基于物候特征参数的山西煤矿区典型复垦植被分类  

Typical reclamation vegetation classification based on phenological feature parameters for coalfields in Shanxi Province

作　　者：张彦彬[1] 安楠[2] 刘佩艳[1] 贾坤[3] 姚云军[3] 

机构地区：[1]山西省自动化研究所,太原030012 [2]堪萨斯州立大学农学系环境与农业空间分析实验室,美国堪萨斯州66506 [3]北京师范大学地理学与遥感科学学院,北京100875

出　　处：《国土资源遥感》2017年第1期170-177,共8页Remote Sensing for Land & Resources

基　　金：国际合作项目"利用卫星遥感技术对煤矿复垦生态环境的动态监测及分析"(编号:2013DFA91870);中国科学院数字地球重点实验室开放基金项目"低空间分辨率遥感数据时相特征改善高分辨率数据农作物分类精度研究"(编号:2014LDE011)共同资助

摘　　要：基于2001―2013年获取的MOD13Q1 NDVI数据,采用低通平滑Savitzky-Golay(S-G)滤波方法、插值法及切比雪夫多项式(Chebyshev Polynomial)拟合对NDVI时序数据进行重构;通过提取植被生长季开始日期、生长季长度、生长季结束日期、生长季NDVI最大值及NDVI最大值出现日期等关键物候特征参数,对研究区典型复垦植被类型进行分类。结果表明:研究区不同植被的物候特征具有显著差异,从生长季开始日期及NDVI最大值出现日期来看,农作物较有规律;而林地的生长季NDVI累积总值则明显区别于农作物及草地;农作物、草地和林地基于植被物候特征参数分类取得了较好结果,总体分类精度达到89.67%,优于采用多时相非监督分类的结果;该研究为山西省煤炭矿区生态环境恢复评价提供了一定的数据基础。In this paper, the authors reconstructed MOD13Q1 time -series NDVI data from 2001 to 2013 using Savitzky- Golay filter and Chebyshev Polynomial methods for classifying vegetation types in the six coalfields in Shanxi Province. The key phenological parameters were extracted from the reconstructed NDVI data, such as the beginning dates of the growing season, length of the growing season, the ending dates of the growing season, the maximum NDVI value and the responding dates. The results show that different vegetation types of the six major coalfields in Shanxi have different phenological features. Cropland has distinguishable differences from grass and forest. Similarly, forest is distinguished from grass and cropland by integration of total growth. It is shown that the classification of vegetation types can achieve better results by extracting and analyzing the phonological parameters compared with multi - temporal unsupervised classification. The overall classification accuracy reaches 89.67%. This study provides a robust method for assessing long- term ecological conditions and monitoring vegetation coverage changes of the six major coalfields in Shanxi Province.

关 键 词：遥感数据 MODIS NDVI 物候参数 植被分类 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置] S127[自动化与计算机技术—控制科学与工程]