Research On Snow Monitoring Of Transmission Corridor Based On Combination Of Multi Remote Sensing Data

Snow is one of the most active natural elements of the earth surface and has a close influence on human activity. With the issue on global climate changing becoming more and more serious, China has been deeply affected by the snow disaster. Snow disaster is an important factor which causes the paralysis of the power network. In 2008, a dozen of provinces of southern China was affected by a large area of ice covering which was hard to see. Because of the continuous rain and snow, large area of the power lines were covered by ice, the power towers collapsed, power facilities suffered unprecedented devastation, which caused great impacts and losses to people. So, it’s necessary to do the research of monitoring the snow covering of power towers.With the development of human technology, remote sensing has become an important tool for monitoring snow. Remote sensing images commonly include optical remote sensing images and SAR images. Because of the different imaging mechanism, the two type images have the advantage and disadvantage to the precision of the recognition of snow pixels. Optical images are much accurate, but are often affected by the weather. SAR images can work in different weather conditions, but have less accuracy. The power towers are often located in the mountainous regions which have complex terrain. The changeable climate, large area shadow are great tests to the precision of the snow pixels.The common SNOMAP algorithm has a significant problem that it can’t extract the snow pixels in mountainous regions. In this paper, the real reflectance of the pixels which are in shadow can be calculated by the radiation transfer relationship which is based on ground surface, atmosphere, sensor by 6S models and the terrain parameters generated by the DEM data. Finally the snow pixels are identified by the SNOMAP algorithm. By using this algorithm in the research of identifying snow pixels in the regions of Bogda Peak of TianShan Mountain and Gongga Mountain and doing precision analysis by the confusion matrix, The conclusion is that the algorithm can identify 80% of snow pixels which are in shadow regions successfully and guaranteeing the identifying precision of snow pixels in non-hatched region at 80%.SAR images can solve the significant deficiency of optical images that it can’t work in cloudy weather. The snow pixels are recognized by the method of coherence analysis in this paper. The snow pixels in the eastern and western mountainous region of Daqingliangzi segment of the province highway can be identified by the conclusion that the snow pixels have lower coherence. The coherence analysis can’t solve the problem that the mix of snow pixels and bushes pixels, because the imaging of SAR images use X-Band which has a lower penetration. But this method can be used as the primary classification of snow cover mapping.Finally,Aiming at the problem that the optical images can’t distinguish the dry snow and wet snow pixels, this paper puts forward a method of combining optical images and SAR images to recognize the snow pixels in mountainous regions. By the algorithm(SNOMAP algorithm based on terrain radiant correction) that has been realized, total snow pixels can be recognized in mountainous region. By SAR images, total wet snow pixels can be recognized in mountainous region. At last, the classification graph of dry snow pixels and wet snow pixels can be obtained in part region in Qilian Mountain.