Abstract Among various trace evidence, the soil is very crucial evidence because it can link the suspect with the crime and crime with its geographical region. Based on this hypothesis, the aim of present research is to characterize, differentiate and classify the soil samples collected from various geological regions of northwestern India via using attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy combined with chemometrics. The organic and inorganic parts of surface and depth soil samples are successfully characterized. Sieving and heating of soil in a muffle furnace (650 °C) affect the spectral features considerably. The differentiation of soil samples is done by utilizing two approaches i.e. qualitative analysis and multivariate analysis. Both approaches provide a high discriminating power (i.e. qualitative analysis, surface = 99.35%, depth = 97.38%, and multivariate analysis, surface = 100%, depth = 100%). A classification model based on canonical discriminant function analysis is also built for grouping of soil to its particular geographical region. The developed model provides 100% correct classification of soil samples after the leave-one-out cross-validation. Therefore, the current study provides useful methods of soil analysis which can further be utilized by soil as well as forensic expert dealing with such cases.

中文翻译：

---

土壤取证：痕量证据的光谱检查

摘要 在各种痕迹证据中，土壤是非常重要的证据，因为它可以将犯罪嫌疑人与犯罪联系起来，并将犯罪与其地理区域联系起来。基于这一假设，本研究的目的是通过使用衰减全反射傅里叶变换红外 (ATR-FTIR) 光谱结合化学计量学对从印度西北部各个地质区域收集的土壤样品进行表征、区分和分类。成功表征了表层和深层土壤样品的有机和无机部分。在马弗炉 (650 °C) 中对土壤进行筛分和加热会显着影响光谱特征。土壤样品的区分是通过使用两种方法完成的，即定性分析和多变量分析。这两种方法都提供了很高的辨别力（即 定性分析，表面 = 99.35%，深度 = 97.38%，多元分析，表面 = 100%，深度 = 100%）。还建立了基于典型判别函数分析的分类模型，用于将土壤分组到其特定的地理区域。开发的模型在留一法交叉验证后提供了 100% 正确的土壤样品分类。因此，当前的研究提供了有用的土壤分析方法，土壤以及处理此类案件的法医专家可以进一步利用这些方法。开发的模型在留一法交叉验证后提供了 100% 正确的土壤样品分类。因此，当前的研究提供了有用的土壤分析方法，土壤以及处理此类案件的法医专家可以进一步利用这些方法。开发的模型在留一法交叉验证后提供了 100% 正确的土壤样品分类。因此，当前的研究提供了有用的土壤分析方法，土壤以及处理此类案件的法医专家可以进一步利用这些方法。