Remote Sensing Monitoring Of Maize Lodging Disaster Based On Unmanned Aerial Vehicle Images

Maize is one of the main food crops in China.According to the National Bureau of Statistics,the maize planting area is 41.28 million hectares and the production is 261 million tons in 2019.Its yield fluctuation is of great significance to the national food security.Lodging stress is one of the important disasters affecting maize yield due to unreasonable cultivation methods and weak lodging resistance of maize varieties,as well as the frequent occurrence of strong winds and torrential rains in recent years.The conventional observation method based on artificial field investigation can no longer meet the timeliness needs of large-scale disaster monitoring.Ultra-low altitude unmanned aerial vehicle(UAV)system has become the focus of crop lodging monitoring because of its advantages of low cost,high resolution and image acquisition under cloud.It is of great practical significance to explore the hyperspcctral response mcchanism of maize lodging and realize the monitoring of maize lodging grades based on UAV images.It has been paid more and more attention by the agricultural insurance industry,agricultural management departments,new business entities and large grain growers.It is of great practical significance to obtain the spatial distribution information of maize lodging grades timely and accurately for yield loss assessment,post-disaster management and insurance claims settlement rapidly.The purpose of this study is to explore the ability of UAV imaging technology to monitor maize lodging disaster.With the support of maize lodging control experiments and actual lodging cases,the response rules between canopy structure parameters and hyperspectral characteristic parameters with different lodging stress were analyzed,the diagnosis of maize lodging disaster based on UAV hyperspectral image was realized,and a method of rapid monitoring maize lodging grade by UAV multispectral image was proposed.The following conclusions are drawn:(1)The changes of stem-leaf ratio and canopy spectral response of maize under different lodging stress were analyzed.The results showed that the reflectance of stem is higher than leaf.The canopy reflectance of maize with different lodging stress intensity was compared and analyzed,and it was found that the maize canopy reflectance increased with the increase of lodging stress intensity.In the view of spectral detection at the canopy scale,with the increase of lodging stress grades,the proportion of maize leaves blocking each other and the shadow caused by lodging and the gap between maize plants gradually decreased,and the proportion of stem and leaf gradually increased,especially in the stem components;in addition,the pixel reflectance is higher in the canopy image.(2)The change characteristics of canopy chlorophyll density(CCD)of maize with different lodging grades were analyzed.The original canopy spectra were transformed by continuous wavelet and vegetation index.The monitoring model of maize CCD under lodging stress was constructed by the sensitive characteristic parameters of canopy spectra,wavelet coefficient and vegetation index.It was found that the model accuracy of CCD based on wavelet coefficients and vegetation index are high(the determination coefficient R2>0.6).And the model accuracy of vegetation index is the highest(R2=0.63,RMSE=0.36g/m3).The maize lodging grades were evaluated according to the model inversion results of CCD.(3)With the support of multispectral image of UAV in the real lodging area,the transformation of principal components,texture features,vegetation indices,and combination of texture features and vegetation indices based on original multispcctral image were performed.Different lodging grades of maizc based on five feature images were extracted by maximum likelihood supervised classification.Classification accuracy was evaluated using a confusion matrix.The results showed that compared with a multispectral image,the principal components,texture features,the combination of texture features and vegetation indices were improved by varying degrees.The overall accuracy and Kappa coefficient of the combination of texture features and vegetation indices were the highest,which were 86.61%and 0.83,respectively.The classification results were increased by 3.03%and 0.04.In this study,the UAV images were used to analyze the spectral features with different lodging grades of maize.The monitoring model of CCD under lodging stress based on the hyperspectral characteristic parameters was completed.Different lodging grades of maize were classified and the remote sensing monitoring of different lodging types of maize was realized based on the multi-spectral image feature combination.It verifies the application ability of UAV monitoring technology for remote sensing monitoring of maize lodging disaster of different lodging grades.