基于多时相Sentinel-2数据的中牟县冬小麦种植面积提取

Extraction of Winter Wheat Planting Area in Zhongmu County Based on Multi-temporal Sentinel-2 Data

农作物的遥感识别监测是开展农作物长势监测、估产的前提和出发点,是农业遥感的焦点之一.利用遥感数据监测作物种植面积,日益成为农业管理及政策制定的重要技术支撑.以中牟县为研究区,选取2020—2021年期间的6幅Sentinel-2遥感影像数据,主要依据农作物的物候期光谱特征曲线变化和时序归一化植被指数曲线的变化,利用面向对象提取方法提取中牟县冬小麦面积及空间分布.将提取结果与谷歌地球高分辨率影像结合野外调查得到的验证样本进行精度验证,评价结果显示:总体精度93.33%,Kappa系数86.38%,结果能够较好地反映中牟县冬小麦种植面积及空间分布情况.该方法为大区域农作物的面积提取提供了重要方法参考.

Remote sensing identification and monitoring of crops is the premise and starting point of crop growth monitoring and yield estimation,and it is one of the focal points of agricultural remote sensing.Monitoring crop planting area by remote sensing satellite data is becoming a very important approach in the agricultural production and management.In this study,we selected six Sentinel-2 remote sensing image data of Zhongmu County during 2020-2021 to analyze the changes of spectral characteristics and temporal normalized vegetation index(NDVI)of major crops at different growing stages,and used object-oriented extraction method to extract the area and spatial distribution of winter wheat in Zhongmu County.The accuracy was verified by the extraction results and Google Earth high-resolution images combined with the validation samples obtained from field investigation.The confusion matrix was used to evaluate the classification accuracy.The evaluation results showed that the overall accuracy was 93.33%,and the Kappa coefficient was 86.38%,which could better reflect the planting area and spatial distribution of winter wheat in Zhongmu County.This method provides an important reference for crop area extraction in other large regions.