遥感信息
遙感信息
요감신식
2015年
2期
105-110
,共6页
夏玉米%MODIS-EVI%时序谱段组合%黄淮海
夏玉米%MODIS-EVI%時序譜段組閤%黃淮海
하옥미%MODIS-EVI%시서보단조합%황회해
summer maize%MODIS-EVI%time-series spectral group%Huang-Huai-Hai
针对夏玉米难以精确识别和分类时数据冗余的问题，提出夏玉米最佳时序谱段组合识别模式。基于时间序列的 MODIS EVI 数据，利用马氏距离（Jeffries-Matusita Distance，J-M）和构造的加权平均分离距离（Weighted Average Separability，DWAS ）得到夏玉米区别于其他作物的增强型植被指数（Enhanced Vegetation Index，EVI）时序图像，并将其进行组合，从而构建了识别夏玉米的最佳时序谱段组合；并利用其提取了2001年～2010年黄淮海地区的夏玉米。同时利用典型试验样区 TM 影像分类结果和野外实地采集样地对提取结果进行了验证：典型样区 MODIS 与 TM 分类误差最大为11．4％，空间匹配度达到91．29％，基于 TM 数据提取的典型样区夏玉米种植面积（STM ）与基于 MODIS 数据得到的夏玉米种植面积（S M ）的精度均大于70％；地面506个样地检验的混淆矩阵总体精度达到81％。
針對夏玉米難以精確識彆和分類時數據冗餘的問題，提齣夏玉米最佳時序譜段組閤識彆模式。基于時間序列的 MODIS EVI 數據，利用馬氏距離（Jeffries-Matusita Distance，J-M）和構造的加權平均分離距離（Weighted Average Separability，DWAS ）得到夏玉米區彆于其他作物的增彊型植被指數（Enhanced Vegetation Index，EVI）時序圖像，併將其進行組閤，從而構建瞭識彆夏玉米的最佳時序譜段組閤；併利用其提取瞭2001年～2010年黃淮海地區的夏玉米。同時利用典型試驗樣區 TM 影像分類結果和野外實地採集樣地對提取結果進行瞭驗證：典型樣區 MODIS 與 TM 分類誤差最大為11．4％，空間匹配度達到91．29％，基于 TM 數據提取的典型樣區夏玉米種植麵積（STM ）與基于 MODIS 數據得到的夏玉米種植麵積（S M ）的精度均大于70％；地麵506箇樣地檢驗的混淆矩陣總體精度達到81％。
침대하옥미난이정학식별화분류시수거용여적문제，제출하옥미최가시서보단조합식별모식。기우시간서렬적 MODIS EVI 수거，이용마씨거리（Jeffries-Matusita Distance，J-M）화구조적가권평균분리거리（Weighted Average Separability，DWAS ）득도하옥미구별우기타작물적증강형식피지수（Enhanced Vegetation Index，EVI）시서도상，병장기진행조합，종이구건료식별하옥미적최가시서보단조합；병이용기제취료2001년～2010년황회해지구적하옥미。동시이용전형시험양구 TM 영상분류결과화야외실지채집양지대제취결과진행료험증：전형양구 MODIS 여 TM 분류오차최대위11．4％，공간필배도체도91．29％，기우 TM 수거제취적전형양구하옥미충식면적（STM ）여기우 MODIS 수거득도적하옥미충식면적（S M ）적정도균대우70％；지면506개양지검험적혼효구진총체정도체도81％。
The optium model of time-series spectral group was put forward to solve accurate identification and classification for summer maize.The MODIS EVI data were picked up from time-series MODIS EVI images using J-M distance and weighted average separability distance (DWAS )structured.On the basis of images separated,the model of identification for summer maize was constructed.Then summer maize in Huang-Huai-Hai from 2001 to 2010 were classified based on the classifying model.And the results were verified by that of TM in test regions and samples collected from fields.The maximal error of spatial distribution in test regions between MODIS and TM is 11.4%,the total consistency is 91.29%,the accuracy of area between SM and STM were higher than 70%,the overall accuracy of confusion matrix between ground samples and results derived MODIS is 81%.
