中文摘要：

云是光学遥感影像解译及地表参数定量反演的主要障碍,破坏了遥感产品的时空连续性;而随着遥感技术的推广和应用,大量的遥感参数产品已经被用于各个领域,比如卫生领域的疾病监控、地质灾害领域的滑坡监测以及城市雨洪预警等等;云导致的遥感数据质量下降与各领域对遥感数据质量要求的不断增加构成了不断激化的矛盾。针对这一矛盾,结合当前关于云消除研究的现状;提出了一个具有物理机制的云像元反演算法。该算法在计算过程中考虑了地表在不同入射、观测方向上的二向反射分布函数（BRDF）,并引入时空权重因子实现对时空异质性的考虑;这些使该算法能够保证在各种情况下均具有良好的整体精度,从而保证该算法能够胜任定量遥感领域的研究需求。应用该算法,在黑河中游地区展开实验,反演结果与中分辨率成像光谱仪（MODIS）多个波段的数据对比均展现出较好精度。其中,整体均方根误差RMSE达到0.012,整体均方根误差百分比RMSE%为8.8%,整体相关系数平方R~2达到0.98,与对比算法相比较具有极大的提升;这表明该算法理论基础扎实,并具有较好的实用性。

英文摘要：

As an important factor for diminishing quality of remote sensing data,cloud results in the ineffective observation of sensors abroad on satellites in most cases,which deteriorates quality of remote sensing data and destroys the spatio-temporal continuity of remote sensing data; but,the requirement for the quality of remote sensing data is continually improved with the deepening of its application in various field（ e. g.,disease monitoring,landslide monitoring and flood warning）. So,the remote sensing data with cloud restricts its applying range and depth in social and economic field. A method having physical mechanism was proposed which considers the Bidirectional Reflectance Distribution Function（ BRDF） of land surface on different solar-viewing geometry and spatiotemporal weight was introduced to cancel error. Such a design makes sure a good precision of the proposed method under different condition; then,the result from the proposed method can be applied for retrieving high level remote sensing product with high accuracy. Based on the proposed method,a test was carried out in Heihe river basin by using Moderate Resolution Imaging Spectroradiometer（ MODIS） band reflectance,and the result shows a total RMSE of 0. 012,a total percent of RMSE% of 8. 8%,and a total R2 of 0. 98,which is better than other methods.Based on the results and comparison,a conclusion is that the proposed method has a better accuracy,and can be used more widely in various field of society,economy and environment.