Hyperspectral Inversion Model For Leaf Moisture Content Of Walnut

Water is the basic factor affecting plant life activities.Leaf water content is an important indicator to describe plant water status.Rapidly and accurately monitoring or diagnosing leaf moisture content and timely adjustment water management measures are of great significance to ensure the normal growth,development and fruit yield of walnut trees.Field walnut trees and potted walnut saplings in the hilly area of Northeast China were used as research objects in this study.The data of potted water control test and leaf dehydration test were used to analyze the response characteristics of walnut leaf hyperspectral to water change,extract the water sensitive band and sensitive spectral indices,construct the inversion model of walnut leaf water content,campare the differences among the five varieties of walnut leaf moisture content and hyperspectral reflectance,such as“Guanmao”,“Shizitou”,“Luguo 2”,“Liaoning 1”and“Liaoning 5”,and verify the proposed spectral index and its estimation model in this study.The main results are as follows:?1?The hyperspectral reflectance of leaves of walnut seedlings increased with the decrease of leaf relative moisture content?LMC?and leaf dry matter moisture content?DMC?in the range of 750?1400 nm,and increased with the decrease of the leaf equivalent water thickness?EWT?in the range of 1405?2500nm.The correlation between the spectral reflectance of the first derivative and the leaf water content was greatly improved.The correlation between reflectance and spectral indices proposed by the predecessors and EWT is higher than that of LMC and DMC.Therefore,EWT was preferred as the moisture index,and using the first derivative reflectance to improve the inversion accuracy of leaf moisture.?2?The hyperspectral reflectance of“Xiangling”leaves increased with the decrease of leaf water content in the range of 760?2500 nm,and the moisture sensitive wavelength was 1456 nm and d1978 nm.The inversion accuracy of the spectral index estimation model was higher than that of the sensitive band model.The regression coefficient of d RSI_?1310,1405?and d NDSI_?1310,1405?and leaf moisture content were the highest?R²=0.74?,and the model verification accuracy were berrer.Therefore,EWT=-0.0102×d RSI²_?1310,1405?+0.1340×d RSI_?1310,1405?-0.0012 and EWT=0.0510×d NDSI²_?1310,1405?+0.1628×d NDSI_?1310,1405?+0.1111 were the inversion models of walnut leaf moisture content.?3?The spectral reflectance characteristics of five different varieties of walnuts,were similar,and the correlation between leaf moisture content and leaf reflectance was not affected by moisture indicators.The correlation between leaf moisture and reflectance of different varieties of walnut was significant in 621?703 nm and 1244?2500 nm,and different varieties of walnut leaves had different water-sensitive bands in the range of 704?1243 nm,showing obvious differences between varieties.?4?The spectral indices d RSI_?1310,1405?and d NDSI_?1310,1405?and their regression models constructed by the"Xiangling"varieties were verified by the data of reflectance and water content of the other five different varieties of walnut in the same climate zone.The regression coefficient of d RSI_?1310,1405?and d NDSI_?1310,1405?and moisture content of other varieties of walnut leaves was higher,and the applicability was better.In practical applications,adjustting the coefficients of the regression model in different varieties to improve the accuracy of the model inversion.