The Study On Extraction Of Snow Information Of The Area Using TM Image In Binggou Watershed, Qilian Mountains

Binggou Watershed, located in the middle of the northern slope of the Qilianmountains, is a larger second tributary of the east branch of the upper reaches of theHeihe River Basin. And it belongs to seasonal snow. Snow in the mountains is animportant supplement to the main supply of many rivers and groundwater. In aridregions of northern China, alpine snow is the most important water resources. Thesnow in Qilian mountains is not only providing a lot of water for the agriculturalirrigation in Hexi Corridor, but is also the basis of the economic construction and thepeople’s lifeline. Of which50percent is from the snowmelt of the Qilian mountains.Therefore, keeping abreast of the snow resource changes status of Qilian mountains innorthwestern China is particularly important.By analyzing the correlation between snow depth and snow reflectance, thefeasibility and accuracy of extracting snow depth based on TM data in complexterrain areas was explored. Based on RS and GIS, using the measured snow depthdata and the image data, the snow depth estimation models were established. Thus thesnow depth extraction of the study area was completed. Further, the main factors ofthe snow depth simulation using the TM data were preliminarily discussed. Theresults show that:(1) The distribution of snow in the study area was obtained by the application ofsnow cover index method to identify the snow. Most of the Binggou Watershed wascovered by snow. Because of there being a lot of spring snow on March17,2008, inthe early of the snowmelt, the snow cover is very high.(2) Based on the sample data, the correlation analysis between different bandsreflectance and the measured snow depth was done. There has a significantly positivecorrelation between snow depth and the reflectance of the TM visible light band andnear infrared band. While, a significant negative correlation between the snow depthand the TM shortwave infrared reflectance was presence.(3) It is feasible that extraction of snow depth by the use of TM or ETM+. By thereflectance ratios between the TM2and TM5, the simulation model was established. Itis better to estimate the snow depth, and the retrieval accuracy is77.8%. (4) By the contrast analysis of the estimation values and the measured values, itshows that there exists some connections between the accuracy of the extraction andthe snow depth. When the snow depth is between15centimeters and30centimeters,the average error, the positive average error, mean absolute error and root meansquare error are smaller.(5) The terrain and weather conditions affect the snow distribution of the study area,which results in the heterogeneity of the spatial distribution of snow depth, and has aeffect on the accuracy of the simulating results. Meanwhile, the topographic affectslargely the reflectivity which received by the satellite. Thus, when the inversion ofsnow depth, many errors were produced under the complex terrain conditions.