Winter Wheat Aboveground Biomass and Yield Estimation Based on Multi-Source Information from UAV Imagery

GUO Yan; JING Yuhang; HE Jia; ZHANG Huifang; JIA Dewei; WANG Laigang

doi:10.15933/j.cnki.1004-3268.2023.12.017

Abstract：Winter wheat aboveground biomass is an important indicator to characterize yield,and rapid and non-destructive monitoring of winter wheat aboveground biomass by UAV remote sensing technology can grasp the growth of winter wheat in time,which is of great significance to the estimation of winter wheat yield.In this study,based on the spectral information and texture characteristics of UAV digital orthophoto map(DOM)and plant height(HDSM)extracted by digital surface model(DSM)during the booting,flowering,and filling stages of winter wheat,multiple linear regression(MLR),partial least squares regression(PLSR),and random forest(RF)methods were used to construct the winter wheat aboveground biomass and yield estimation models.The results showed that when using DOM information for winter wheat aboveground biomass estimation,the models constructed by integrating SIs+TFs were better than those constructed by a single spectral index or a texture feature;the accuracy of the winter wheat aboveground biomass estimation model constructed by incorporating HDSM information was improved,the RF model at the flowering stage had the highest accuracy;when incorporating the HDSM information into the aboveground biomass estimation of winter wheat,the accuracy of the estimation model was most obviously improved by TFs+HDSM.In the early estimation of winter wheat yield,the logarithmic function model had the highest accuracy in fitting the measured aboveground biomass to yield,and the R2 of the models for the booting,flowering,and filling stages were 0.87,0.88,and 0.92,respectively.The optimal models for aboveground biomass and yield estimation were coupled to estimate the yield,and the highest accuracy of the estimation model was obtained at the filling stage,with R2,RPD,and RMSE of 0.90,2.77,and 244.61 kg/ha,respectively.Therefore,the integration of multi-source information from UAV imagery and machine learning algorithms,can be used to quickly and efficiently estimate the aboveground biomass and yield of winter wheat,which is of great significance for the accurate formulation of food security policies.