基于MODIS-EVI的内蒙古沿黄平原区作物种植结构分析

【目的】获取内蒙古沿黄平原区农业资源信息,提高大区域中低分辨率遥感影像上作物种植结构分类的效率和精度。【方法】利用多时相作物分类方法在内蒙古沿黄平原区构建增强型植被指数(EVI)时间序列曲线,依据不同农作物EVI时间序列曲线的差异对6种主要农作物小麦、玉米、葵花、西葫芦、番茄和苜蓿的种植结构进行了分析。【结果】小麦、玉米、葵花、西葫芦、番茄、苜蓿和其他作物的用户精度分别为:79.59%、80%、83.67%、78.18%、75.93%、82.22%、68.75%,制图精度分别为78%、80%、82%、86%、82%、74%、66%,农作物总体分类精度达到78.29%,Kappa系数为0.747。经统计沿黄灌区种植的玉米实地统计面积合计为7912.17 km²,文中提取出的面积为7412.75 km²,相对误差为6.31%。【结论】通过对MODIS-EVI时间序列的分析可以较为准确地识别大尺度测区内的主要作物,该方法能够在大区域中低分辨率影像上实现较好的分类结果。

Using MODIS-EVI to Identify Cropping Structure in Plains Along the Yellow River in Inner Mongolia

【Background and objective】Remote sensing has been increasingly used to monitor cropping structure at regional and continental scales because of its low cost and high precision.The purpose of this paper to use it to extract agricultural resource information in plains proximal to the Yellow river in Inner Mongolia,with a view to help improve agricultural managements in this region.【Method】Multi-temporal crop classification method considering seasonal change in characteristic traits of different crops was used to identify temporal change in the crops,from which a time series of enhanced vegetation index(EVI)were constructed for crops in the Yellow river plain in Inner Mongolia.Based on the difference in EVI between different crops,we estimated the planted areas of wheat,corn,sunflower,zucchini,tomato and alfalfa in the studied region.【Result】Comparison with ground-true data showed the accuracy of the method for estimating planted areas of wheat,corn,sunflower,zucchini,tomato,alfalfa and other crops was 79.59%,80%,83.67%,78.18%,75.93%,82.22%and 68.75%respectively,with their associated mapping accuracy being 78%,80%,82%,86%,82%,74%,74%and 66%,respectively.For example,data from the statistical bureau showed the planted maize areas in the studied region were 7912.17 km²,while the areas estimated from our method were 7412.75 km²,giving rise to a relative error of 6.31%.The overall classification accuracy for all crops was 78.29%,with the kappa coefficient being 0.747.【Conclusion】The MODIS-EVI time series can be used to identify cropping structure at large scale with a reasonable accuracy,and can be used to aid agricultural and irrigation managements.