利用多源时序遥感数据提取大范围水稻种植面积

Mapping paddy fields in large areas, based on time series multi-sensors data

摘要: 为避免大范围作物面积提取中的分区问题，提出了一种新的采用时序MODIS日反射率产品（MOD09GA）和HJ-1CCD数据的大范围水稻种植面积提取方法。利用样区多时相高分辨率HJ-1CCD数据，进行高分辨率水稻制图，获取水稻移栽期和生长期的特征。再利用时序MODIS日反射率产品自动提取移栽期稻田水分信息和生长期生长信息，实现大范围的水稻种植面积测量。以江苏省为试验区对该方法进行了试验，结果表明该方法可以较高精度的测量大范围的水稻种植面积，水稻提取精度为93.3%，Kappa系数0.88。

Abstract: In order to avoid the regionalization problem of crops discrimination in large areas, this paper proposed a novel rice extraction method by combination of time series of MODIS daily surface reflectance product (MOD09GA) and HJ-1CCD data. Using HJ-1 CCD data, parameters were extracted in rice transplanting period and growing period for the sampled area. Then rice area were mapped by distinguishing water information in transplanting period and NDVI change information in growing period using time series MODIS data. Validation of the method was conducted in Jiangsu Province and high mapping precision of 93.3% was acquired with Kappa Coefficient of 0.88.