决策树分类提取宁夏耕地复种指数
|
摘要
针对运用Savitzky-Golay滤波处理的增强型植被指数曲线会出现伪峰或将真实峰去掉的情况,对复种指数的提取造成误差的问题,提出一种基于ENVI软件的决策树分类法快速准确提取宁夏复种指数的分析方法。在对2009年和2010年中等分辨率MOD13Q1的EVI进行预处理后,进行SG滤波法重构时间序列曲线,然后利用TIMESAT软件的TSF_seas2img功能模块提取宁夏植被的关键物候信息,最后基于物候信息运用决策树分类,分离出宁夏一年一熟及一年两熟的作物面积。对提取的结果利用高分辨率的2009年Landsat-5TM影像以及统计年鉴提供的数据进行了验证。结果表明,宁夏主要种植方式为一年一熟,少量的一年两熟区域大多集中于固原市。基于关键物候信息提取宁夏复种指数的精度较高,能够比较准确地反映宁夏复种作物信息,表明中分辨率遥感影像在提取复种指数方面也能发挥积极作用,丰富了复种指数的分析,也为该方面的研究与应用提供了参考。
The use of Savitzky-Golay filtering EVI(enhanced vegetation index)curve will appear spurious peaks or the real peak removed,which causes the error in the extraction of multiple cropping index.In this paper,a decision tree classification method based on ENVI software is proposed for the rapid and accurate analysis of the multiple cropping index in Ningxia.After preprocessing the 2009 and 2010 medium resolution MOD13 Q1 EVI,reconstruct the time sequence curve by SG filtering method,then utilize the TSF_seas2 img TIMESAT software function module to extract the key Ningxia vegetation phenology information.Finally,use the decision tree classification to isolate the crop area of Ningxia one year one ripe and two crops a year.The results are verified by using high resolution Landsat-5 TM images in 2009 and the statistical yearbook data.The results show that the main cropping system in Ningxia is one year one ripe,and a small amount of a year two cooked areas are mostly focused on the natural Guyuan city.The extraction of Ningxia multiple cropping index based on the key phenology information has higher accuracy,which can accurately reflect the Ningxia double-cropping information,showing that moderate resolution remote sensing image can also play apositive role in extracting multiple crop index.
The use of Savitzky-Golay filtering EVI(enhanced vegetation index)curve will appear spurious peaks or the real peak removed,which causes the error in the extraction of multiple cropping index.In this paper,a decision tree classification method based on ENVI software is proposed for the rapid and accurate analysis of the multiple cropping index in Ningxia.After preprocessing the 2009 and 2010 medium resolution MOD13 Q1 EVI,reconstruct the time sequence curve by SG filtering method,then utilize the TSF_seas2 img TIMESAT software function module to extract the key Ningxia vegetation phenology information.Finally,use the decision tree classification to isolate the crop area of Ningxia one year one ripe and two crops a year.The results are verified by using high resolution Landsat-5 TM images in 2009 and the statistical yearbook data.The results show that the main cropping system in Ningxia is one year one ripe,and a small amount of a year two cooked areas are mostly focused on the natural Guyuan city.The extraction of Ningxia multiple cropping index based on the key phenology information has higher accuracy,which can accurately reflect the Ningxia double-cropping information,showing that moderate resolution remote sensing image can also play apositive role in extracting multiple crop index.
引文
[1]AUDSLEY E,PEARN K R,SIMOTA C,et al.What can scenario model-ing tell us about future European scale agricultural land use,and what not?[J].Environmental Science&Policy,2006,9(2):148-162.
[2]STEHFEST E,HEISTERMANN M,PRIESS J A,et al.Simulation of global crop production with the ecosystem model Day Cen[J].Ecological Modelling,2007,209(4):203-19.
[3]卞新民,冯金侠.多元多熟种植制度复种指数计算方法探讨[J].南京农业大学学报,1999,22(1):11-15.
[4]梁书民.我国各地区复种发展潜力与复种行为研究[J].农业经济问题,2007,28(5):85-90.
