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

Accueil du site → Master → Chine → 2018 → A Method for Detecting the Damage of Apocheima Cinerarius Erschoff in Populus Euphratica Forest Based on Remote Sensing

University of Chinese Academy of Sciences (2018)

A Method for Detecting the Damage of Apocheima Cinerarius Erschoff in Populus Euphratica Forest Based on Remote Sensing

汪航;

Titre : A Method for Detecting the Damage of Apocheima Cinerarius Erschoff in Populus Euphratica Forest Based on Remote Sensing

Auteur : 汪航;

Grade : Master’s Theses 2018

Université : University of Chinese Academy of Sciences

Résumé
Populus euphratica is the earliest type of the desert riparian forest in the Tarim River Basin in Xinjiang.It has been limiting the Taklamakan Desert from expansion,protecting the habitat of people in the Tarim River Basin and creating living space for animals and plants in the harshest desert environment.In recent years,the damage of Apocheima Cinerarius Erschoff in Xinjiang province has seriously threatened the survival and green barrier function of natural Populus euphratica.Due to the large coverage and high tree body of Populus euphratica,traditional monitoring methods are difficult to meet the requirement of rapid monitoring of pest.In this thesis,Landsat OLI images were employed for land cover classification of the study area.Damage monitoring and analysis of Apocheima Cinerarius Erschoff were performed based on the Populus euphratica forest distribution from classification.NDVI time series data sets were reconstructed from MODIS data from 2014 to 2017,and the NDVI time series curves were further fitted by three filters.The sensitivities of NDVI values in the characterization of the pests damage were analyzed temporally and spatially.Damaged areas of Apocheima Cinerarius Erschoff were extracted from 2014 to2017.Combining climatic and environmental factors,the study analyzed the spatial-temporal distribution patterns of damage areas and provided reference data for such work like pest limiting and Populus protection in the future.The thesis contents and conclusions were as follows :(1)Information extraction of Populus euphratica:The red and near-infrared bands spectral reflectance of Populus euphratica is different from those of cropland.The reflectance values of various wavelengths of shrub are lower than those of Populus euphratica,and the reflectance values of various wavelengths of mixed types of shrubs and Populus euphratica are intertwined with shrubs and Populus euphratica forests.Using spectral index can effectively distinguish various land cover types.NDVI performed better than EVI and RVI after experiments.In addition,the autumn images were helpful in discrimination of cropland,Populus euphratica and shrubs.The combination of texture information further improved the classification accuracy,in which three features of mean,correlation,and contrast were effective,and the result was best when the sliding window size is 3.Support vector machine classification outperformed the maximum likelihood classification,with a classification accuracy of 87.2%.The classification results show that the Populus euphratica forests are distributed along Tarim River riparian corridors,with an area about 166.78 Km 2,and the mixed forest of Populus euphratica is about 273.08 Km 2.(2)The method for detecting the damage of Apocheima Cinerarius Erschoff based on remote sensing images:The NDVI time series can be used to characterize the growth status of Populus euphratica forest because the maximum,minimum and growth trends of the curve from random sites are highly consistent with the vegetation growth characteristics in the Tarim River Basin.NDVI value is sensitive to external environmental factors,especially rainfall.The filter fitting algorithm can effectively remove the non-image-quality noise points.In the three fitting algorithms,the asymmetric Gaussian function fitting method and the dual Logistic fitting function method are excessively over-fitting to the original values.The S-G filter retains more detail information of the original NDVI curve on the basis of removing outliers,and is suitable for the NDVI time series analysis in this study.Compared with the fitted results,the original NDVI curve had obvious and regular fluctuations during the pests damage period from March to May,and comparisons between the NDVI curve of the fly defense areas and the undefended defense areas further validated the fluctuations were associated with the insects of the Apocheima Cinerarius Erschoff.Damage can be detected by monitoring mutation points based on this.(3)Monitoring results of the damaged area:Pests damage distribution of Populus euphratica forest from 2014 to 2017 shows that disaster area is spatially clustered and change dynamically without obvious regularity.

Mots clés : Populus Euphratica; Apocheima Cinerarius Erschoff damage; Landsat 8; MODIS NDVI; Time series analysis;

Présentation (CNKI)

Page publiée le 18 mai 2019