新疆于田绿洲土壤盐渍化风险遥感定量评估研究

In arid area, assessment of soil salinization risk has explored new fields for the study of regional ecological risk. Nevertheless, the quantitative assessment of soil salinization risk based on remote sensing is still relatively rare in the study on soil salinization in Xinjiang. In this study, Keriya Oasis, the typical arid land oasis of Xinjiang Uyghur Autonomous Region, will be used as a core study area. Remote sensing based quantitative monitoring of soil salinization and establishment of the risk assessment model will be conducted as two aspects relying each other closely. Firstly, on the basis of full investigation, making good use of multi-source remote sensing data, and using image fusion technique and support vector machine classifier, quantitative monitoring method of soil salinization, which is based on remote sensing information and quantitative parameters, will be build and further improved. Secondly, the theory of ecological risk will be introduced into soil salinity risk assessment of the Keriya Oasis ecosystem. Then utilizing data assimilation techniques, making GIS spatial analysis for the multi-source remote sensing and observational data, and then the index system for assessment of soil salinization risk will be established and quantitative assessment model of soil salinization risk based on remote sensing will be constructed. Finally, the regional soil salinity risk will be quantitatively assessed and graded, and the research results will be verified. This assessment and mapping of the soil salinization risk in Keriya Oasis should pave the way for finding out new ideas and approaches of the quantitative study and evaluation of soil salinization, moreover it could provide scientific basis for rehabilitation of saline soil, preventing from soil salinization, and sustainable development of oasis agriculture and eco-environmental stability.

干旱区土壤盐渍化风险评估已为区域生态风险研究开拓了新的领域。然而，盐渍化风险遥感定量评估在新疆土壤盐渍化研究中的应用尚有不足。本项目选择具有长期研究基础的新疆于田绿洲为典型研究区，将干旱区土壤盐渍化遥感定量监测及其风险评估模型的构建为相互依托的两个层次。首先，在充分调研的基础上，利用多源遥感数据，采用融合及支持向量机分类方法，构建并进一步完善基于遥感信息及定量参数的土壤盐渍化定量监测模型。其次，将生态风险理论引进于田绿洲生态系统的土壤盐渍化风险评估中，运用数据同化技术，对多源遥感及观测数据进行GIS空间分析，建立盐渍化风险评价指标体系，构建土壤盐渍化风险遥感定量评估模型。最后，对研究区土壤盐渍化风险状况进行定量评估与分级，并对研究结果进行验证。研究成果能为土壤盐渍化的定量研究和评价开辟新的研究思路和途径，可为于田绿洲盐渍化土壤的科学改良、绿洲生态环境及农业的可持续发展提供科学依据。

土壤盐渍化是全人类长期关注的世界性土地退化及自然灾害问题之一。目前我国西部干旱、半干旱地区，特别是新疆地区的土壤盐渍化问题仍然极为突出，是影响当地生态环境、农业生产和土地资源可持续利用的关键问题。因而，利用遥感与地理信息系统技术的宏观、实时、动态、快速、可视化及信息量大等优势，对干旱区土壤盐渍化风险进行动态、有效、实时监测及定量评估尤为重要，意义重大。. 本项目选择具有长期研究基础的于田绿洲为研究区，针对干旱区土壤盐渍化信息挖掘及风险定量评估问题，以3S集成技术为依托，结合传统的野外实测手段，采用多种极化目标分解方法，构建基于最佳特征子集和最优分类参数的WFS-SVM分类模型，实现了对干旱区土壤盐渍化全极化微波遥感监测。集成光学与多源微波遥感数据，进行S-PCA融合，构建最优SVM分类模型及决策树分类方案，并利用这些模型进行研究区土壤盐渍化多源遥感监测，进而充分挖掘了光学遥感数据的丰富光谱信息、微波遥感数据的后向散射系数特征及其空间纹理信息。通过采用多光谱、高光谱及高分辨率等多源遥感数据，定量反演了研究区土壤盐分，探讨了地下水埋深、土壤水分、植被覆盖度等主要参数的时空异质性，进一步分析了研究区土壤水盐时空变化特征。利用遥感定量反演和GIS空间地学分析的技术，获取并创建盐渍化风险评价因子的RS和GIS栅格数据集，完成了研究区属性、矢量和栅格数据的有机集成和数据同化，创建空间数据库，建立了研究区土壤盐渍化风险评估指标体系，计算评估因子风险权重，引入灰色分析系统，构建盐渍化风险灰色定量评估模型，进而开展了研究区盐渍化风险状况的定量评估与分析。并在此基础上，设计开发了干旱区土壤盐渍化风险WebGIS评估系统，进一步实现了土壤盐渍化风险评估的本地与远程动态实时化。本课题成果为干旱区土壤盐渍化有效治理、土壤资源的科学管理提供崭新的技术手段和可靠的参考依据，以及为该区域农业和生态环境的可持续发展提供科学依据。