根据2001~2007年江汉平原四湖地区增强型植被指数(EVI)的时序数据,运用时间序列谐波分析法(HANTS)进行去云处理,计算各个像元EVI时序值的平方和N,将N〉0.25的像元划为植被覆盖区,N〈0.25的像元划为非植被覆盖区。对植被覆盖区采用二次差分法判断EVI时序数据的峰值出现次数,年内峰值出现1次的为林地和草地,出现2次的为耕地,出项3次及以上的为蔬菜地,采用监督分类方法划分非植被覆盖区中的城镇、水域和植被覆盖区中的林地、草地,制作了四湖地区2001~2007年的土地覆盖分类图,与基于2001年多时相ETM数据的土地覆盖分类进行了比较,耕地面积相差在10%以内。利用MODIS的EVI数据能够快速、有效、低成本进行土地覆盖分类。动态分析结果发现,耕地面积处在不断减少过程中,其余覆盖类型的面积呈增加趋势。
Based on the time-series data of the Enhanced Vegetation Index(EVI) from 2001 to 2007 in Four-lake area in Jianghan Plain,after the cloud removing process by Harmonic Analysis of Time Series(HANTS),the sums of square of the EVI in one year for every single pixel were calculated.The pixels with sum of square of EVI less than 0.25 were identified as area uncovered by plant,contrariwise,area covered by plant.For the area covered by plants,the occurrence time of the peak of EVI were counted by quadratic differential method.The area was divided into forest and pasture land,crop land and vegetable land when the frequency of occurrence of peak value was 1,2 and 3 or upwards,respectively.Supervised classification methods were used to identify towns and water in the area uncovered by plants,and the forestry and pasture in the area covered by plants.The figures of land cover types in Four-lake area from 2001 to 2007 were made.The results were compared with the classification come from multi-temporal ETM in 2001.The difference of the cropland areas between the two classifications was less than 10%.So it could be possible to monitor the change of land cover quickly and cheaply by Modis-EVI data.It was found that the area of vegetable land and crop land was decreasing,and the area of other types was increasing in the studied region.