摘要
为研究GF-1时间序列影像的识别技术对大尺度葡萄信息识别提取的可行性,探索大尺度葡萄快速精确提取的新思路。基于GF-1/WFV时间序列影像数据分析试验区主要植被类型的归一化植被指数(Normalized difference vegetation index,NDVI)、增强型植被指数(Enhanced vegetation index,EVI)的时序变化特征,结合葡萄物候期构建决策树,提取宁夏贺兰山东麓葡萄的空间分布信息。结果表明,该方法可以有效提取贺兰山东麓葡萄分布信息,分类总体精度为95%,Kappa系数为0.91。葡萄提取的生产精度为93%,用户精度为96%。在时间序列数据中,葡萄提取的窗口期为3—5月掩埋期和7—9月生长旺盛期。NDVI时间序列能够较好地区分作物和防护林,EVI时间序列能够区分葡萄地和防护林。
In order to study the feasibility of GF-1 time series image recognition technology for large-scale grape information recognition and extraction,a new idea for rapid and accurate extraction of large-scale grape was explored.Based on GF-1/WFV time series image data,the time sequence change characteristics of normalized difference vegetation index(NDVI) and enhanced vegetation index(EVI) of the main vegetation types in the experimental area was analyzed,and the spatial distribution information of grape in the eastern foot of Helan mountain in Ningxia province was extracted by combining the establishment of decision trees in grape phenology.The results showed that this method could effectively extract the information of grape planting distribution in the eastern foot of Helan mountain,with a total classification accuracy of 95% and a Kappa coefficient of 0.91.The production precision of grape extraction was 93%,and the user precision was 96%.In the time series data,the window period of grape extraction was grape burial period which from March to May and vigorous growth period which from July to September.NDVI time series could distinguish crop and protection forest,and EVI time series could distinguish grape and protection forest.
引文
[1] 杜国华.宁夏贺兰山东麓葡萄产业发展历程与目标[J].农业科学研究,2010,33(3):50-54.
[2] 任建强,陈仲新,周清波,等.基于时序归一化植被指数的冬小麦收获指数空间信息提取[J].农业工程学报,2010,26(8):160-167.
[3] 刘佳,王利民,姚保民,等.基于多时相OLI数据的宁夏大尺度水稻面积遥感估算[J].农业工程学报,2017,33(15):200-209.
[4] 杨闫君,占玉林,田庆久,等.基于GF-1/WFV NDVI时间序列数据的作物分类[J].农业工程学报,2015,31(24):155-161.
[5] REED B C,BROWN J F,VANDERZEE D,et al.Measuring phenological variability from satellite imagery[J].Journal of Vegetation Science,1994,5(5):703-714.
[6] MURAKAMI T,OGAWA S,ISHITSUKA N,et al.Crop discrimination with multitemporal SPOT/HRV data in the Saga Plains,Japan[J].Remote Sensing,2001,22(7):1335-1348.
[7] BROWN J C,KASTENS J H,COUTINHO A C,et al.Classifying multiyear agricultural land use data from Mato Grosso using time-series MODIS vegetation index data[J].Remote Sensing of Environment,2013,130(4):39-50.
[8] ZHANG X,SUN R,ZHANG B,et al.Land cover classification of the North China Plain using MODIS/EVI time series[J].Isprs Journal of Photogrammetry & Remote Sensing,2008,63(4):476-484.
[9] 康峻,侯学会,牛铮,等.基于拟合物候参数的植被遥感决策树分类[J].农业工程学报,2014,30(9):148-156.
[10] 索玉霞,王正兴.基于决策树和MODIS植被指数时间序列的中亚土地覆盖分类[J].世界地理研究,2008,17(3):126-130.
[11] 张赫林,彭代亮,邓睿,等.基于Landsat时间序列数据的祁连山区域土地利用变化[J].北京工业大学学报,2017,43(5):665-676.
[12] 梁守真,陈劲松,吴炳方,等.应用面向对象的决策树模型提取橡胶林信息[J].遥感学报,2015,19(3):485-494.
[13] 杜宏娟,张磊,赵立斌,等.贺兰山东麓酿酒葡萄适宜越冬埋土深度调查研究[J].现代农业科技,2010(20):119,123.
[14] 马丽,徐新刚,刘良云,等.基于多时相NDVI及特征波段的作物分类研究[J].遥感技术与应用,2008,23(5):520-524.
[15] 张晶,占玉林,李如仁.高分一号归一化植被指数时间序列用于冬小麦识别[J].遥感信息,2017,32(1):50-56.
[16] 裴志远,杨邦杰.多时相归一化植被指数NDVI的时空特征提取与作物长势模型设计[J].农业工程学报,2000,16(5):20-22.
[17] 郭昱杉,刘庆生,刘高焕,等.基于MODIS时序NDVI主要农作物种植信息提取研究[J].自然资源学报,2017,32(10):1808-1818.
[18] 祁建春,李洪建,李莘莘,等.HJ-1-CCD与MODIS数据的NDVI比较分析[J].遥感信息,2013,28(3):44-49.
[19] 李文梅,覃志豪,李文娟,等.MODISNDVI与MODISEVI的比较分析[J].遥感信息,2010(6):73-78.
[20] 陈燕丽,罗永明,莫伟华,等.MODISNDVI与MODISEVI对气候因子响应差异[J].自然资源学报,2014,29(10):1802-1812.