The Estimation Of Surface Solar Radiation Considering Cloud Shadow Distortion On Complex Terrain

Surface solar radiation controls exchange process of energy and flux in enrage system, which causes the spatial difference of surface vegetation and the various biological processes. The accurate estimation of DSSR plays an important role in climate prediction, solar energy utilization and the vegetation productivity estimation, etc. Cloud shadow distortion and complex terrain are the main influence factors of radiation estimation by remote sensing, and then make the radiation balance of surface not so accurate. So the research of influence on surface solar radiation by cloud and topography is of great significance.Most research separated the two factors and rarely consider the relationship of cloud-cloud shadow-terrain and the change of cloud shadows under different angles of the sun、satellite and complex terrain. Some research while considered the effect of 3D geometry effect of the cloud, but only on horizontal surface, lack of research on the effect of cloud shadow distortion. In this case, obtaining the location of cloud shadow pixels in complex terrain becomes a question to solve. In order to solve these problems, we used the HJ-1B CCD and IRS data, explored the high-resolution cloud height calculating method, and cloud shadow identification method on complex terrain. At the same time, we analyzed the effects of terrain and cloud shadow distortion on surface solar radiation estimation, carried out following conclusion:1. A cloud edge height matching algorithm based on Euclidean distance transform is proposed. Through which, the edge heights calculated by thermal infrared data could be matched to the corresponding high resolution image by the method of contour search based on Euclidean distance transform. Experimental showed that the results not only followed the law of distribution and diversification of heights information of the original background, also get the corresponding high resolution cloud heights distribution. The precision of cloud heights with an error of 70% between 0.1 and 0.3 km,12% is less than 0.1 km, with an average of 0.26 km. To a certain extent, it solves the problem when using heat infrared technology to obtain height in the high resolution.2. To more accurately describe the shadow distortion caused by the terrain, a geometrical method was used to calculate the true position of cloud shadows on complex terrain. Which could cloud obtain the position of shadow of corresponding cloud pixels in complex terrain, implemented the recognition of cloud shadows in any terrain conditions. Calculated and visual shadow cover were with high consistency, the comprehensive interpretation of shadow cover also fit well with calculated cover, r2 was 0.78, RMSE was 3.49, which was much better than results not considering the terrain. Calculated results of not considering the terrain was with big difference with shadow cover by interpretation, whose r2 was 0.12, RMSE was 15.14. Our method could make up for the defects of geometry method when it is applied to describe cloud shadows distribution in rugged topography.3. Considering complex terrain、cloud three-dimensional geometric effect and cloud shadow distortion, we made comprehensive analysis the influence of various factors on the surface radiation estimation:In sunny slope with cloud shadow, if we do not make any correction, results were obviously higher, the average deviation was about 400 W/m2.The results without considering shadow distortion was more, whose average deviation was about 500 W/m2. In the shady slope with cloud shadow, the result without considering shadow distortion was higher, the average deviation was about 300 W/m2; deviation will increase further if do not make any correction, with an average of about 400 w/m2. In sunny slope without cloud shadow, the result without considering shadow distortion was less, the average deviation was about 400 W/m2. In shady slope without the shadow, the result was less without considering distortion with an average deviation of about 300 W/m2. The results without any correction are mostly high, with an average of about 350 W/m2. Considering various factors could raise the accuracy of the estimated value about 300-500 W/m2. In some superposition area between cloud and shadow was not with very big difference, generally around 100 W/m2.