摘要
Terra／MODIS雪被产品在牧区积雪灾害的动态监测中，具有重要的应用价值。利用新疆北部地区2001—2005年4个积雪季的Terra／MOD10A1雪被产品和20个气象台站的观测资料，对比分析了MODIS积雪制图算法的分类精度。结果表明：①在晴空天气条件下，MODIS积雪制图算法分类的总精度达98．5％，积雪分类精度为98．2％，积雪多分误差小于漏分误差。分类精度在不同观测台站之间存在较大的差异，其总精度介于94．0％～100％，积雪分类精度介于77．8％～100％。②MODIS积雪算法的分类精度同海拔、坡度和坡向之间没有稳定的相关关系，雪深和土地覆盖是导致精度差异的主要原因。MODIS无法识别雪深低于0．5cm的积雪；当雪深介于0．5～3cm时，积雪识别精度较低，介于45．2％～76．2％。当雪深大于等于3cm时，积雪识别精度随雪深的增加而增大，平均值达98．6％以上。③MODIS积雪制图算法在草地区的分类精度最高，达98．9％；其次为农田、灌木林和城市与建设用地区，分别为97．9％，97．2％和96．9％。在农区积雪漏分误差低于多分误差，在草地、灌木林和城市与建设用地区则高于多分误差。
Terra/MODIS snow cover products play an important role in monitoring snow disasters in pastoral areas. In this paper, the accuracy of MODIS snow cover mapping algorithm in classification of snow cover in north Xinjiang is analyzed by using the MOD10A1 snow cover products and the meteorological data observed by 20 meteorological stations in north Xinjiang during the 4 snow cover seasons from 2001 to 2005. The results are as follows： （1） On fine days, the accuracy of MODIS snow cover mapping algorithm is as high as 96.5%, and the accuracy of snow cover classification is 96.6 %. The classification accuracy is quite different from different meteorological stations and varies in a range of 94.0 % - 100 %, and the accuracy of snow cover classification varies in a range of 77.8 % -100 % ; （2） There is no consistent correlation between the classification accuracy of MODIS snow cover algorithm and the elevation, slope and aspect. Snow cover depth and land use types are the main causes affecting the accuracy. Snow cover thinner than 0.5 cm can not be identified with MODIS. The accuracy of identifying snow cover in depth of 0.5 - 3.0 cm is low and varies in a range of 45.2 % - 76.2 %. When snow cover depth is thicker than 3 cm, the identification accuracy is increased with the increase of snow cover depth and higher than 98.6 % in average; （3） The accuracy of MODIS snow mapping algorithm is the highest in pastoral regions, it is as high as 98.9%, then in croplands, shrubbery lands and urban areas, and the accuracies are 97.9 %, 97.2 % and 96.9 % respectively. The omission error is lower than the commission error in croplands, but it is higher than the commission errors in pastoral regions, shrubbery lands and urban areas.
出处
《干旱区研究》
CSCD
北大核心
2007年第4期446-452,共7页
Arid Zone Research
基金
国家自然科学基金项目(30571316)资助
