青藏高原地區(qū)季節(jié)性積雪遙感制圖與應用研究.pdf

1、 后結(jié)果進行精度驗證。結(jié)果表明，MODIS 每日無云積雪產(chǎn)品總分類精度較高，為94.7%。考慮到光學遙感對薄雪的探測精度本身較差，若去掉雪深<=3cm 的樣本，總分類精度達 96.6%。用戶開發(fā)的“青藏高原 MODIS 逐日無云積雪面積”分類算法可用于研究區(qū)無云積雪產(chǎn)品的獲取。 3.基于 MODIS 無云積雪產(chǎn)品長時間序列分析。 采用用戶開發(fā)的 MODIS 無云積雪圖像合成算法， 批量生產(chǎn)青藏高原 2002-2015 年的無云積雪

2、產(chǎn)品。 對 10 余年的 MODIS無云積雪圖像進行分析計算，包括各像素每年的積雪覆蓋日數(shù)、積雪開始時間和結(jié)束時間，總結(jié)長時間以來積雪的時空分布規(guī)律和變化，預測未來積雪的變化趨勢以及對氣候環(huán)境的影響。分析結(jié)果表明：十余年來，永久積雪大致呈減少趨勢，永久陸地呈上升趨勢，未來將繼續(xù)按照這個趨勢變化，這與全球變暖有著直接關(guān)系。 關(guān)鍵詞 ：青藏高原，MODIS，積雪，去云，精度驗證，長時間序列分析 研究類型：應用研究 reliability

3、is poor .Method based on the spatial combination of neighboring pixels removes scattered cloud with a high accuracy rate of 95.5%, but removed less cloud.The approach based on elevation applies in mountain area.While in

4、the hinterland of the plateau,error of classification is larger due to the smaller slope.Algorithm combined with passive microwave remote sensing data for cloud removing depends on the its recognition rate on cloud, ofte

5、n has a larger error.Method based on mathematics has a weak physical meaning in the Qinghai-Tibet Plateau,where the snow is broken and melts quickly.The methods with high accuracy can be used in the Qinghai-Tibet plateau

6、,including combination of Terra and Aqua satellite images，spatial combination of neighboring pixels and elevation filtering method. 2. An improved algorithm of cloud removal based on MODIS snow cover data.In view of prev

7、ious conventional algorithm,and considering snow and mountain terrain features,8 steps are used to remove all the cloud.Then, daily insitu measurement of snow depth data at 176 meteorological stations in two snow seasons

8、 of 2009 to 2011 were collected to verify the accuracy of each steps and the final results.The results indicates that, MODIS cloudless snow products prove a high overall classification accuracy of 94.7%. Considering that

9、 the accuracy of optical remote sensing is poor in detecting thin snow , if snow samples whose depth is less than 3cm are removed,the general classification accuracy reaches up to 96.6%. The classification algorithm of “

10、MODIS daily cloudless snow products in the Qinghai-Tibet Plateau“ can be used in study area. 3. Long time series analysis based on MODIS daily cloudless snow data.Using the cloud removal algorithm above，we produce cloudl

11、ess snow products form 2002 to 2015 of the Qinghai-Tibet Plateau.Snow parameters for more than 10 years were calculated and analyzed, including snow cover days each year of each pixel , snow start time and end time.Lastl

12、y, paper analyzes the changes of snow during this time,and forecasts the snow trends. The results show that: in the past twelve years，the permanent snow is generally decreasing and the permanent land is rising, the trend