[5]左丽君,张增祥,董婷婷,等.耕地复种指数国内外研究进展[J].自然资源学报,2009,24(3):553-560.
[6]闫慧敏,曹明奎,刘纪远,等.基于多时相遥感信息的中国农业种植制度空间格局研究[J].农业工程学报,2005,21,(4):85-90.
[7]左丽君,张增祥,董婷婷,等.MODIS/NDVI和MODIS/EVI在耕地信息提取中的应用及对比分析[J].农业工程学报,2008,24(3):167-172.
[8]张晗,任志远,韦振锋.2000—2012年陕西省复种指数时空分布及潜力分析[J].干旱区研究,2015,32(3):492-498.
[9]王立辉,黄进良,孙俊英.基于时序MODIS-EVI监测华中地区耕地复种指数[J].长江流域资源与环境,2010,19(5):529.
[10]梁守真,马万栋,施平,等.基于MODIS NDVI数据的复种指数监测:以环渤海地区为例[J].中国生态农业学报,2012,20(12):1657-1663.
[11]吴岩.中国东部季风区耕地复种指数遥感监测研究[D].南京:南京信息工程大学,2008.
[12]杨忍,刘彦随,陈玉福,等.环渤海地区耕地复种指数时空变化遥感反演及影响因素探测[J].地理科学,2013,33(5):588-593.
[13]左丽君,张增祥,董婷婷,等.MODIS/NDVI和MODIS/EVI在耕地信息提取中的应用及对比分析[J].农业工程学报,2008,24(3):167-172.
[14]王正兴,刘闯,陈文波,等.MODIS增强型植被指数EVI与NDVI初步比较[J].武汉大学学报(信息科学版),2006,31(5):407-410.
[15]王志慧,李世明,刘良云,等.基于MODIS NDVI时间序列的土地覆盖分层分类方法研究[J].遥感技术与应用,2013,28(5):910-919.
[16]裴雪原,臧淑英,那晓东.MODIS MOD13Q1植被产品介绍及快速预处理[J].哈尔滨师范大学(自然科学学报),2004,30(2):65-67.
[17]陈俊,王文,李子扬,等.Landsat-5卫星数据产品[J].遥感信息,2007,32(3):85-88.
[18]左丽君,董婷婷,汪潇,等.基于MODIS/EVI的中国北方耕地复种指数提取[J].农业工程学报,2009,25(8):141-146.
[19]SAKAMOTO T,VAN N.Detecting temporal changes in the extent of annual flooding within the Cambodia and the Vietnamese Mekong Delta from MODIS time-series imagery[J].Remote Sensing of Environment,2007,109(3):295-313.
[20]JAKUBAUSKAS M E,LEGATES D R,KASTENS J H.Crop identification using harmonic analysis of time-series AVHRR NDV data[J].Computers and Electronics in Agriculture,2002,37(1/2/3):127-139.
[21]潘瑞鑫,赖格英,李秀娟,等.基于MODIS-NDVI的梅江流域耕地复种指数提取研究[J].江西科学,2012,30(6):764-768.
[22]辜智慧.中国农业复种指数的遥感估算方法研究:基于SPOT/VGT多时相NDVI遥感数据[D].北京:北京师范大学,2003.
[23]唐鹏钦,姚艳敏,吴文斌,等.基于遥感技术的耕地复种指数研究进展[J].中国农业资源与区划,2010,31(2):21-27.
[24]苏伟,姜方方,朱德海,等.基于决策树和混合像元分解的玉米种植面积提取方法[J].农业机械学报,2015,46(9):289-295.
[25]王正海,张红军.SAM和决策树结合的Hyperion数据分类方法[J].武汉科技大学学报(自然科学版),2006,29(5):478-481.
[2]STEHFEST E,HEISTERMANN M,PRIESS J A,et al.Simulation of global crop production with the ecosystem model Day Cen[J].Ecological Modelling,2007,209(4):203-19.
