Sedimentary geochemistry is very often associated with the measurement of isotopic composition of carbon and oxygen from carbonates. The usual technique combining acid digestion and mass spectrometry analysis is slow, costly and non-ideal for spatially resolved analyses. When carbonates are processed using laser calcination and the gas produced during calcination is analyzed by infrared spectrometry, the time required for isotopic analysis is reduced to around 15 min to analyze 30 mg of carbonate in situ. Although the time saved is significant, it is hardly reasonable to carry out a high-resolution isotopic mapping of large samples. A fully resolved isotopic mapping, for example, of a sample with 25 cm2 surface area at resolution of a tenth of a millimeter, would require a continuous measurement carried out for a month. The aim of this study is, therefore, to explore possible strategies for constructing an isotopic map with a minimum number of analyses. Two approaches are pro-posed: (i) a mathematical approach that seeks to establish a correlation between the position of the sample and the carbon or oxy-gen isotopes, and (ii) an approach that looks for a correlation between the color (spectral characteristics) of the sample surface subdomains and their isotopic compositions. The choice of the second approach stems from the assumption that color contains a priori information about geological or geochemical processes. Several algorithms were developed and tested, notably using artificial intelligence tools. To testify the isotopic maps produced by these algorithms, posteriori isotopic measurements are taken and compared with the predictions from computed isotopic maps.

中文翻译：

---

碳酸盐中激光-激光同位素测量的标测应用


沉积地球化学通常与碳酸盐中碳和氧的同位素组成的测量有关。结合酸消解和质谱分析的常用技术速度慢、成本高昂且不适合空间分辨分析。当使用激光煅烧处理碳酸盐并通过红外光谱分析煅烧过程中产生的气体时，同位素分析所需的时间减少到原位分析 30 mg 碳酸盐所需的时间约为 15 分钟。尽管节省的时间很长，但对大样品进行高分辨率同位素映射几乎不合理。例如，以十分之一毫米的分辨率对表面积为 25 cm2 的样品进行完全解析的同位素映射，需要连续测量一个月。因此，本研究的目的是探索以最少的分析次数构建同位素图的可能策略。提出了两种方法：（i） 一种数学方法，旨在建立样品位置与碳或氧同位素之间的相关性，以及 （ii） 一种寻找样品表面子域的颜色（光谱特性）与其同位素组成之间的相关性的方法。选择第二种方法源于这样一个假设，即颜色包含有关地质或地球化学过程的先验信息。开发和测试了几种算法，特别是使用人工智能工具。为了验证这些算法生成的同位素图，进行了后验同位素测量并与计算同位素图的预测进行了比较。