Analysis On Characteristics Of Desert Adult Populus Euphratica Leaves Water Content In Spring Using Ground Hyperspectral Data

Moisture is one of the important factors what controls the plant respiration, photosynthesis and biomass.The plant lack of moisture will directly both reflect the growth of plant and morphological structure. In desert areas, saline plants have the ability of windbreak and sand-fixation, can curb the spread of desertification, maintain the stability of ecosystem. So the real-time monitoring for the moisture of saline plant growth conditions,is very important. It can provide the basic for recovery of ecological environment and evaluation in the study area. With the formation and the continuous development of remote sensing technology, since the 1980 s, hyperspectral remote sensing is gradually matured, which is a large area simultaneous observations, obtained the data timeliness, continuous, non-destructive and other characteristics, then make remote sensing technology is widely used in the plant monitoring, which provides a new way to realize the real-time monitoring of a large area of water content of plant.In this paper, desert Populus euphratica leaves of Ebinur Lake Basin acquisition were measured reflectance and leaves water content of indoor spectrum, on the original spectral data of 9 mathematical transformations include:(the original reflectance first derivative differential(FDR),reflectance reciprocal(1/R)、the reflectance reciprocal first derivative differential((1/R)’)、reflectance logarithms(lg R)、a derivative of logarithm of reflectance differential((lg R)’)、the reflectance reciprocal logarithmic(lg(1/R))、the reflectance reciprocal logarithmic first derivative differential(lg(1/R)’)、reflectance RMS( R)and the root mean square derivative differential reflectivity(( R)’).Analysis of 10 kinds of spectral reflectance curve characteristics. Analysis the spectrum curve characteristics of leaves water content in different water content and different soil moisture content and different elevation. The spectral reflectance and Populus euphratica leaves moisture content to do correlation analysis, to extract the characteristic bands, using the multiple stepwise regression(SMLR), principal component regression(PCR) and partial least squares regression(PLSR) three kinds of modeling method to establish model. The main conclusion as following:(1)This paper makes 9 mathematical transformation(FDR、1/R、(1/R)’、lg R、(lg R)’、lg(1/R)、lg(1/R)’、 R �'�( R)’)of the original spectral data, these mathematical transformation can magnify the spectrum characteristics of information, analysis of the relationship between the spectral reflectance and leaves water content for better. By analyzing the spectral curve characteristics found that with the increase of leaves water content, the Populus euphratica leaves spectral reflectance is less; With the decrease of the soil moisture content, the Populus euphratica leaves spectral reflectance will increase; With the increasing of elevation, the Populus euphratica leaves spectral reflectance will increase.(2)10 spectral indices have shown the correlation coefficient in the near infrared band is higher than that of visible light wave band, the1/R、(1/R)’、lg R、lg(1/R) spectral emission rate dependence and leaves water content of the best, the maximum value of the absolute value of the correlation coefficient all above 0.89; The characteristic bands selected for 1300~1450nm, 1470~1600nm, 1830~1950nm.(3)Multivariate stepwise regression(SMLR), the principal component regression(PCR) and partial least squares regression(PLSR) three modeling methods, modeling is the best effect of log spectral reflectance first-order differential transformation after the establishment of the PLSR prediction model, modeling of R2 is 0.76, RMSE is 6.7%, R2 is 0.89, RMSE test as of 5.0%; Principal component regression model is more precise than the stepwise regression model, the prediction model of the highest accuracy is the reciprocal of reflectance transformation after the establishment of the modeling of R2 is 0.78, RMSE is 6.4%, R2 is 0.83, RMSE test is 6.1%; Three kinds of modeling method of the highest accuracy of model in using different mathematical transformation, illustrates the different mathematical transformation is applicable to different modeling methods.