基于高光谱的春玉米大斑病害遥感监测指数选择

发布时间：2018-04-19 08:38

本文选题：遥感 + 病害　；参考：《农业工程学报》2017年05期

【摘要】：玉米大斑病是春玉米主要病害之一,采用地面光谱观测的方式构建遥感监测指数,是实施区域遥感监测的基础,也是卫星传感器谱段设计的主要依据。该文在陕西省眉县设计了人工控制小区试验,针对高抗、抗、感和高感4个品种,通过人工接种不同浓度大斑病分生孢子的方法,获得了正常、轻微、中度以及严重等4个病害感染梯度小区,并在春玉米抽雄、吐丝、乳熟以及成熟4个生长期进行了地面高光谱观测。为了实现对春玉米大斑病的遥感监测,该项研究在春玉米冠层光谱数据基础上,分析了不同种植区不同生长期春玉米冠层光谱反射率和光谱一阶微分特征,并以此确定了大斑病敏感波段位置以及病害适宜监测期,同时根据敏感波段位置的光谱特征构建了专门的春玉米大斑病的遥感监测指数,最后结合180个光谱观测样本,对比了所提指数以及其他病害指数与病害严重度之间的相关性,并通过聚类分析了所建遥感指数的稳定性。研究结果表明,乳熟期的春玉米大斑病在红边波谱内的响应较为敏感,尤其红边核心区(725~740 nm)的光谱一阶微分与病害严重程度间存在明显地单调变化关系,具有非常显著的负相关性;同时,该文所提病害监测指数与病情指数具有较高的相关性,其相关系数达到了0.995 0,最后结果表明利用红边一阶微分指数的对病害程度的聚类总体精度达到100.0%,指数值分布稳定性也更高,具有在遥感监测业务中应用的潜力。
[Abstract]:Maize spot disease is one of the main diseases of spring maize. It is the basis of regional remote sensing monitoring and the design of satellite sensor spectrum to construct remote sensing monitoring index by means of ground spectrum observation.An artificial control plot experiment was designed in Meimei County, Shaanxi Province. For four varieties with high resistance, resistance, sensitivity and high sensitivity, normal and slight results were obtained by inoculating conidial spores with different concentrations.Four moderate and severe diseases were infected in gradient plots, and the ground hyperspectral observations were carried out in 4 growing stages of spring maize, such as heading, silking, milking and maturing.Based on the spectral data of spring maize canopy, the spectral reflectance and first-order differential characteristics of spring maize canopy in different growing periods were analyzed.According to the spectral characteristics of the sensitive bands, a special remote sensing monitoring index of spring maize spot was constructed. Finally, 180 spectral observation samples were used.The correlation between the index and the severity of the disease was compared, and the stability of the remote sensing index was analyzed by clustering.The results showed that the response of spring maize blotch in milk-ripening stage was more sensitive to the red edge spectrum, especially the relationship between the first order differential spectrum of red edge core and the severity of the disease.At the same time, the disease surveillance index proposed in this paper has a high correlation with disease index.The correlation coefficient is up to 0.995. The result shows that the clustering accuracy of the first order differential index of red border is 100.00.The stability of index value distribution is also higher, which has the potential to be applied in remote sensing monitoring.
【作者单位】：中国农业科学院农业资源与农业区划研究所;
【基金】：国家重点研发计划“粮食作物生长监测诊断与精确栽培技术”课题“作物生长与生产力卫星遥感监测预测”(2016YFD0300603)
【分类号】：S435.13;S127

【参考文献】