Monitoring Of Maize Lodging Disaster Based On Multi-source Remote Sensing Data

Lodging stress is one of the main natural disasters in the process of maize growth in China.Due to the characteristics of maize varieties,cultivation management,diseases and insect pests,external environment and other factors,maize is vulnerable to lodging threat in nutrition(heading stage)and reproductive(filling stage)growth stage.Lodging seriously affects the grain yield,quality and mechanical harvesting ability of maize,limits the high yield and high quality of maize,becomes one of the main natural disaster stresses in the process of maize growth,and thus to seriously threaten food security.In recent years,the rapid development of remote sensing technology provides scientific technical means in lodging disaster monitoring and disaster assessment because of its rapid,objective,economic and regional advantages.It is of great significance to the field management of lodging maize,the evaluation of yield loss and the implementation of related measures after the disaster.In this study,the National Precision Agriculture Demonstration Research Base in Xiaotangshan,Changping District,Beijing is selected as the lodging area control test research area,and Gaocheng District,Shijiazhuang City,Hebei Province is the wild nature lodging research area.Taking the lodging maize population as the research object,under different lodging periods and different lodging treatments,the change characteristics of the physiological index parameters of the lodging maize population were analyzed based on the canopy hyperspectral data of the lodging maize population.The inversion model of physiological index of lodging maize population was constructed,and the mechanism response analysis between physiological indexes and spectral parameters of lodging maize population was discussed.At the same time,the lodging disaster index which can characterize the disaster intensity of maize was constructed by using the field investigation data,and the changes of remote sensing image information of maize population structure parameters before and after lodging were analyzed by GF-1 optical remote sensing images.According to the change information of remote sensing images,the lodging disaster monitoring model was constructed,and the regional scale maize lodging disaster monitoring based on real remote sensing images was carried out.The main contents and conclusions of this study are as follows:(1)Based on the lodging plot control experiment,according to the field measured results of the actual lodging type and growth period of maize,the changes of population physiological indexes of maize after lodging were studied from the relevant agronomic parameters such as leaf area index(LAI),leaf area density(LAD)and height.The canopy spectrum of maize was measured by ASD high spectrometer,and the response relationship between physiological indexes and spectral characteristic parameters of lodging maize population was analyzed.The continuous projection algorithm,optimal index method and Akaike information criterion were used to analyze the data,and the hyperspectral estimation model of maize population physiological index under lodging stress was constructed.It was found that lodging stress could lead to the change of maize canopy structure,and the canopy spectral reflectance of different intensity lodging maize was different,and the canopy spectral reflectance of maize was different from that of non-lodging maize at different growth stages.The canopy spectral reflectance increased in varying degrees after lodging,and the more serious the lodging was,the higher the spectral reflectance was.Lodging maize itself has a certain recovery ability,and under lodging stress,lodging maize plants decline rapidly,resulting in a gradual decrease in canopy spectral reflectance in the later stage,LAI changes are not obvious under lodging stress,but LAD constructed by introducing plant height is related to lodging intensity,the more serious the lodging is,the higher the LAD value is,which is proportional to the lodging intensity.Compared with LAI,the effect of LAD model constructed by various spectral features is better than that of LAI,and single feature,and the model constructed by combined features is more stable;LAD is suitable to be used as a parameter to characterize the physiological characteristics of lodging maize population,which can provide necessary prior knowledge for remote sensing monitoring of lodging stress disaster in maize.(2)Based on the regional scale maize lodging disaster evaluation based on lodging disaster index,the regional maize lodging disaster monitoring is studied by using GF-1 optical remote sensing image.By using the measured lodging ratio and inclination angle after lodging,the lodging degree evaluation index is constructed,and the lodging disaster grade is divided,and the difference changes of band reflectance and vegetation index before and after lodging are calculated.The optimal variable combination sensitive to the evaluation index of maize lodging degree was selected to construct the remote sensing monitoring model of maize lodging disaster.It is found that the more serious the lodging disaster is,the higher the spectral reflectance is,and the optimal combination of variables selected by CARS algorithm is spectral band Δ B2,ΔB4 and vegetation index △ NDVI,△ GNDVI,△ BNDVI,△NPCI;The lodging disaster index based on lodging ratio and lodging inclination angle can effectively judge the actual lodging occurrence.The model trains R2=0.96,RMSE=0.08,verifies that R2=0.88,RMSE=0.12,and the prediction accuracy of lodging grade is 79%.It can be used as an evaluation index for remote sensing to monitor the degree of lodging disaster.The optical GF-1 image is used to monitor the maize lodging disaster at the regional scale,and the remote sensing mapping of the maize lodging disaster grade at the regional scale is realized,which provides technical support for disaster assessment and post-disaster remediation.