土壤漏油含量光谱预测实验研究与分析

发布时间：2018-07-28 07:02

【摘要】：针对土壤漏油污染的检测、分析和含量预测等实际问题,用关中平原常见普通土壤、柴油、机油及油土混合物为研究对象,设计被不同油种污染以及不同污染程度的土壤光谱特征实验。采用光谱分析的方法来获取光谱指数,测量并分析渗漏相同含量两个油种的油土混合物以及渗漏不同含量同一油种的油土混合物的光谱特征及其差异,根据分析结果建立一个基于反射特征的土壤柴油渗漏含量的光谱预测模型。研究表明:(1)渗漏相同含量不同油种时,渗漏柴油的土壤的反射率低于渗漏机油的土壤;渗漏柴油和机油的土壤均在1 740和2 328nm附近出现双吸收谷特征,且在每一个吸收谷点处渗漏柴油的土壤的光谱吸收深度和光谱吸收指数均低于渗漏相同含量机油的土壤。(2)建立的土壤柴油渗漏含量的光谱预测模型相关系数r的值达到0.854,说明模型具有较强的稳定性;评价模型预测能力的均方根误差RMSE值为0.016证明该模型具有较好预测能力,可以对土壤柴油渗漏含量进行有效的预测。
[Abstract]:Aiming at the practical problems of oil spill pollution detection, analysis and content prediction, the common soil, diesel oil, oil and oil-soil mixture in Guanzhong Plain were used as the research objects. The experiments of soil spectral characteristics with different oil pollution and different pollution degree were designed. The spectral index was obtained by spectral analysis, and the spectral characteristics and differences of the oil-soil mixture with the same leakage content and the same oil content with the same oil content were measured and analyzed. Based on the analysis results, a spectral prediction model of soil diesel leakage content based on reflection characteristics was established. The results showed that: (1) the reflectivity of the soil with the same leakage content of different oil species was lower than that of the oil leakage soil, and the double absorption valley was found in both the oil and diesel oil leakage soils near 1 740 and 2 328nm. The spectral absorption depth and spectral absorption index of diesel-leaking soil at each point of absorption valley are lower than those of the soil with the same oil content. (2) the correlation coefficient r of the spectral prediction model for the leakage content of soil diesel oil is established. The value reaches 0.854, which indicates that the model has strong stability. The root-mean-square error (RMSE) of the evaluation model was 0.016, which proved that the model had good prediction ability and could be used to predict soil diesel leakage effectively.
【作者单位】：中国科学院西安光学精密机械研究所;西安交通大学电子与信息工程学院;中国科学院大学;
【基金】：国家自然科学基金项目(41005019) 国家(863计划)项目资助
【分类号】：X833

【相似文献】