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

Accueil du site → Master → Chine → 2018 → Scale Effect Analysis of Soil Salinization Modeling Based on Remote Sensing

Xinjiang University (2018)

Scale Effect Analysis of Soil Salinization Modeling Based on Remote Sensing

冯娟

Titre : Scale Effect Analysis of Soil Salinization Modeling Based on Remote Sensing

Auteur : 冯娟

Grade : Master’s Theses 2018

Université : Xinjiang University

Résumé
Remote sensing data is matched with the low resolution model in multi-model coupling and system integration.The scale mismatch is the difficulty of multi-scale verification based on high resolution data,direct up-scale verification based on ground sampling and cross validation of remote sensing products in different space scales.In order to better grasp the spatial information distribution of salinization,different scale data needs to be converted or fused.Analysis of the data acquisition,difficulty of model and feasibility,up-scale conversion can become the breach of the scale research.It is of great significance to reveal and grasp the complicated changes of salinization.Weigan-Kuqa oasis is the study area,which is typical oasis in arid area.Using different resolution remote sensing image of MODIS and Landsat as basic data,To screen the best index for the inversion of salt with the correlation between various spectral indices and salt.It obtains two scales of salinization quantitative inversion results,combining the trend surface method,transforming the space scale of Landsat image and MODIS image,by comparing the characteristic parameters and texture characteristics,to evaluate up-scale model.The main conclusions of this study are as follows:1.The spectral index is closely related to salinization,and the average spectral index obtained by Landsat has better correlation with salt.And the NDVI,SI,and SRSI indexes of the two images in all spectral indices have good association over the years,and the SRSI is the best,which is 0.641 for Landsat and 0.504 for MODIS.Using 5 transform forms of the linear and nonlinear model to establish better-accuracy model of SRSI and salt.At the same time,the results of qualitative inversion of landsat-srsi and modis-srsi indexes were consistent with the overall trend of the actual distribution by using density segmentation with salinization degree region.2.Space trend surface method is used to realize spatial scale conversion,through quantitative analysis the spatial scale conversion effect compared with theneighboring resampling method and window average method with multiple eigenvalues,and compared with the MODIS-SRSI salinization monitoring results to qualitative analysis the texture feature and spatial distribution.The result shows that the spatial scale conversion effect of trend surface analysis is better than the other two,and the scale conversion information is rich and can better describe the spatial distribution of salinization.3.Based on the trend surface method to get improvement combining with the method of mathematical statistics,by comparing the eigenvalues,texture,spatial distribution,and the consistency and error analysis acquiring the best improvement effect of weighting method based on the dominant classes,and arithmetic mean method for model getting no improvement.According to the above,the study gets through correlation between the average spectral index and salt screening of two different resolution inversion data of best salt index,and uses the method of trend surface to achieve pushing salt spatial scales,and combining with the dominant class weights and weight arithmetic mean method to improved push-up model.And the dominant class weight method improvement effect is obvious,that gives some reference to scale change research for salinization in arid areas.

Mots clés : soil salinization; Spectral index; Scale conversion; Trend surface analysis; Dominant class weight method;

Présentation (CNKI)

Page publiée le 5 mai 2019