Study On Rapid Monitoring Water Of Winter Wheat Based On Imaging Spectroscopy

Aiming to screen out optimal spectral parameter,building accurate and universal monitoring model and providing technical support for water monitoring and precision irrigation of winter wheat,this study selected winter wheat as experimental material and set up four water treatments,i.e.,W1?0 mm?,W2?30 mm?,W3?60 mm?,W4?90 mm?,respectively.Using the hyperspectral imager,we analyzed canopy spectral reflectance characteristics and spectral image features of winter wheat in different growth stages and built fitting models of plant water content,leaf relative water content,soil water content,leaf water potential and soil water potential,which were tested later.The results were as follows:1 Monitoring model of plant water content of winter wheatThe results of model fitting indicated that determination coefficients of image parameter calculated vaule in different growth stages were all superior to those of vegetation indexes.As for monitoring of plant water content in winter wheat whole period,fitting model based on combination of spectral parameter was better than single spectral parameter.The best fitting model of plant water content in winter wheat whole period was based on GSAO_green?Gray Scale Anti Atmospheric Operation Value?and GNO?Gray Normalized Operation Value?,i.e.,y=-1.04×GSAO_green+0.07×GNO+0.75,whose determination coefficient was 0.831 and MRE,RMSE were 2.27%and 2.1%,respectively.Thus,this study was likely to provide technical reference for accurately monitoring plant water content of winter wheat.2 Monitoring model of leaf relative water content of winter wheatThe results of model fitting indicated that monitoring models based on Calculated value of image parameter at jointing stage,heading stage,flowering stage,early-filling stage and late-filling stage were all superior to those of models based on vegetation index.The best fitting model of leaf relative water content in winter wheat whole period was based on GWC?Gray Water Calculation Value?and GSAO_green,i.e.,y=0.56×GWC-0.87×GSAO_green+0.16,whose determination coefficient was 0.836 and MRE,RMSE were 4.22%and 4.6%,respectively.Thus,this study was likely to provide technical reference for accurately monitoring plant water content of winter wheat.3 Monitoring model of soil water content of farmlandThe results of model fitting indicated that the best monitoring depth for simulating soil water content of farmland was 0-20 cm soil layer and the best fitting model in winter wheat whole period was based on GREO2?Gray Red Edge Operation Value?and mGSRO705?Gray Improved Specific Value of Red Edge Operation Value?,i.e.,y=-0.30×GREO2+0.51×mGSRO705-0.04,whose determination coefficient was 0.815 and MRE,RMSE were13.52%and 1.6%,respectively.The results showed that this method of monitoring soil water content was feasible and it provided a certain theoretical basis and technical support for remote sensing of soil water content.4 Monitoring model of leaf water potential of winter wheatThe results of model fitting indicated that monitoring models based on Calculated value of image parameter in different growth stages were all superior to those of models based on vegetation index and as for the fitting model in winter wheat whole period,the former was better than the latter.Furthermore,the best fitting model of leaf water potential in winter wheat whole period was based on GSOSA?Grey Scale Value of Optimization Soil-adjusted?and GEO2?Gray Scale Operation Value of Enhanced?,i.e.,y=1.62×GSOSA+2.66×GEO2-4.84,whose determination coefficient was 0.849 and MRE,RMSE were-8.02%and21.0%,respectively.5 Monitoring model of soil water potential of farmlandThe results of model fitting indicated that with increase in soil depth,the monitoring effect of optimal model reduced and the best monitoring depth for simulating soil water potential of farmland was 0-20 cm soil layer.Moreover,the best fitting model of soil water potential in winter wheat whole period was based on GREO1?Gray Scale Operation Value of Red-enhanced?and GREO2 and the determination coefficient was high.The fitting model was y=-0.75×GREO2+2.80×GREO1-4.79,whose determination coefficient was 0.800and MRE,RMSE were-15.67%and 13.9%,respectively,after tested.