县域尺度上基于GF-1PMS影像的冬小麦种植面积遥感监测

发布时间：2018-11-23 21:07

【摘要】：为探究县域尺度上基于高分一号卫星(GF-1)PMS影像进行冬小麦遥感监测的可行性及精准性,以河南省滑县为研究区,遴选2015年2月上旬GF-1 PMS影像6景,对影像进行辐射定标、FLAASH大气校正、NNDiffuse融合、几何精校正、地图投影转换等预处理后,在外业调查和样本分析的基础上构建一种新的冬小麦决策树分类模型,模型第1层决策方案中NDVI0.311的像元为冬小麦,得到冬小麦的粗分类结果;在此基础上进行第2层决策分类,以进一步提高冬小麦的分类精度,分类方案为第1波段地表反射率0.146、第2波段地表反射率0.148、第3波段地表反射率0.135、第4波段地表反射率0.250的像元为冬小麦。对分类结果进行形态学滤波处理,以消除或减少分类结果中孤立的像元。分别基于决策树分类模型与ENVI软件自带的IsoData非监督分类模型,对比分析GF-1PMS影像和同时期Landsat-8OLI影像在冬小麦面积提取上的精度。结果表明:基于新构建的决策树分类模型,2015年滑县冬小麦种植面积为115 715.81hm2,混淆矩阵检验总体精度为99.62%,Kappa系数为0.99;PMS影像提取冬小麦的混淆矩阵总体精度比OLI影像高出9个百分点。说明县域尺度上基于单时相GF-1PMS影像在冬小麦收获前提取冬小麦种植面积是可行的,提取精度较高。
[Abstract]:In order to explore the feasibility and accuracy of remote sensing monitoring of winter wheat based on PMS images of Gao Fen 1 satellite (GF-1) at county scale, six scenes of GF-1 PMS images were selected from Huxian County, Henan Province, in the first ten days of February, 2015, in order to investigate the feasibility and accuracy of remote sensing monitoring of winter wheat. After pre-processing such as radiometric calibration, FLAASH atmospheric correction, NNDiffuse fusion, geometric correction, map projection conversion and so on, a new classification model of winter wheat decision tree was constructed on the basis of field survey and sample analysis. In the first layer of the model, the pixel of NDVI0.311 is winter wheat, and the coarse classification result of winter wheat is obtained. In order to further improve the classification accuracy of winter wheat, the classification schemes are as follows: the first band surface reflectivity 0.146, the second band surface reflectivity 0.148, the third band surface reflectivity 0.135, the third band 0.135, the second band 0.148, the third band 0.135. Winter wheat is the pixel of surface reflectance 0.250 in band 4. The classification results are processed by morphological filtering to eliminate or reduce the isolated pixels in the classification results. Based on the decision tree classification model and the IsoData unsupervised classification model with ENVI software, the accuracy of GF-1PMS image and Landsat-8OLI image in winter wheat area extraction were compared and analyzed. The results showed that based on the newly constructed decision tree classification model, the planting area of winter wheat was 115,715.81hm2in 2015, and the overall accuracy of confusion matrix test was 99.62kappa coefficient 0.99. The overall accuracy of PMS image extraction of winter wheat confusion matrix is 9% higher than that of OLI image. It is feasible to extract the planting area of winter wheat before harvest on the county scale based on monochronous GF-1PMS image, and the precision of extraction is high.
【作者单位】：中国科学院遥感与数字地球研究所遥感科学国家重点实验室;中国科学院大学环境与资源学院;河南大学环境与规划学院;
【基金】：国家自然科学基金项目(41301390,4137138) 国家“973”计划项目(2013CB733405) 国家“863”计划项目(2014AA06A511) 云南省科技计划(2010AD004) 高分辨率国家重大专项(20-Y30B17-9001-14/16)
【分类号】：S127;S512.11

【相似文献】