A Spectral Inversion Study Of Physiological Parameters Of Poplar Based On PROSPECT Model

In this study,a typical desert riparian forest plant,poplar,in the Tarim River basin of Xinjiang,as the only tree building in desert riparian ecosystems,plays a key role in regulating the balance of fragile ecosystems in the desert oasis transition zone.As the only source of water for poplar,groundwater is one of the most important factors affecting the life activities of poplar.In recent years,the groundwater level on both sides of the Tarim River has been gradually declining under the influence of human activities,causing a rapid reduction in the area of natural poplar forests,so further protection and research into poplar forests has become crucial to stabilising the ecological balance.In this study,five groundwater depth gradients of 0～2m,2～4 m,4～6 m,6～8 m and 8～10 m were set in the study area,and the hyperspectral data and physiological parameters of poplar under each depth gradient were obtained.The physiological parameters of poplar under different depths of groundwater burial and the spectral characteristics of the leaves were analyzed,as well as the canopy spectra of poplar were simulated using the PROSPECT-SAIL model,and the differences between the leaf spectra and the canopy simulated spectra were compared and analyzed,and finally the simulated spectra of the canopy and the physiological indexes were then used for model construction.It is expected to characterize the groundwater depth range of poplar and provide a theoretical basis for promoting ecological environment construction and vegetation restoration in arid areas.The main findings of the study are as follows:(1)The water content of poplar was mostly higher at the burial depths of 0～2 m and 2～4 m in different periods,with the trend of decreasing and then increasing with the enhancement of drought.The nitrogen content of the leaves showed an’M’-shaped variation with increasing depth of groundwater burial,decreasing under mild drought stress,but gradually increasing as the drought deepened.The changes in sodium(Na),magnesium(Mg),manganese(Mn),zinc(Zn),iron(Fe)and calcium(Ca)were not identical as the depth of groundwater burial increased.When poplar was subjected to drought stress,the Na,Mg and Zn contents in its leaves all decreased,while Mn,Fe and Ca contents all appeared to increase,all of which are manifestations of poplar’s resistance to drought stress.(2)The spectral curves of the same treatment at different burial depths in different periods are basically the same,and all of them have the typical"peak and valley"characteristics of vegetation.Among them,logarithmic,log-inverse and envelope removal spectra can amplify certain features in the visible or near-infrared bands,facilitating the study of the spectral variation characteristics of poplar at different burial depths.The spectral position parameters of poplar in different periods with different burial depths are less different,and the position of the green peak increases with the burial depth,and it moves in the long-wave direction,in which the red edge mainly shows obvious"blue shift"or"red shift"in August,September and October.The vegetation index was highly correlated with poplar water content,nitrogen and sodium with correlation coefficients greater than 0.4,while the correlations of all other indicators were poor.(3)The canopy spectra of poplar were simulated using the PROSPECT-SAIL model,and the canopy simulated spectra of poplar were compared with the leaf spectra.The results showed that the canopy spectra of poplar differed significantly with the depth of groundwater burial at the high-reflection platform,and the canopy spectra increased with the depth of groundwater burial at the water absorption valley at 970 nm,while the absorption valley feature weakened;Including the original canopy simulated spectra and six data processed canopy simulated spectra with poplar physiological indicators,A BP neural network(BP),partial least squares regression(PLSR)and principal component regression(PCR)model for estimating the physiological index content of poplar,nitrogen content and sodium content based on BP could pass the accuracy verification,but only the estimation model of sodium content achieved a good prediction ability,with modeling accuracy R~2 was 0.8372,RMSE was 2.1112 mg/g,and RPD was 2.4820,and the prediction accuracy R~2 was 0.8151,RMSE was 2.2508 mg/g,and RPD was 2.3025.(4)Eleven vegetation indices,including the Normalised Vegetation Index(NDVI),Difference Vegetation Index(DVI)and Simple Ratio Index(SRI),were constructed using simulated poplar canopy spectra,and three physiological indices(water content,nitrogen content and sodium content of poplar leaves)with high correlation with the vegetation indices were modelled to establish BP,PLSR and PCR models for estimating the content of poplar physiological indices.A model was developed to estimate the content of the physiological indicators by BP,PLSR and PCR.It was found that the physiological index content estimation model of vegetation index,only the BP-based sodium content estimation model of poplar leaves could be verified by accuracy with modeling accuracy of R~2 of 0.7254,RMSE of 2.7470 mg/g,and RPD of 1.9076,and prediction accuracy of R~2 of 0.7272,RMSE of 2.7238 mg/g,and RPD of 1.9026,and the model just reached rough predictive ability and poor predictive structure,reducing the model effect compared to the model for estimating physiological index content from canopy simulation spectra,but narrowing the difference in model effect between modeling methods.