基于多源遥感与GIS的马铃薯晚疫病早期空间预警技术OA
Early Spatial Warning Technology for Potato Late Blight Based on Multi-Source Remote Sensing and GIS
为解决马铃薯晚疫病传统监测方式时效性差、空间覆盖不足的问题,以贵州省贵定县为研究区,融合Sentinel-2多光谱影像、气象数据及地形数据,构建基于随机森林的病害早期识别模型,并结合GIS空间分析方法建立三级空间预警框架.研究发现红边类指数(NDRE、NDVI705)对晚疫病早期侵染响应最为敏感,在健康与轻度发病植株间的区分显著性优于NDVI(P<0.05),随机森林模型总体识别精度达88.3%(Kappa系数为0.847).基于随机森林的马铃薯晚疫病早期识别与空间预警技术体系经独立田间样点回溯验证,可提前3～5 d(平均4 d)发出病害预警,预警准确率达84.6%、漏报率9.2%、误报率6.2%.研究结果可为贵定县及周边冬作区马铃薯晚疫病精准防控提供决策支持.
This study integrated Sentinel-2 imagery,meteorological and terrain data to construct a random forest-based early disease identification model and a GIS-driven three-level spatial early warning framework for Guiding County,Guizhou Province in order to address the limitations of traditional potato late blight monitoring.Red-edge indices(NDRE and NDVI705)were most sensitive to early infection,distinguishing healthy from mildly infected plants better than NDVI(P<0.05);and the random forest model achieved an overall accuracy of 88.3%(Kappa=0.847).Verified against independent field samples,the framework issued warnings 3-5 days(4 days on average)in advance with a warning accuracy of 84.6%(omission and commission errors of 9.2%and 6.2%,respectively).These findings would provide decision support for precise potato late blight control in winter potato-growing areas of Guiding County and surrounding regions.
陈天慧
贵州省黔南州贵定县新巴镇人民政府,贵州 黔南 551306||国家开放大学,北京 100039
农业科技
多源遥感地理信息系统马铃薯病害
multi-source remote sensinggeographic information systempotatodiseases
《中国马铃薯》 2026 (2)
155-160,6
评论