Hyperspectral Response Analysis And Quantitative Monitor On The Water Characteristic Parameters Of Winter Wheat Under Drought Conditions

As an important index to measure the physiological condition of crop water stress,water content is of great significance for scientific monitoring of crop water profit and loss as well as rational irrigation.In this study,water stress tests in different growth stages of winter wheat were carried out to determine the canopy spectral reflectance of winter wheat under water stress at different growth stages,and to analyze the water characteristic parameters and changes in the canopy spectrum under different water treatments.Through the correlation analysis of the full spectrum band,spectral characteristic band,"trilateral" parameters,vegetation index and moisture characteristic parameters,the response characteristics of winter wheat water characteristic parameters and canopy spectrum were studied.By using PLSR,SVM and RF methods,hyperspectral estimation models based on full spectral band,characteristic band,"trilateral" parameters and vegetation index were established,and these model were verified and evaluated.The main conclusions are as follows:(1)The trends of plant water content,leaf water content and equivalent water thickness in different growth stages were similar,and all gradually decreased with the advancement of growth;Under different water stress,plant water content,leaf water content,equivalent water thickness and canopy water content will gradually decrease with the increase of water stress degree.(2)Under different water stress,the spectral reflectance at different growth stages basically showed an upward trend with the increase of water content,In the near infrared region,spectral reflectance increases with the increase of water content,While for "trilateral" parameters,the yellow and blue parameters decrease,and the red parameter increases.(3)The water characteristic parameters and canopy spectrum were positively correlated at about 400-760 nm,and negatively at 761-1350 nm.The absolute correlation coefficients of plant water content,leaf water content and canopy spectrum were all 0.8,and the absolute correlation coefficients of equivalent water thickness,canopy water content and canopy spectrum were all 0.7.(4)Based on the “trilateral” parameters,the PLSR model had the best effect on plant water content,leaf water content and canopy water content,while SVM model had the best effect on equivalent water thickness,The estimation effect of canopy water content was the best(R2=0.753,RMSE=0.024,RPD=2.133).For the model constructed in full spectrum bands,the SVM model had the best effect on plant water content,equivalent water thickness and canopy water content,and the PLSR model had the best effect on leaf water content,The effect of equivalent water thickness was the best(R2=0.646,RMSE=0.036,RPD=1.834).For the model constructed by the characteristic waveband,SVM model had the best effect on plant water content,while PLSR model had the best effect on leaf water content,equivalent water thickness and canopy water content.Based on the model of vegetation index,SVM model can predict the four water indexes well,The estimation effect of canopy water content was the best(R2=0.673,RMSE=0.026,RPD=1.714).