摘要
中分辨率成像光谱仪(M OD IS)已在全球资源环境监测中发挥了重要作用,但是它的低分辨率成为提高分类精度的阻碍。利用M OD IS的高时间分辨率弥补其低空间分辨率的不足,设计分类器改善分类精度。利用2003年23个时相的M OD IS_EV I图像,构建华北平原植被指数图像时间立方体。在谐波分析去噪标准化基础上,从EV I时间谱上提取5个表征物候差异的特征向量,结合表征地气交互作用差异的地表温度(LST)信息及表征地表固有的空间分异特征的坡度信息,建立分类二叉树进行土地覆盖分类。结果表明,与2000年TM分类结果的总体一致性为75.5%,K appa系数为0.68。而NA SA U SG S基于M OD IS分类精度为66.0051%,K appa系数为0.3209。进一步与2003年耕地面积的官方统计资料的比较表明,该文的估算误差为34.0507 khm2,而NA SA U SG S的估算误差高达66.1205 khm2。研究表明利用高时间分辨率的M OD IS植被指数时间序列获得较高精度的土地覆盖分类结果是可能的。
MODIS data play an important role in global environmental and resource researches. But its low spatial resolution sometimes becomes a regretful factor by some people in pursuit of more precise classification results. In this research, MODIS high temporal resolution was used to improve the accuracy of land cover classification of the North China Plain using MODIS_EVI time-series of 2003. Harmonic Analysis of Time Series（HANTS） was performed on the MODIS_EVI image time series to reduce the cloud or other noise effects. Based on five phenological features derived from EVI profiles, as well as on Land Surface Temperature（LST） and topographic slope, a simple but reasonable decision tree was built to distinguish the ambiguous land cover classifications. The overall accuracy of the final land cover map was 75.5%, and the kappa coefficient is 0. 68. While the overall accuracy and kappa coefficient of NASA USGS product are 66. 0051% and 0. 3209 respectively. When compared with the cropland area from official statistics, the classification in the paper shows much higher consistence with an overall mean square root error of 34. 0507 kilo-hectare versus 66. 1205 kilo-hectare by USGS product, indicating that land cover classification using MODIS EVI time series and decision tree is feasible and promising.
出处
《农业工程学报》
EI
CAS
CSCD
北大核心
2006年第12期128-132,F0003,共6页
Transactions of the Chinese Society of Agricultural Engineering
基金
国家自然科学基金项目(40271085)
973计划子课题(2002CB412506)
关键词
MODIS
时间谱
物候特征
决策树
土地覆盖分类
耕地
MODIS
temporal profile
phenological feature
decision tree
land cover classifications
cropland
