返回Google圖書搜尋

Image-processing Techniques to Retrieve Cloud-free Snow Covers from Remote-sensing Modis Images

出版	University of California, Irvine, 2011
ISBN	126713285X9781267132857
URL	http://books.google.com.hk/books?id=GySmAQAACAAJ&hl=&source=gbs_api

註釋To improve the usability of remote-sensing observation of snow in practical hydrological and water-resources applications, this dissertation explores and studies image-processing techniques to solve two major problems on the topic of snow cover: (1) retrieve cloud-free snow-cover data from cloud-hindered MODIS snow-cover images, and (2) based on the cloud-free snow-cover maps, analyze the spatial distribution of snow-melting rates. The first problem is solved in two steps: the first step attempts to mitigate the cloud hindrance, and the second step clears the remaining clouds. In the first step, three types of analyses are used to process the MODIS snow-cover images. The first type of analysis relies on the multiple data sources; they can be data from different sensors or the same sensor on different satellites. The second type is spatial analysis, conducted in a neighborhood. (of what?) Spatial variables, such as elevations, help to improve the results based on physical mechanisms. The third type is temporal analysis, short-term or long-term, taking advantage of the persistent characteristics of snowpack. After the first step, more than two thirds of the clouds can be eliminated. Clouds can't be reduced, but the number of them can be). The remaining clouds are cleared by the variational interpolation method, which assumes that the snow accumulating or snow-melting process follows the minimum energy theorem and, therefore, can be modeled with radial basis functions. The dominant-point detection technique helps to increase the computing efficiency and improves the results. The variational interpolation process is validated using a statistical bootstrap validation using snow-melting events occurring over five days and four years of observed cloud library over the Sierra Nevada Mountains, California; the results are satisfying. Based on the cloud-free snow-cover maps, we use the optical flow method to calculate the snow-melt rate field. The method is used in three basins in California, and the average snow-melt rate in the region is correlated with the increased outlet discharge. The results indicate that the method can be used to monitor snowmelt and avoid flood disasters.