The Study To Monitor Cotton Canopy Water Condition Using Hyperspectral Remote Sensing Based On Photosynthetic Physiological Parameters

[Research object]: In this study,the spectral data and photosynthetic parameters of two varieties of cotton were determined in the key growth period under different water treatments,Construct the fitting equations.Finally,the vegetation index and cotton yield and fiber quality index were analyzed,and the best vegetation index could be selected to estimate cotton yield and quality.Which provided the effective,fast and non-destructive diagnosis and monitoring of cotton growth process and the further study on the monitoring of cotton growth,production and quality in the future.[Research methods]: Canopy hyperspectral dates of the two cotton cultivars with four levels water treatments were obtained in a field experiment by ASD spectrometer radiometer at cotton key growth stages,calculations of reflectance spectrum data,such as the original spectral reflectance,first derivative,the "three edge" parameters,vegetation indexes.At the same time,we obtained the photosynthetic physiological indexes of two cotton cultivars in key growth period by Li-6400,such as net photosynthetic rate?Pn?,gas conduction?Gs?,transpiration rate?Tr?,Intercellular CO2 concentration?Ci?.Finally,we obtained the yield and fiber quality in the harvest period,Analyzed the correlation between spectral variables and the indexes,and established a remote sensing model to monitor the growth status of cotton.[Research results]:?1?The spectral reflectance showed a significant difference in the infrared band under different water treatments."Three edge" curve appeared obvious “peak” and“valley”,the "red edge" area showed a strong significant difference than the "red edge" slope under different water treatments.So the reflectance in the infrared band?the obvious “peak” and“valley” and the "red edge" area can be used to monitor the moisture content of cotton.?2?The trend of photosynthetic parameters increased first and then decreased in the whole growth period,reached the maximum in flowering boll stage,With the different water treatments,photosynthetic parameters increased gradually with the increase of irrigation amount,showed significant difference in the later growth stage,the difference in Xinluzao 45 was more obvious than Xinluzao 62.?3?There was a significant negative correlation between the spectral reflectance and Pn at 780 nm,It reached the maximum positive correlation which is 0.53.The first derivative reflectance has a significant positive correlation with Pn at 693 nm,with a correlation coefficient of 0.67.Furthermore,the correlation between estimated Pn and measured Pn at the 780 nm reflectance and the first derivative reflectance at 693nm is analyzed.The results show that the first derivative reflectance accuracy is higher than the spectral reflectance(r_{693 nm???????Pn-??Pn}=0.71**,RMSE=2.42;r_{780 nm?????Pn-??Pn}=0.67**,RMSE=2.76,n=40)?The estimation accuracy is over 80%.The"red edge"slope was correlated with Pn,Tr,The correlation is shown as rXinluzao 45 Dr-Tr=0.66**,"Red edge"area and Pn,Tr,Gs showed significant correlation,The correlation is shown as rXinluzao 45SDr-Tr=0.68**,and the"red edge"area has higher accuracy of Pn than"red edge"slope(r_SDr??Pn-??Pn=0.67**,RMSE=2.51,r_Dr??Pn-??Pn=0.63**,RMSE=2.77,n=40)?The best estimation accuracy is over 85%.It showed the best correlation between the vegetation index RENDVI and RVI of the infrared nearinfrared band reflectance ratio and the photosynthetic characteristic parameters.Which RENDVI and Pn showed a strong significant linear correlation(r_{Xinluzao 45 RENDVI-Pn}=0.89**,r_{Xinluzao 62 RENDVI-Pn}=0.71**,n=20),it showed a significant logarithm correlation with Tr(r_{Xinluzao 45 RENDVI-Tr}=0.78**,r_{Xinluzao 62 RENDVI-Tr}=0.64**,n=20).RVI also showed a strong correlation with Pn,Tr(r_{Xinluzao 45 RVI-Pn}=0.87**,rXinluzao 45 RVI-Tr=0.78**;rXinluzao 62 RVI-Pn=0.70**,r_{Xinluzao 62 RVI-Tr}=0.72**,n=20).Do analysis between the predicted Pn by RENDVI and RVI and measured Pn,the r value reached the most significant level(r_RENDVI??Pn-??Pn=0.82**,RMSE=2.13;r_RVI??Pn-??Pn=0.86**,RMSE=1.7,n=40).The best results are predicted by RVI,and the estimation accuracy is over 90%.Therefore,the cotton vegetation index RENDVI and RVI can be used as a good indicator to predict the photosynthetic characteristic parameters,and the model of the two can be used to monitor the growth of cotton.?4?The NDVI and RVI have highest correlation with lint yield at the full flowering stage(r_NDVI-yield)=0.97**,r_RVI-yield=0.97**,n=20),NDVI and RVI also have highest correlation with fiber quality at the flowering stage.So constructing the model in flowering stage can predict the yield and quality better.