【摘要】：鑒于土地利用中耕地類型的遙感光譜特征差異大,以及我國北方農(nóng)牧交錯帶中撂荒地、耕地、裸地和草地混淆嚴重,耕地信息的獲取難度大、精度低,提出了利用長時間序列遙感數(shù)據(jù),通過多級再分類技術方法(multilevel reclassification,MLRC)提取可耕種區(qū)域。首先利用最大似然法對長時間序列的多期遙感數(shù)據(jù)進行監(jiān)督分類,提取出耕地區(qū)域,之后在初級分類的基礎上,通過統(tǒng)計不同區(qū)域在多期分類結果中被判定為耕地的次數(shù)進而確定可耕作區(qū)域的范圍。通過對閃電河濕地實驗區(qū)的研究表明,利用MLRC方法的精度達到了82.56%。
[Abstract]:In view of the great difference in remote sensing spectral characteristics of cultivated land types in land use, and the abandoned land in the agricultural and pastoral ecotone of northern China, the confusion between cultivated land, bare land and grassland is serious, the acquisition of cultivated land information is difficult and the precision is low. In this paper, using long time series remote sensing data, the arable area is extracted by multilevel reclassification technique (MLRC). Firstly, the maximum likelihood method is used to monitor and classify the multi-period remote sensing data of long time series, and then the cultivated land area is extracted, and then on the basis of primary classification, The range of arable areas was determined by counting the number of times that different regions were identified as cultivated land in the results of multi-period classification. The study on the experimental area of Lightning River Wetland shows that the accuracy of MLRC method is 82.56.
【作者單位】： 北京師范大學地理科學學部;北京師范大學全球變化與地球系統(tǒng)科學研究院;河北農(nóng)業(yè)大學林學院;
【基金】：河北省科學技術廳項目(14274203D)
【分類號】：S127
，

本文編號：2130038