基于FY-2G总云量的日照百分率估算及检验

Estimated and Tested of Sunshine Percentage Based on Total Cloud Amount Form FY-2G

摘要: 首先以考虑地形遮蔽的分布式可照时间理论模型结合气象台站观测的日照时间计算了日照百分率，然后对我国第二代静止气象卫星(FY-2G)的总云量遥感影像产品重采样后，根据日照百分率和云量的负相关性，分别建立了基于遥感总云量和观测站云量(总云量、低云量)的日尺度日照百分率估算模型，并以气象站点分布稀疏的新疆区域为例，对两种估算模型使用IDW和Kriging插值方法的模拟效果进行了检验，得出以下结论：1) 实际地形下新疆区域可照时间的空间分布受地形影响较大，全区四季平均可照时间分别为：春季1165 h，夏季1286 h，秋季964 h，冬季823 h。2) 基于遥感云量的条带状重采样方案考虑了日照轨迹和云的区域移动变化，重采样后的云量值与日照百分率的相关性有所提升，相关系数为0.756。3) 建立的单站分季节遥感集成日照百分率模型相关系数冬夏差异明显，夏季最高，春秋次之，且分布形态较为一致，冬季最低，低值主要集中在北疆沿天山一带。4) 从模拟的效果来看，遥感集成日照百分率模型(平均绝对误差为14.8%)要明显优于基于观测站云量模型模拟的结果，由于遥感集成日照百分率充分发挥了卫星在空间上连续观测的优势，空间分布更为连续，通过检验后的该方法可以在站点稀少的西部地区进行业务应用。

Abstract: The sunshine percentage was calculated using the theoretical model of the distributed exposure time considering the terrain factors and the sunshine hours observed by the meteorological station. After re-sampling the total cloud amount remote sensing image products by the second generation geostationary meteorological satellite (FY-2G) in China, the daily scale sunshine percentage estimation model was established based on remote sensing total cloud amount and observation station cloud amount (total cloud amount and low cloud amount) used the negative correlation between cloud cover and sunshine percentage. The spatial simulations are completed of the two interpolation methods IDW and Kriging in Xinjiang. Finally, simulation results are verified by the test station. The conclusions are as follows: 1) The spatial distribution of the four seasons in Xinjiang has obvious changes. The average seasons in the whole region are 1165 h in spring, 1286 h in summer, 964 h in autumn and 823 h in winter. 2) The resampling scheme based on remote sensing cloud cover takes into account the cloud movement. The correlation was improved after resampling between cloud cover and sunshine percentage, and the correlation coefficient was 0.756. 3) The correlation coefficient of sunshine percentage model based on Remote Sensing has obvious differences in the two seasons in winter and summer; the highest correlation was found in summer; the distribution pat-terns were similar in spring and autumn; the correlation coefficient was the lowest in winter and represented along the Tianshan Mountains in the northern Xinjiang. 4) The sunshine percentage accuracy of the RS integrated model is obviously better than that based on the observation station cloud cover model. The mean absolute error of the sunshine percentage was 14.8%. As that gives full play to the superiority of the continuous observation of satellites in space, the spatial distribution is more continuous. The post-verification method can be used for business applications in the sparsely stationed western region.