基于机器学习订正ERA5的甘肃省地表太阳辐射时空分布
吴欣华, 王思晨, 王菲菲, 王天河, 杜源, 陈涛, 牛亮亮, 赵怀宇, 张昊天

Spatiotemporal distribution of surface solar radiation in Gansu Province based on machine learning correction of ERA5
WU Xinhua, WANG Sichen, WANG Feifei, WANG Tianhe, DU Yuan, CHEN Tao, NIU Liangliang, ZHAO Huaiyu, ZHANG Haotian
图1 未订正及不同机器学习方法订正的2022—2024年ERA5逐小时地表下行太阳辐射与地面站点观测值的散点图
(a)未订正,(b)LGBM订正,(c)RF订正,(d)SVM订正
(黑色实线为1∶1基准线,灰色阴影表示高斯核密度估计值)
Fig.1 Scatter plots of hourly ERA5 surface downward solar radiation against in-situ observations before correction and corrected by different machine learning methods during 2022-2024
(a) Uncorrected,(b) LGBM-corrected,(c) RF-corrected,(d) SVM-corrected
(The black solid line indicates the 1∶1 reference line,and the gray shading represents the gaussian kernel density estimate)