Winter Wheat Growth Monitoring And Yield Estimation Based On UAV Multi-spectral Remote Sensing

This study used winter wheat as the main sample and collected vegetation index information related to its growth status.By applying advanced unmanned aerial vehicle multispectral remote sensing technology,the growth status of winter wheat can be quickly and timely observed,allowing for more accurate prediction of growth and yield estimation.This article analyzes the correlation between the aboveground biomass,LAI value,SPAD value,and vegetation index of winter wheat during the jointing,heading,filling,and maturity stages,and constructs the optimal model that can effectively monitor the growth changes of winter wheat at different growth stages in each period,as well as accurately predict its yield.The specific results are as follows:(1)A monitoring model based on aboveground biomass was constructed during the four growth stages of winter wheat.The study found that the monitoring model constructed based on RVI,SAVVI,NDVI,and DVI had the best effect,and the model’s R ~2 The values are 0.531,0.6217,0.6191,and 0.6217,respectively.These models are more suitable for monitoring aboveground biomass during the middle and late stages of winter wheat growth.(2)For the LAI value monitoring models of four growth stages of winter wheat,research has shown that the monitoring models constructed based on SAVI,NDVI,TVI,and DVI have better performance,and their R ~2 They are 0.5913,0.6371,0.6298,and0.5712,respectively.By using these models,better monitoring of LAI can be achieved.(3)For the monitoring models of SPAD values in the four growth stages of winter wheat,the monitoring models composed of NDVI,SAVI,GRVI,and CCCI performed well,and their R ~2 The values reached 0.6316,0.6318,0.6486,and 0.4986 respectively,indicating that these indicator combinations can effectively reflect the growth of winter wheat.The monitoring effect of these models is good during the jointing and heading stages,but the monitoring effect is poor during the filling and mature stages.(4)The equal weight method was used to build the comprehensive monitoring index of winter wheat growth,and three machine learning methods,random forest,BP neural network and convolution neural network,were used to build the comprehensive monitoring model of winter wheat growth in four growth periods,and the performance of various models was compared.In jointing stage,filling stage and mature stage,the model constructed by random forest method has the best effect ~2 They are 0.7644,0.7984,and0.7362,respectively.At the heading stage,the convolutional neural network model has the best effect,R ~2 It is 0.7816.(5)This study also explored the correlation between different vegetation indices and winter wheat yield,and established corresponding models.Research has found that different plant indicators have little impact on yield during the jointing and heading stages,making it difficult to accurately predict them;As the growth period increases,the final yield can be more accurately predicted.During the filling stage of winter wheat,monitoring models based on RVI,MNVI,and EVI performed the best,while during the mature stage,multivariate linear fitting models based on MNVI,SAVI,and RVI performed the best.According to this study and the model constructed by integrating various winter wheat growth stages,different vegetation indices are suitable for monitoring different growth parameters.NDVI and NDVI are the best choices for monitoring aboveground biomass,while LAI,DVI,and PPR are the best choices for monitoring SPAD values.In addition,the study also found that when detecting aboveground biomass,the modeling effect was better during the filling and mature stages.When monitoring LAI values,the modeling effect was better during the filling stage,and the modeling effect was better during the heading stage and the monitoring of chlorophyll content during the jointing stage.However,the modeling effect on chlorophyll content was poor during the mature stage.In order to improve accuracy,appropriate plant indicators and monitoring techniques should be selected based on the growth characteristics and developmental stages of plants.The results of this study provide certain reference significance for the growth status and yield prediction of winter wheat,thus providing new ideas for agricultural development.