基于无人机多光谱遥感的水稻二化螟冬前虫量测算  

Estimation of the pre-winter population of Chilo suppressalis in rice field based on unmanned aerial vehicle multi-spectral remote sensing

作　　者：曹梦娇 白石 唐攀攀 徐红星[3] 王晔青 周国鑫 CAO Mengjiao;BAI Shi;TANG Panpan;XU Hongxing;WANG Yeqing;ZHOU Guoxin(Jiaxing Soil Fertilizer,Plant Protection and Rural Energy Station,Jiaxing 314100,China;Big Data Technology Research Center,Nanhu Laboratory,Jiaxing 314100,China;Institute of Plant Protection and Microbiology,Zhejiang Academy of Agricultural Sciences,Hangzhou 310000,China;College of Advanced Agricultural Sciences,Zhejiang A&F University,Hangzhou 311300,China)

机构地区：[1]嘉兴市土肥植保与农村能源站,浙江嘉兴314100 [2]南湖实验室大数据技术研究中心,浙江嘉兴314100 [3]浙江省农业科学院植物保护与微生物研究所,浙江杭州310000 [4]浙江农林大学现代农学院,浙江杭州311300

出　　处：《江苏农业学报》2025年第2期305-312,共8页Jiangsu Journal of Agricultural Sciences

基　　金：浙江省重点研发计划项目(2022C02034);浙江省粮油产业技术项目;浙江省农业重大技术协同推广计划项目(2023ZDXT01-5)。

摘　　要：为实现稻田二化螟冬前虫量的精确测算,本研究在二化螟差异化防控的基础上,利用无人机获取水稻灌浆期和蜡熟期的双时相多光谱数据,并结合虫量稳定期的冬前虫量田间调查,基于线性回归、支持向量机回归、随机森林回归、岭回归、Lasso回归和贝叶斯回归等方法构建稻田二化螟冬前虫量的遥感估算模型。结果表明,灌浆期450 nm(b1)、660 nm(b3)波段的光谱反射率和蜡熟期的归一化植被指数(NDVI)与稻田二化螟冬前虫量存在极显著的线性相关;不同回归方法下,采用双时相数据建立的稻田二化螟冬前虫量遥感估算模型的估算值与观测值的相关性整体上优于单时相数据,其中,基于双时相遥感数据和随机森林回归模型建立的估算方法最佳,测试集和训练集的估算虫量和观测虫量相关系数分别达0.85和0.94,且此方法下稻田二化螟冬前虫量的估算结果更符合田间实际情况。本研究基于无人机技术建立的稻田二化螟冬前虫量估算方法,可为稻田二化螟的精确防控提供依据。In order to accurately estimate the pre-winter population of Chilo suppressalis in paddy fields,based on differentiated prevention and control of C.suppressalis,this study used unmanned aerial vehicle(UAV)to obtain double-phase multi-spectral data of rice at filling stage and wax ripening stage.And combined with the field survey of pre-winter population in the stable period of insect population,based on linear regression,support vector machine regression,random forest regression,ridge regression,Lasso regression and Bayesian regression,the remote sensing estimation model of pre-winter population of C.suppressalis in paddy fields was constructed.The results showed that the spectral reflectance of 450 nm(b1)and 660 nm(b3)bands at the filling stage and the normalized difference vegetation index(NDVI)at the ripening stage were in extremely significantly linear correlation with the pre-winter population of C.suppressalis in paddy fields.Under different regression methods,the correlations between the estimated value and the observed value of the remote sensing estimation model of the pre-winter population of C.suppressalis in rice fields established by using double-phase data were better than those of the single-phase data.Among them,the estimation method based on double-phase remote sensing data and random forest regression model was the best.The correlation coefficients between the estimated and observed population of C.suppressalis in the test set and the training set were 0.85 and 0.94,respectively,and the estimation results of the pre-winter population of C.suppressalis in rice fields under this method were more in line with the actual situation in the fields.Based on UAV technology,this study established an estimation method for the pre-winter population of C.suppressalis in paddy fields,which provided a basis for accurate prevention and control of C.suppressalis in paddy fields.

关 键 词：二化螟 冬前虫量 多光谱 随机森林 双时相 

分 类 号：S431.9[农业科学—农业昆虫与害虫防治]