MODIS NDVI和AVHRR NDVI对草原植被变化监测差异

Differences between MODIS NDVI and AVHRR NDVI in monitoring grasslands change

以草地作为研究载体,对比分析草原植被AVHRRNDVI和MODISNDVI两种NDVI序列的年内、年际变化特征,讨论两种NDVI序列对降水量、平均气温和水汽压3种气候因子的响应差异,为合理选择NDVI序列对植被进行监测研究提供参考。结果表明:(1)两种NDVI序列所反映的草原植被年内变化趋势相似,但MODISNDVI对各类草原的区分度优于AVHRRNDVI;(2)两种NDVI序列所反映的2000年—2003年草原植被年际变化差异明显。较之于MODISNDVI,AVHRRNDVI变化趋势分类图表现出更强的植被改善趋势,植被改善面积在AVHRRNDVI变化趋势分类图中占94.25%,在MODISNDVI中为83.33%;两种NDVI变化趋势分类图反映的植被变化趋势吻合度为52.88%。(3)两种NDVI序列与水汽压、降水量相关性差异显著。MODISNDVI与各站点平均气温的相关系数均大于GIMMSNDVI;而MODISNDVI与水汽压的相关系数83%(10个站点)小于GIMMSNDVI,与降水量的相关系数67%(8个站点)小于GIMMSNDVI。

In this paper, with grassland of Xilingol as the research site, the average method, unary linear regression model, correlation analysis are used to compare the annual and interannual characteristics of NDVI series, including the MODIS NDVI and the AVHRR NDVI. The precipitation, mean temperature, vapor pressure are taken as climatic variables to compare the relationship between NDVI series and climatic factors over different types of grasslands. The objective is to provide references for reasonable use of NDVI series to monitor vegetation changes. The results show that, （1） Annual characteristics of the MODIS NDVI and the GIMMS NDVI are similar, yet the MODIS NDVI can distinguish each steppe better than that of the GIMMS NDVI. （2） Different interannual characteristics are observed between MODIS NDVI and GIMMS NDVI of grassland vegetation during 2000 to 2003. Compared with MODIS NDVI, vegetation increase trend is stronger in GIMMS NDVI change trend classification image, with a proportion of vegetation increasing area in GIMMS NDVI of 94.25%, while only 83.33% for MODIS NDVI. Consistency of the two NDVI classification images is only 52.88%. （3） Correlation differences between the two NDVI series and climatic factors are significant. Correlation coefficients between MODIS NDVI and mean temperature are larger than that of GIMMS NDVI. The proportion of correlation coefficients between MODIS NDVI and vapor pressure are lower than that of GIMMS NDVI with a value of 83% （about 10 meteorological stations）. Similar results are also observed for the precipitation with a proportion of 67% （about 8 meteorological stations）.