[3]卞新民,冯金侠.多元多熟种植制度复种指数计算方法探讨[J].南京农业大学学报,1999,22(1):11-15.
[4]梁书民.我国各地区复种发展潜力与复种行为研究[J].农业经济问题,2007,28(5):85-90.
[5]左丽君,张增祥,董婷婷,等.耕地复种指数国内外研究进展[J].自然资源学报,2009,24(3):553-560.
[6]闫慧敏,曹明奎,刘纪远,等.基于多时相遥感信息的中国农业种植制度空间格局研究[J].农业工程学报,2005,21,(4):85-90.
[7]左丽君,张增祥,董婷婷,等.MODIS/NDVI和MODIS/EVI在耕地信息提取中的应用及对比分析[J].农业工程学报,2008,24(3):167-172.
[8]张晗,任志远,韦振锋.2000—2012年陕西省复种指数时空分布及潜力分析[J].干旱区研究,2015,32(3):492-498.
[9]王立辉,黄进良,孙俊英.基于时序MODIS-EVI监测华中地区耕地复种指数[J].长江流域资源与环境,2010,19(5):529.
[10]梁守真,马万栋,施平,等.基于MODIS NDVI数据的复种指数监测:以环渤海地区为例[J].中国生态农业学报,2012,20(12):1657-1663.
[11]吴岩.中国东部季风区耕地复种指数遥感监测研究[D].南京:南京信息工程大学,2008.
[12]杨忍,刘彦随,陈玉福,等.环渤海地区耕地复种指数时空变化遥感反演及影响因素探测[J].地理科学,2013,33(5):588-593.
[13]左丽君,张增祥,董婷婷,等.MODIS/NDVI和MODIS/EVI在耕地信息提取中的应用及对比分析[J].农业工程学报,2008,24(3):167-172.
[14]王正兴,刘闯,陈文波,等.MODIS增强型植被指数EVI与NDVI初步比较[J].武汉大学学报(信息科学版),2006,31(5):407-410.
[15]王志慧,李世明,刘良云,等.基于MODIS NDVI时间序列的土地覆盖分层分类方法研究[J].遥感技术与应用,2013,28(5):910-919.
[16]裴雪原,臧淑英,那晓东.MODIS MOD13Q1植被产品介绍及快速预处理[J].哈尔滨师范大学(自然科学学报),2004,30(2):65-67.
[17]陈俊,王文,李子扬,等.Landsat-5卫星数据产品[J].遥感信息,2007,32(3):85-88.
[18]左丽君,董婷婷,汪潇,等.基于MODIS/EVI的中国北方耕地复种指数提取[J].农业工程学报,2009,25(8):141-146.
[19]SAKAMOTO T,VAN N.Detecting temporal changes in the extent of annual flooding within the Cambodia and the Vietnamese Mekong Delta from MODIS time-series imagery[J].Remote Sensing of Environment,2007,109(3):295-313.
[20]JAKUBAUSKAS M E,LEGATES D R,KASTENS J H.Crop identification using harmonic analysis of time-series AVHRR NDV data[J].Computers and Electronics in Agriculture,2002,37(1/2/3):127-139.
[21]潘瑞鑫,赖格英,李秀娟,等.基于MODIS-NDVI的梅江流域耕地复种指数提取研究[J].江西科学,2012,30(6):764-768.
[22]辜智慧.中国农业复种指数的遥感估算方法研究:基于SPOT/VGT多时相NDVI遥感数据[D].北京:北京师范大学,2003.
[23]唐鹏钦,姚艳敏,吴文斌,等.基于遥感技术的耕地复种指数研究进展[J].中国农业资源与区划,2010,31(2):21-27.
[24]苏伟,姜方方,朱德海,等.基于决策树和混合像元分解的玉米种植面积提取方法[J].农业机械学报,2015,46(9):289-295.
[25]王正海,张红军.SAM和决策树结合的Hyperion数据分类方法[J].武汉科技大学学报(自然科学版),2006,29(5):478-481.