摘要
摘要:运用遥感精准识别技术提取灌溉耕地可以为区域内农业耗水管理提供关键数据支持，但受限于遥感数据的时空分辨率，对于光谱特性相似容易混淆的灌溉耕地作物识别有一定难度。为此，本研究通过增强型自适应反射率时空融合模型（Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model，ESTARFM）对MODIS和Landsat数据进行融合，得到察汗淖尔流域时间分辨率为8 d、空间分辨率为30 m的NDVI数据，并与同期真实Landsat NDVI进行对比验证，相关系数为0.94。利用HANTS滤波建立ESTARFN NDVI时间序列数据，选取灌溉耕地样本通过支持向量机进行灌溉耕地空间分布的提取，以弥补数据源、特征提取等限制因素对复杂灌溉耕地提取的空缺。结果表明：流域内灌溉耕地总面积为1958.24 km²，商都、兴和、尚义、康保和化德5县耕地总面积占流域耕地总面积的94%，灌溉耕地分别为616.67 km²、337.36 km²、409.85 km²、290.93 km²和239.38 km²，主要种植葵花、甜菜、马铃薯等生长季从4月初到9月底的长生育期作物和生长季从5月初到8月初的蔬菜；张北、察哈尔右翼前旗、后旗和镶黄旗耕地占流域耕地总面积为6%，4县灌溉耕地面积共64.05 km²。最后，通过真实样本进行验证，总分类精度为93.18%，Kappa系数为0.91。结果表明，用该数据融合模型获得的NDVI时间序列能反映作物真实变化情况，并且使用支持向量机提取察汗淖尔流域灌溉耕地效果较好。
关键词:灌溉耕地/
NDVI/
ESTARFM/
数据融合/
支持向量机
Abstract:Remote sensing extraction technology for crops is a promising research method for remote sensing applications. However, it is limited by the spatial and temporal resolution of the remote sensing data; it is difficult to identify crops with similar spectral characteristics. In this study, the enhanced spatial and temporal adaptive reflectance fusion model (ESTARFM) was used to fuse MODIS and Landsat data to obtain normalized difference vegetation index (NDVI) time series data of the Chahannur Basin with a temporal resolution of 8 days and a spatial resolution of 30 m. The correlation coefficient between the ESTARFM NDVI and the Landsat NDVI during the same period was 0.94. Harmonic analysis of the NDVI time series filter was used to smooth the ESTARFM NDVI data. Finally, the irrigated arable land was extracted using a support vector machine (SVM) with irrigated cultivated land samples. The results showed that the total area of irrigated arable land in the watershed is 1958.24 km². The total area of Shangdu, Xinghe, Shangyi, Kangbao, and Huade Counties accounts for 94% of the watershed, and the irrigated arable land is 616.67 km², 337.36 km², 409.85 km², 290.93 km², and 239.38 km², respectively. The region primarily grows sunflowers, beets, potatoes, and other long-growing crops from the beginning of April to the end of September and vegetables (including celery, Chinese cabbage, cabbage) from the beginning of May to the beginning of August. Cultivated land in Zhangbei, Chahar Youyiqianqi, Houqi, and Xianghuangqi Counties accounts for 6% of the total cultivated land in the basin, and the total irrigated land in the four counties is 64.05 km². Real sample verification indicates that the total classification accuracy is 93.18%, and the Kappa coefficient is 0.91. The results show that the NDVI time series obtained from the data fusion model reflects real changes in crops, and the use of SVMs to extract irrigation arable land in the Chahannur Basin is suitable.
Key words:Irrigated land/
Normalized difference vegetation index (NDVI)/
Enhanced spatial and temporal adaptive reflectance fusion mode (ESTARFM)/
Data fusion/
Support vector machine (SVM)



PDF全文下载地址:

http://www.ecoagri.ac.cn/article/exportPdf?id=7c64a49f-aa64-4192-ad20-236464a24082