| 摘要: |
| 【目的】为了弥补地面观测辐射数据的不足，获取高精度的太阳辐射空间分布资料。【方法】利用贵州2018年3月1日—2020年4月30日7个地面辐射站逐日太阳辐射资料，分月建立了基于BP神经网络的FY-4A SSI产品订正模型，利用该模型对FY-4A总辐射进行订正与分析，并把订正结果和采用一元线性回归模型所得到的结果进行了对比。【结果】结果表明：（1）FY-4A反演总辐射值存在对“低值辐射高估、高值辐射低估”的偏差分布特征，整体上FY-4A总辐射值偏高；（2）FY-4A总辐射值与站点值的相关系数R在0.7670.926之间，平均绝对误差MAE在5.046.98 之间，平均误差ME在4.436.68 之间，均方根误差RMSE在5.707.44 之间；（3）采用线性方法订正后两者的R不变，MAE在1.453.09 之间，ME在-0.250.42 之间，RMSE在2.124.70 之间；（4）采用BP神经网络方法订正后两者的R在0.8560.962之间，MAE在1.022.43 之间，ME在-0.480.23 之间，RMSE在1.363.77 之间。【结论】因此，基于BP神经网络模型订正后的FY-4A总辐射值与站点值之间的误差较小，订正精度高于线性订正，有较好的稳定性，可适用于贵州高原山区FY-4A地面太阳辐射产品的订正。 |
| 关键词: 太阳辐射；FY-4A-SSI；BP神经网络；一元线性回归；订正 |
| DOI： |
| 投稿时间：2024-06-06修订日期：2024-07-30 |
| 基金项目:贵州省气象局科研业务项目（黔气科登[2024]04-05号）：FY-4A地面太阳辐射产品在贵州山区的本地化订正研究 |
|
| Research on the Application of BP Neural Network Algorithm in FY-4A-SSI Product Correction |
| linxuefei,zhujun,tianpengju,liguangyi,lifengdan,hujingjing |
| (Zunyi Meteorological Bureau;Guizhou Climate Center (Climate Change Center);Guizhou Ecological Meteorology and Agrometeorology Center;Wenzhou Air Traffic Management Station) |
| Abstract: |
| In order to compensate for the shortcomings of ground observation radiation data and obtain high-precision solar radiation spatial distribution data, a monthly FY-4A SSI product correction model based on BP neural network was established using daily solar radiation data from seven ground radiation stations in Guizhou from March 1, 2018 to April 30, 2020. The model was used to correct and analyze the total radiation of FY-4A, and the correction results were compared with those obtained using a simple linear regression model. The results show that: (1) FY-4A inversion has a biased distribution characteristic of overestimating low value radiation and underestimating high value radiation, and overall, the total radiation value of FY-4A is relatively high; (2) The R between the total radiation value of FY-4A and the station value is between 0.7670.926, the mean absolute error MAE is between 5.046.98 , the mean error ME is between 4.436.68 , and the root mean square error RMSE is between 5.707.44 ; (3) After linear correction, the R of the two remains unchanged, with MAE between 1.453.09 , ME between -0.250.42 , and RMSE between 2.124.70 ; (4) After using the BP neural network method to correct, the R of the two is between 0.8560.962 , MAE is between 1.022.43 , ME is between -0.480.23 , and RMSE is between 1.363.77 . Therefore, the error between the total radiation value of FY-4A corrected base on the BP neural network model and the station value is small, and the correction accuracy is higher than linear correction. and it has good stability, which can be applied to the correction of FY-4A surface solar irradiance products in mountainous areas of Guizhou Plateau. |
| Key words: Solar radiation; FY-4A-SSI; BP neural network; simple linear regression; correction |
