土壤盐渍化水－盐遥感监测最优尺度研究

Due to the difference of remote sensing data in different platforms and different resolutions, how to select optimal scale and implementation of scale transformation has become the main problem of current remote sensing monitoring of soil salinization. In this study, the soil salinization was taken as the research object in the Wei-Ku oasis， (1) Soil spectral signatures of different soil moisture, salinity and water - salt interaction were analyzed. The relationship between Salty soil spectrum and soil water and salt condition was revealed and corresponding relations between different water- salt contents and spectral bands were confirmed (2) The AIEM model and the Water-Cloud model were used to develop the soil moisture retrieval model of bare land and vegetation coverage area; Integrating remote sensing and sensing technology, the soil salinity inversion model based on multi source data was established, and the multi-scale water salt synergistic inversion model was developed by using the feature space theory (3) Variety of precision evaluation indices were applied to analyze the problem of scale effect of remote sensing monitoring of soil salinization and the scale transformation methods, such as data assimilation and data fusion were adapted for realizing model conversion on the “Ground - Air - Satellite" scale; finally, the optimal scale of "water salt monitoring" in the study area was determined. Improving the accuracy and efficiency of remote sensing monitoring of soil salinization is to provide scientific basis and technical support for the evaluation and improvement of Oasis saline soil.

由于不同平台、不同分辨率的遥感数据尺度存在差异，如何选取最优尺度及实现尺度转换成为当前土壤盐渍化遥感监测面临的主要问题。本研究以渭－库绿洲为研究区，以土壤盐渍化为研究对象，（1）分析不同水分、盐分及水-盐交互作用下的土壤光谱特征；揭示盐渍土土壤光谱与土壤水-盐状况的内在联系，确定不同水盐含量与光谱波段对应的关系（2）采用AIEM模型和Water-Cloud模型分别构建裸露地表及植被覆盖区域的土壤水分反演模型；集成遥感和近感技术，建立基于多源数据的土壤盐分反演模型；利用特征空间理论，构建多尺度水－盐协同反演模型。（3）运用多种精度评价指标，分析土壤盐渍化遥感监测中的尺度效应问题，采用数据同化和数据融合等尺度转换方法，实现“地面-航空-卫星”尺度模型转换，确定研究区“水－盐”监测的最优尺度。提高土壤盐渍化遥感监测的精度与效率，为绿洲盐渍化土壤的评价和改良提供科学依据和技术支持。

本项目针对干旱、半干旱地区土壤盐渍化监测及其最优尺度选取问题，以新疆维吾尔自治区渭干河－库车河绿洲作为研究区，土壤水分、盐分及水－盐监测作为主要研究目标，利用地理学、数学等相关理论知识结合野外采样与室内实验，采用分数阶微分预处理法、优化光谱指数法、相关分析法、波段组合算法、TGBM（梯度提升机）模型、PLSR（偏最小二乘回归）模型、VIP（变量重要性准则）模型、GWR（地理加权回归）模型、BP神经网络模型等方法，将不同尺度遥感影像数据重采样到相同的光谱尺度，探讨珠海1号高光谱影像定量反演盐渍化土壤含水量的可能性以及分数阶微分优化光谱指数的应用效果；针对高光谱数据，利用优化光谱指数提高光谱异质性土壤的含盐量估算精度；利用不同变换运算下的最佳高光谱参数建立土壤电导率估算模型从而实现土壤盐分信息的高效监测；基于土壤采样尺度的有机质含量、重金属砷含量与室内高光谱尺度数据探讨优化光谱指数方法应用于高光谱检测露天煤矿区土壤重金属砷含量的可行性；通过对高光谱数据进行数学变换的微分预处理大幅提升了高光谱建模精度和稳健性。本项目研究成果为提高高光谱遥感反演精度提供新的研究思路；为绿洲盐渍化土壤的评价和改良提供科学依据和技术支持；为在区域尺度上快速准确寻求卫星传感器中监测春小麦冠层叶绿素含量的最佳波段提供依据；通过探讨不同尺度数据对盐渍化的响应，分析了土壤盐渍化遥感监测中的尺度效应问题，对区域盐渍化防治与调控，农业可持续发展以及生态文明建设具有重要意义。通过四年的研究，完成了项目计划基本要求。本项目执行期间，已发表学术论文6篇，其中SCI收录论文3篇，其他中文核心期刊论文3篇，专业专著一本。