China daily ground surface direct and diffuse solar radiation dataset

This dataset is produced based on a machine learning model. The satellite-derived surface global solar radiation dataset ISCCP-ITP-CNN was fed into the trained model and a daily direct and diffuse radiation component dataset for China was produced with a spatial resolution of 10 km and a time span of 1983-2018. This dataset has been validated using the surface solar radiation component observations China Meteorological Administration (CMA). It has higher accuracy compared with the rest of the current datasets.The data format is mat file, which is generated by the scientific computing software MATLAB and can be read using either matlab or Python. The diffese radiation file is named 'dif_year', the storage variable is 'dif2', and the data are in the form of 400*700 arrays, representing 70-140° E and 15-55° N, with a resolution of 0.1 degrees, i.e., a grid of 10km. The direct radiation file is named 'dir_year', the storage variable is 'dir3', and the data form is the same as that of the scattering radiation. No latitude and longitude information is stored in any of the files.

本数据集基于机器学习模型构建。将卫星反演地表全球太阳辐射数据集ISCCP-ITP-CNN输入至训练完成的模型后，生成了覆盖中国区域的日尺度直接辐射与散射辐射分量数据集，其空间分辨率为10 km，时间跨度为1983年至2018年。本数据集已通过中国气象局（China Meteorological Administration，CMA）的地表太阳辐射分量观测数据完成验证，相较于当前其他同类数据集，具备更高的精度。该数据集采用科学计算软件MATLAB生成的MAT文件格式，可通过MATLAB或Python读取。其中散射辐射文件命名为"dif_year"，存储变量为"dif2"，数据以400×700数组形式存储，对应空间范围为东经70°至140°、北纬15°至55°，空间分辨率为0.1°，即单网格尺度为10 km。直接辐射文件命名为"dir_year"，存储变量为"dir3"，数据格式与散射辐射文件完全一致。所有文件均未存储经纬度信息。