,|MB|>1℃).The accuracy in autumn(RMSE=1.54℃,MB=0.35℃)is better than that in other seasons(RMSE>1.58℃,MB>0.38℃).