The Reseach Of Specific Cloud Properties Based On MODIS Remote Sensing Data

Cloud properties have significant meaning in aspects of earth-atmosphere system budget balance of radiation, climate forecasts and atmosphere radiation transfer calculation. The data used for cloud properties researching is acquired through space-based satellite’s remote sensing. The advantages of wide spectral region and subdivided bands of MODIS(Moderate Resolution Imaging Spectroradiometer) remote sensing data are taken in this dissertation to make theoretical research and calculations of specific cloud properties, such as cloud detection, cloud phase detection, cloud top height and cloud thickness. In addition, the results and corresponding products are compared. The main research contents are as follows:(1) Different types of stripe noises have been removed in MODIS_L1B level’s MOD02 data, and cloud detection with the MOD02 data based on threshold method has been accomplished. As the detector-to-detector stripes, after all the max-stripes were judged exactly with the max mean of each swath, the stripes were removed by the method of multi-line adjacent interpolation. As the mirror-side stripes and random stripes, the stripes were removed by median filter. The cloud detection theoretical basis of atmospheric physics was analysed, in addition, the theory of the reasons why to choose different bands in cloud detection based on threshold method, and the arithmetic that combined threshold value model with confidence coefficient to detect cloud were researched. After cloud detection with two sets of MOD02 data based on threshold method, and compared the results with corresponding MOD35 cloud mask products, it shows that the detection effect was great.(2) The arithmetic of cloud phase detection based on BP neural network has been researched. After analysed the radiation theory of different cloud phase, four sets of characteristic data were constituted by five chosen special bands data which were reflectance 1.38 R, brightness temperature 6.7BT, brightness temperature difference 11 3.9BT ?BT and 11 12 BT ?BT. These characteristic data’s optical properties and three dimensional scatter diagrams at different cloud phase were detailed analysed. Then, the theory of BP neural network model arithmetic was researched. Utilized four sets of characteristic data combined with BP neural network model, cloud phase were detected with two sets of MOD02 data, and compared the results with corresponding MOD06 cloud phase products, different cloud phase could be detected commendable.(3) The arithmetic of cloud thickness based on empirical relationship method has been modified. After matching the longitude and latitude of the MODIS data and CloudSat data, two sets of single layer cloud MODIS data’s top height were calculated that utilized atmospheric static method and pressure-height profile interpolation. In addition, two sets of single layer cloud MODIS data’s thickness were calculated based on modified empirical relationship method with MODIS and CloudSat matching data. It turned out that these methods could calculate the single layer cloud of MODIS data’s top height and thickness relatively accuracy.(4) The software to arithmetic specific cloud properties based on MODIS remote sensing data has been developed. The arithmetic software was designed and developed by OpenCV library based on Microsoft Visual C++ 6.0 platform. With the software, any MOD02 data could be chosen, and the data could be showed as grayscale image, pseudo color image or RGB bands composite image, in addition, different types of stripe noises could be removed, cloud and cloud phase could be detected, and the results could compared with the corresponding products.