中文摘要：

MODIS以其时间分辨率、光谱分辨率的优势成为全球及区域土地覆盖研究的主要数据源,但如何快速准确的提取所需土地覆盖信息一直是科学界研究的焦点问题。对于NDVI时序数列分类方面的研究很多,其中影响分类精度的一个重要因素就是NDVI的数据质量问题。本文利用1年MODIS旬最大值合成数据经过Savizky-Golay滤波器平滑滤波处理,应用Logistic模型提取东北地区6个关键物候参数,利用特征参数的主成分影像实现研究区植被覆盖分类,结果显示关键物候指标能提取NDVI时间变化曲线中的有效信息,去除造成植被类型混分的噪声,利用关键物候参数分类可提高植被类型的分类精度,对区域土地覆盖分类精度的提高提出了新思路。

英文摘要：

Moderate Resolution Imaging Spectroradiometer （MODIS） offers a unique combination of spectral,temporal,and spatial resolutions compared to previous global sensors,making it a good candidate for large scale land cover classification.It can provide high quality multi-temporal Normalized Difference Vegetation Index （NDVI） which is closely related to vegetation growth.NDVI time series profiles were different from various vegetation types.Numerous studies on land cover classification based on clustering do not perform quantitative analysis on the unique characteristics of vegetation,which means if two vegetation categories had similar NDVI profiles,they would be grouped into the same class.The primary objective of this study was thus to develop a land cover classification scheme from multi-temporal MODIS data through analyzing the characteristics of vegetation phenology.Data used in this study was 10-day composite NDVI time series from MODIS with a spatial resolution of 250 m from Jan 1 to Dec 31,2004.The Savitzky-Golay smoothing filter was performed on the multi-temporal NDVI time series to eliminate anomaly values.First,a variety of indices which would be effective for vegetation classification were selected.Through analyzing the vegetation categories of the study area,five key plant phenological variables were extracted,i.e.,the time of maximum leaf area index,the start of the growing season,the end of the growing season,the maximum NDVI value in growing season and the lowest NDVI value.Second,the ＂S-shaped＂ growth curve was simulated with a logistic function and divided into two phases.Such values of indices were derived by means of ERDAS IMAGINE.On the basis of the principal component analysis （PCA） performed on indices of vegetation phenology,the ISODATA method was used for vegetation category identification.In addition,validation data for the classification were obtained in Sep,2004,including every vegetation types of the study area.Combined the validation data with field samples,the overall acc