Cloud Recognition Using MODIS Data

Cloud is an important factor which impacts the accuracy of passive remote sensing data. The cloud covers the view scope of the visible band, and the water vapor and various aerosols in the cloud absorb the infrared seriously. The particles in the cloud can also cause scattering, reflection and refraction. Determining the scope and phase of cloud accurately can not only enhance the precision of passive remote sensing inversion of the land, but also help explore the water vapor and atmospheric physics composition among the cloud.MODIS cloud detection algorithm can be divided into five groups. Thick high clouds with emitted radiation threshold tests can be detected using the brightness temperatures of BT11,BT_13.9,BT_6.7. Thin clouds are identified with brightness temperature difference tests of BT8.6-BT11,BT11-BT12,BT11-BT3.9,BT11-BT6.7. Reflectance thresholds for R0.66,R0.87,R0.87/R0.66 reflectance ratio tests are helpful for low clouds detection. Upper tropospheric thin clouds can be detected with a reflectance threshold test for R1.38. And cirrus is sensitive to brightness temperature difference tests BT3.7-BT12,BT11-BT₁₂. According to the nature of surface and solar radiation, the detection region can be divided into 10 types, including daytime land, daytime water, night land, night water, daytime desert, night desert, daytime ice/snow, night ice/snow, daytime coastline, and night coastline. Each type is classified using one or more detection algorithms in order to get the pixel detection value. The results of MODIS detection are based on single pixel, and most results of single pixel test depend on the threshold. The threshold is not global, so credibility is used to show the consistency between observation radiation signal and the expected cloud radiation signal. The ultimate credibility of MODIS cloud detection is composite of four levels: high credibility clear sky, the poor credibility clear sky, uncertain, and cloudy.Observation shows that the cloud can be divided into three categories. The first category is the liquid cloud which formed by water droplets. The second category is composed of ice called ice cloud. The third category is composed of liquid and solid which named mixed cloud. Sometimes the mix cloud contains solid (ice or cold water) only on the top, while liquid water on the bottom, other times may be the whole cloud or part of it is composite of the cooling water droplets and ice crystals, or a mixture of components. Besides, it is likely to encounter that there are ice in the cloud with cooling water cover it. Cloud absorption and scattering is decided by the exponential of refraction and cloud particle size. The 3-spectrometry shows that the 8-11μm brightness temperature difference of clear sky is negative or close to zero, and the 11-12μm brightness temperature difference is little. The difference depends on the size of the margin in the water column. Ice cloud and water cloud can be clearly divided into two categories by in the difference of brightness temperature. The ice cloud is above 45°diagonal and the water cloud is below 45°diagonal, while the mixed-phase cloud is near 45°diagonal. Based on the 3-spectrometry, we can retrieval 5×5 pixels of cloud phase with MODIS 8.55 , 11.03 and 12.02μm bands.The cloud detection products can be obtained by using the MODIS IMAPP (International MODIS / AIRS Processing Package) _SPA (Science Processing Algorithm). Furthermore, the 3-spectrometry model is validated and the cloud phase product also obtained. Therefore, the cloud detection algorithm based on the EOS-MODIS data was efficient, especially when it was used to detect the middle and high cloud. Almost all the high cloud can be detected. MODIS-phase products are much better on high cloud-identification. Meanwhile, we can get the top characteristics of systemic clouds clearly from the retrieval results.