许多运用卫星资料的工作首先要求将卫星观测区域内的云去除。而美国国家航空和宇宙航行局(NASA,National Aeronautics and Space Administration)的中分辨率成像分光辐射计(MODIS,Moderate Resolution Imaging Spectroradiometer)云检测产品MOD35_I,2在光亮表面(如中国西部地区)上空存在较大误差。因而对于云检测结果敏感的工作(如反演气溶胶光学特性),如果简单的采用MOD35_L2,将给最终的结果带来误差。本文提出时间序列方法和MODIS第1(0.659μm)、3(O.470μm)、4(0.555μm)通道多通道方差方法与MOD35_L2中部分云检测方法相结合对中国西部地区进行云检测,得到了较好的云检测结果。
Cloud detection is regarded as the first step in many applications of satellites data. However, the NASAI s MODIS (Moderate Resolution Imaging Spectroradiometer) cloud mask known as MOD35_L2 is undermined by many errors, especially over bright reflectance areas (e. g. the western China). Therefore, as to those studies (e. g. retrieval the aerosol optical properties) which are sensitive to the result of cloud detection, it will bring errors to the ultimate result that adopting MOD35_L2 directly. To resolve this problem, we develop a new method of cloud detection by combining the time-sequence test, multiple-band deviation test and part of the tests executed in MOD35_L2 together. In the multiple-band deviation test, the MODIS 1(0. 659 μm), 3(0. 470 μm) and 4(0. 555 μm) bands are used. The application of the new method over western China indicates great improvement than that of MOD35_L2.