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

Accueil du site → Master → Chine → 2018 → Monitoring and Prediction of Salinity Soil in Oasis Based on High-Resolution Remote Sensing Data

Beijing Jiaotong University (2018)

Monitoring and Prediction of Salinity Soil in Oasis Based on High-Resolution Remote Sensing Data

朱冠华;

Titre : Monitoring and Prediction of Salinity Soil in Oasis Based on High-Resolution Remote Sensing Data

Auteur : 朱冠华;

Grade : Master’s Theses 2018

Université : Beijing Jiaotong University

Résumé
Soil salinization is one of the major forms of soil degradation.Most oasis in China located in arid and semi-arid areas in Xinjiang,where soil salinization is particularly serious.The harm of salinization to the fragile ecological environment in the Oasis region should not be underestimated.The use of remote sensing to understand the distribution of salinized soil is essential for the current oasis economy based on the primary industry.At present,most of the researches use low and middle resolution remote sensing data to extract salinized soil,which has such problems as low positioning accuracy and poor extraction accuracy.In this paper,QuickBird and SPOT high resolution remote sensing images,salinity index and vegetation coverage,combined with random forest classifier was used to extract the salinity soil and other land use in 125 Regime of Bingtuan.Then based on the FLUS model and land use classification results,the salinized soil conditions in 2025 were predicted and analyzed.The main contents and conclusions of this paper are as follows:1)In this paper,QuickBird and SPOT remote sensing images are used to calculate characteristic of salinity index and analyze sampling points,and the Pearson correlation analysis are used to select the more favorable salinity index.In combination with the results of the test area,the salinization index SI1,which is most suitable for the study area,soil information of salt spots in the study area can be extracted effectively.(2)Based on the random forest classifier to extract land use information,the overall accuracy reached 0.9,realizing the accurate identification of the salinized soil in the study area.By comparing the salinized soil characteristics of the two periods in 2005 and 2015,it proves that the salinized soil in the study area shows the tendency of withdrawal,and the severe salinized soil was particularly obvious.The transformation between different degrees of salinized soil is more complex.The distribution of salinized soil in the study area has a certain regularity.(3)By using the FLUS model,a disaster warning for salinized soil is carried out in 2025.The Kappa coefficient reached 0.87,and the overall accuracy is significantly higher than that of Markov-CA model.Compared with the results of land use in 2005 and 2015 and forecast result of 2025,it is found that the overall intensity of land use development of study area gradually decreased,but the change rate of severe salinized soil was still higher than that of other land use types.According to the situation of different salinized soils in the group field,the occurrence of disasters in the selected areas,mainly includes two aspects.First,the water salt movement caused by natural river and hydraulic construction land may cause secondary salinization after the decrease of soil salinity.The second is the stubborn salted soil disaster caused by the soil texture.This area should be constructed as early as possible hydraulic construction,increase technical input,and gradually reduce the soil salt content

Mots clés : high-resolution remote sensing image; object-oriented; salinity index; random forest; oasis area; FLUS model;

Présentation (CNKI)

Page publiée le 26 avril 2019