Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Titre : Study on Hyperspectral Characteristics Identification of Main Desert Shrubs in Western Ordos

Auteur : 赵向苗

Grade : Master 2022

Université : Inner Mongolia Agricultural University

Résumé partiel
As an indispensable part of desert habitat,desert plants play an important role in environmental improvement.Understanding its spectral characteristics not only provides the basis for hyperspectral remote sensing identification of desert plants,but also plays an important role in plant diversity protection and growth monitoring in desert areas.In this study,we selected prunus mongolicum,Ileus mongolicum,Ileus mongolicum,Semilodiphylla and Tetropha chinensis growing under natural conditions in west Ordos National Nature Reserve,measured their canopy spectral data combined with chlorophyll content measurement,studied the spectral similarities and differences of the five shrubs,and analyzed their adaptation specificity.Mahalanobis distance method was used to select bands with different characteristics to identify different desert shrubs.The correlation between spectral characteristics and chlorophyll was analyzed,and the sensitive parameters were selected to predict the regression equation of chlorophyll to understand its growth status.The spectral characteristics in spring,summer,autumn and winter of a.oryzae,the only evergreen broad-leaved shrub in the reserve,were analyzed,the spectral characteristics of A.oryzae of different age levels were determined,and the relationship between the growth status of A.oryzae of different age levels and age levels was obtained.It is expected to provide basic data for spectral identification,growth monitoring and remote sensing interpretation of desert plants in the reserve.The main conclusions are as follows :（1）After pretreatment,the spectral curves of the five desert shrubs in the reserve still showed the spectral curve trend of common plants,but showed strong difference in the wavelength range of visible light.The significant differences are mainly concentrated in visible and near infrared wavelengths（550-654 nm,750nm-1070 nm,1180nm-1270 nm,1650nm-1790nm）.（2）After the original spectrum,first-order differential spectrum and envelope removal spectrum of five desert shrubs were identified by Mahalanobis distance method,the accuracy analysis was carried out by stepwise discriminant method.The results showed that the different bands selected by Mahalanobis distance method could effectively distinguish the five desert shrubs,and all the selected bands were desert plant characteristic bands.Among them,the spectrum recognition accuracy after second order differential transformation is the highest.

Mots clés : Hyperspectral ;Spectral identification ;Desert plants ;Western Ordos ;Chlorophyll content sensitive band ;

Présentation (CNKI)