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

Accueil du site → Master → Chine → 2018 → Hyperspectral Estimation Analysis Biomass and Total Nitrogen of Moss Crust

Xinjiang University (2018)

Hyperspectral Estimation Analysis Biomass and Total Nitrogen of Moss Crust


Titre : Hyperspectral Estimation Analysis Biomass and Total Nitrogen of Moss Crust

Auteur : 徐悦;

Grade : Master’s Theses 2018

Université : Xinjiang University

Moss biocrust is an important desert plant in arid and semi-arid regions.It has important ecological significance for wind protection and sand fixation,and improving soil moisture and nutrition.Chlorophyll,as the first stage of photosynthesis in the process of plant growth,plays an important role in controlling the energy transfer and material circulation of plants.The biomass of plants is an important indicator of plant physiological conditions,plant stress,and characterization of nutritional status,providing an important basis for the relationship between plants and the environment.The nitrogen content of plants is an important indicator for evaluating plant growth,and it is also an important factor for studying global changes and migration of nutrients.Hyperspectral remote sensing technology can monitor the dynamic changes of vegetation in real time in a short period of time and supplement the spectral research results of bryophyte biological crusts in arid regions,providing a scientific basis for the development and protection of desert ecosystems in arid regions.This research area is located in Gurban,which passes through Beishawo of Fukang City in the south of the special desert.It uses ASD spectroscopy to collect the hyperspectral data of bryophyte crusts,and the crust chlorophyll,biomass,and total nitrogen measured by the laboratory.Measured values,an estimation model based on hyperspectral data and biochemical parameters of moss biological crusts was established.Analyze the basic spectral characteristics of moss biological crusts in the study area,and the correlation between the measured spectral reflectance and the three biochemical parameters.Select the relationship between the bryophyte crust,chlorophyll,biomass,and total nitrogen.Higher characteristic bands.PLSR,BP neural network and support vector machine(SVM)models were used to establish an estimation model of bryophyte chlorophyll,biomass and total nitrogen based on measured spectral data.The correlations between biochemical parameters of moss biological crusts and reflectivity of remote sensing images were measured,and sensitive bands were selected through correlation analysis.An estimation model of moss bioclastic biochemical parameters based on Landsat-8 remote sensing image was established,and the coefficient R2 was determined.The RMSE of modeling total root mean square error and the residual prediction bias RPD were used as indicators of model accuracy evaluation,and the model was evaluated.Finally,the best model for predicting bryophyte biological crust based on hyperspectral data and remote sensing image data was obtained.The results show :(1)Using the measured spectral data,analyze the basic spectral features of moss biological crusts and the spectral curve features of three mathematical transformation forms of spectral reflectance : first differential,second differential and reciprocal logarithm.

Mots clés : Moss biocrust; measured spectral data; Landsat-8 remote sensing image; model;

Présentation (CNKI)

Page publiée le 20 mai 2019