基于物候特征参数的山西煤矿区典型复垦植被分类

发布时间：2018-03-28 12:13

本文选题：遥感数据　切入点：MODIS　出处：《国土资源遥感》2017年01期

【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。
[Abstract]:Based on the MOD13Q1 NDVI data obtained in 2001 and 2013, the low-pass smoothing Savitzky-Golay-S-G filtering method, interpolation method and Chebyshev Polynomialfitting are used to reconstruct the NDVI time series data. The key phenological characteristic parameters, such as the end date of growing season, the appearance date of maximum NDVI and maximum NDVI in growing season, were used to classify the typical reclaimed vegetation types in the study area. The results showed that the phenological characteristics of different vegetation in the study area were significantly different. In terms of the beginning date of growing season and the date when the maximum value of NDVI appeared, crops were more regular, while the cumulative value of NDVI in growing season of forest land was obviously different from that of crops and grasslands. The classification of grassland and woodland based on vegetation phenological parameters obtained good results. The overall classification accuracy was 89.677.It was better than the result of multi-temporal unsupervised classification. This research provides certain data basis for ecological environment restoration evaluation of coal mining area in Shanxi province.
【作者单位】：山西省自动化研究所;堪萨斯州立大学农学系环境与农业空间分析实验室;北京师范大学地理学与遥感科学学院;
【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
【分类号】：Q948;TP79

【相似文献】