The Subtropical Vegetation Leaf Area Index Model Constructed From Hyperspectral Data In South China

Leaf Area Index(LAI),which is one of the important physiological and biochemical parameters,is a very important structural parameter in the process of land surface and provides structural quantitative information for the initial energy exchange description of the vegetation canopy.LAI refers to the half of the total surface area of the leaf surface on the surface area per unit area.It is an important land characteristic parameter for crop growth models,net primary productivity models,and water cycle models.It plays a decisive role in crops such as precipitation interception,photosynthesis,and respiration.Not only can it provide real-time information on the growth of crops,but it is also a key input parameter for participating in the decision support system to estimate the growth model of crops.Therefore,scholars from all over the world attach great importance to the study of LAI.In the past half century,one of the hotspots in the field of quantitative remote sensing is the study of LAI inversion.Compared with traditional destructive sampling method to obtain vegetation LAI,hyperspectral remote sensing provides an opportunity for precise LAI inversion by virtue of its strong band continuity and large spectral information.Many scholars at home and abroad mainly use the spectral vegetation index method to carry out a lot of in-depth research on the relationship between crops hyperspectral and LAI,but the research on forest tree species in C hina is still in its infancy,few studies have combined with a certain type of Tree species development stud ies.Therefore,this paper conducts in-depth discussion and research based on the above issues.This study aimed at subtropical trees on the campus of South China Agricultural University,using the theory and methods of hyperspectral remote sensing monitor ing technology,based on the measured spectral reflectance of leaves and LAI data of trees,using the empirical regression model of LAI derived from vegetation indices,analyzing the variation of the spectral reflectance and LAI of leaves from trees,and t he quantitative relationship between LAI and leaf spectral reflectance,constructing an estimation model based on hyperspectral feature parameters and red edge parameters for LAI,completing hyperspectral remote sensing inversion monitoring of physical parameters of trees,in order to achieve key technologies such as timely,rapid,non-destructive and accurate information collection of subtropical vegetation growth in South C hina.The main conclusions of this study are as follows:(1)There was a highly significant correlation between the six hyperspectral characteristic variables and the leaf area index of tree species.The fitting equation R2 of the red side position reflectance(FREP)and the ratio vegetation index RVI all reached 0.8 or more.The correlation coefficients are 0.820 and 0.811,and the estimation models are y = 6.108e-2.503x,y = 0.115 x + 3.289。(2)In the LAI linear regression model,RVI and LAI have the highest degree of fit,reaching a very significant level;In LAI non-linear regression model,FREP and LAI have the highest degree of fit and reach a very significant level.Comparing two kinds of regression models can explain to some extent that there is more than a single linear relationship between hyperspectral vegetation index and leaf LAI of tree species.After accuracy verification,the Root Mean Squared Error(RMSE)estimated by FREP is only 0.13.This regression model is the best model for estimating LAI in typical subtropical tree species.