土壤重金属高光谱遥感定量监测研究进展与趋势

Research progress and trend of quantitative monitoring of hyperspectral remote sensing for heavy metals in soil

与传统的土壤重金属含量定量研究方法相比,高光谱遥感技术具有成本低、效率高、探测范围广、宏观性强、可动态监测等优势。基于空-天-地研究视角,分析了土壤重金属高光谱数据特征、预处理方法与技术流程、应用条件与范围,论述了基于高光谱遥感技术开展土壤重金属监测的发展历程。通过归纳基于土壤重金属、土壤活性成分的光谱特征反演,发现土壤有机质、铁氧化物、黏土矿物等含量是导致土壤特征波段差异的关键因素。总结了土壤中常见活性成分及重金属的特征波谱,认为350~2500nm是预测土壤成分含量的主要特征波段范围。影响反演精度的关键因素包括土壤高光谱响应机制、高光谱数据质量、重金属赋存状态以及反演建模方法,改进光谱优化方法、构建高效的反演模型、明确重金属光谱特征及其赋存机理是进一步提升土壤重金属高光谱反演精度的有效途径和方式。为适应土壤重金属含量定量监测研究需要,未来高光谱遥感技术将会朝着定量化、宏观化、主动化、现场化方向发展。另外,数据多源化、非线性方法与线性方法融合、顾及重金属和活性成分的多特征波段,也是今后高光谱遥感技术发展的重要趋势。

Compared with the traditional methods for quantitative study of soil heavy metal content,the hyperspectral remote sensing technology has greater advantages,such as lower cost,higher efficiency,wider detection range,strong macroscopic property,and dynamic monitoring,etc.Based on air-sky-earth research perspective,this paper analyzed the characteristics,pre-processing methods and technical processes,application conditions and scope of soil heavy metal hyperspectral data;it discussed the development of soil heavy metal monitoring based on hyperspectral remote sensing technology.By analyzing the inversion of spectral features based on soil heavy metals and soil active ingredients,we found that the contents of organic matter,iron oxides,and clay minerals in soil are the key factors leading to the differences in soil characteristic wavebands.In this paper,by summarizing the characteristic wave spectra of common active components and heavy metals in soil,it shows that 350-2500 nm is the main waveband range for predicting their contents,and the characteristic wavebands are very susceptible to soil types.The key factors affecting the inversion accuracy involve the soil hyperspectral response mechanism,hyperspectral data quality,occurrence state of heavy metals,and inversion modeling methods.The effective ways to further improve the accuracy of hyperspectral inversion of soil heavy metals include improving the spectral optimization method,constructing an efficient inversion model,and clarifying the spectral characteristics and occurrence mechanism of heavy metals.The future development of hyperspectral remote sensing technology will be characterized by quantitative,active,macroscopic and on-site.In addition,the data multi-source,fusion of non-linear and linear methods,and considering of multi-featured waveband information are also important trends for hyperspectral remote sensing technology in the future.