中文摘要：

以北疆为研究区,结合气象台站记录的雪情数据,利用地理信息系统方法分析了2004年12月1日至2005年2月28日期间北疆地区90个时相的MODIS每日积雪产品MOD10A1和8日合成产品MOD10A2的积雪分类精度.研究表明：1）当积雪深度≤3 cm时,MOD10A1对积雪的识别率非常低,仅为7.5%;积雪深度为4~6 cm时,积雪识别率达到29.3%;积雪深度为15~20 cm,平均积雪识别率达到45.6%.当积雪深度〉20 cm时,平均积雪识别率为32.2%;2）MOD10A1产品的积雪分类精度受天气状况的严重影响.在晴空状况下,该产品的最大积雪识别率达到58.2%;但是在多云或阴天时,平均积雪识别率仅为17.8%;3）下垫面对MOD10A1的分类结果也会造成影响,在荒漠区MOD10A1的积雪识别率为39.8%,在草原和稀树草原区的积雪识别率为37.2%,农业用地的积雪识别率最低,为29.1%;4）MOD10A2产品可较好的消除云层对地表积雪分类精度的影响,平均积雪识别率达87.5%,可较好的反映地表积雪的分布状况.

英文摘要：

By the use of NASA EOS Terra/MODIS snow products of MOD10A1 and MOD10A2 and climatic data, the snow classification accuracy was analyzed using Geographic Information System （GIS） techniques for 90 temporal daily snow com- posite products of MOD10A1 and 11 temporal eight-day composite products of MOD10A2 from December 1, 2004 to February 28, 2005. Results showed that： 1） When snow depth is less than 3 cm, the precision of snow identified by MOD10A1 is very low, only 7.5% ; as snow depth is between 4 cm to 6 cm, MOD10A1 snow identification accuracy reaches to 29.3% ; and the precision is 45.6 when snow depth is between 15 to 20 cm; The mean accuracy is 31.5% when the snow depth is great than 20 cm; 2） the precision of snow identification for MOD10A1 products is severely affected by climatic situation. Under sunshine weather conitions, the snow identification accuracy of MOD10A1 reaches to 50.6%; but the average of snow identification rate was only 18% when it is cloudy or overcast; 3） The condition of underlying surface is another factor affecting the MOD10A1 classification results, such as, under clear sky conditions, the precision of snow identified by MOD10A1 is 68% for grasslands with sparse trees and shrubs; in desert, the snow identification rate is 64.4%, and only 40% for agricultural land and 4） It can better eliminate the influence of amount of clouds and improve the snow classification preci- sion for MOD10A2 products, as a result, the mean precision of snow identification is 87.5%, which can reflect better the ground snow distribution and plays an important role in snow disaster monitoring in pastoral areas.