基于遥感的乡镇级棉花面积提取与长势监测研究

THE STUDY OF COTTON AREA EXTRACTION AND GROWTH MONITORING USING REMOTE SENSING AT TOWNSHIP LEVEL

棉花面积及长势信息是棉花生产管理的重要依据。本文以山东广饶县丁庄镇为研究区,以CBERS01和HJ1B卫星图像为信息源,选取棉花信息较为突出的时相,经几何校正、腌膜、图像增强等预处理,在分析研究区典型地物光谱特征的基础上,采用决策树分类方法提取棉花种植面积。利用2009年棉花生长期4个不同时相的遥感图像,分析棉花生长过程中植被指数的变化规律和棉花长势的时空变化。结果显示,各时相棉花种植面积的提取精度均在90%以上,棉花长势随各生长时期呈现不同时空分布规律,监测结果与调查统计得到的棉花单产分布相符。研究证明,该方法能够实现对内陆小尺度棉花种植面积及长势信息的适时获取及动态监测,为棉花产量早期预测提供依据,对提高棉花栽培生产与管理水平有积极意义。

Cotton area and growth information are important basis for cotton production and management.This article chose Dingzhuang Town in Guangrao County of Shandong Province as the study area,CBERS01 and HJ1B satellite images as the information source.Selecting similar temporal characteristics with obvious cotton information.The area of cotton was acquired by decision tree classification approach through image pre-processing,including geometric accurate correction,masking and image enhancement,based on the analysis of spectrum characteristics of typical objects in the study area.Four different time remote sensing images of cotton growing season in 2009 was selected to analyze changes of vegetation index cotton conditions in spatial and temporal.The results showed that the extraction accuracy of cotton area were more than 90%.With the growing of cotton in all period,the growth of cotton showed different rules in spatial and temporal distribution.Monitoring results were consistent with cotton yield in distribution by the survey.This study showed that this method can achieve timely access and the dynamic monitoring to small-scale cotton area and growth information.It also can provide the basis for early prediction of cotton production.This research has positive significance to improve the cultivation of cotton production and management.