县域棉花信息遥感提取与棉田精确化管理分区研究  

Extraction of Cotton Information Using Remote Sensing and Precision Management Zoning at County Scale

作　　者：李敏[1] 赵庚星[1] 蔡明庆[2] 赵倩倩[1] 唐建[1] 

机构地区：[1]山东农业大学资源与环境学院,山东泰安271018 [2]山东建筑大学热能工程学院,济南250101

出　　处：《自然资源学报》2012年第11期1971-1980,共10页Journal of Natural Resources

基　　金：高校博士点基金项目(20103702110010);山东省高等学校科技计划项目(J11C11)

摘　　要：棉田信息化的精确管理是提高棉花生产管理水平及我国数字农业发展的客观需求。论文以山东省夏津县为研究区,选取棉花4个不同生长期的HJ-1卫星遥感影像,通过物候历分析和主要农作物的光谱特征,确定棉花识别最佳时相,建立提取模型,获取棉花种植区信息。建立棉花花铃期遥感影像的特征光谱指数,并与实测的5种主要土壤养分数据结合,运用相关分析、主成分分析和K均值聚类分析方法,划分棉花管理区,并对分区结果的合理性进行差异显著性评价分析。结果显示:棉花种植区提取的最佳时相为蕾期,提取精度达到93%以上;优化土壤调节植被指数(OSAVI)能够较好地反映棉花的长势信息,同时与土壤养分具有显著的相关性;利用主成分分区指标将研究区棉田划分为3个管理区:管理一区占总棉田面积的24.67%,土壤有机质和氮磷钾含量最高,棉花长势旺盛;管理二区占47.02%,土壤养分含量中等,棉花长势一般;管理三区占28.31%,土壤肥力最低,棉花长势较差;各管理分区内土壤养分的变异系数都显著减小,分区间土壤养分和光谱指数均值差异显著,显示了分区的合理性。该研究可为实时准确的棉花栽培管理与决策提供科学依据。Precision management of cotton information is an objective need to improve the management of cotton production and promote digital agricultural development.This article took Xiajin County of Shandong Province as a study area and selected remote sensing images of HJ-1 satellite in four different cotton growing seasons.The best phase to identify cotton was determined by analyzing phonological calendar and spectral characteristics of major crops.An extraction model was established to acquire cotton planting area information.Combining characteristic spectral index established at flower and boll stage with five kinds of main soil nutrients measured,cotton management zones were divided by correlation analysis,principal component analysis and K-means cluster analysis.The significant differences of the partition results were analyzed to evaluate the reasonableness.The results showed that the best phase to extract cotton area was bud stage and extraction accuracy was more than 93%.The OSAVI could reflect cotton growth information well and was significantly correlated with soil nutrients.Cotton field of the study area was divided into three management zones using principal component partition indicators.The first management zone accounted for 24.67% of the total area with the highest content of soil organic matter and N,P,K,and its cotton grew well.The second management zone accounted for 47.02% with medium soil nutrients and general cotton grew.The third management zone accounted for 28.31% with the lowest soil fertility and poor cotton grew.In each management partition,coefficient variation of soil nutrients decreased significantly.There were significantly statistical differences between mean soil nutrients and spectral data in different defined management zones,which showed the partition＇s high reasonableness.This research provides scientific basis of real-time and accurate cotton cultivation management for decision-making.

关 键 词：棉花 遥感提取 土壤养分 管理分区 

分 类 号：S127[农业科学—农业基础科学]