Formation Mechanism of Peridot from Yiqisong, Jilin Province and the Application of Machine Learning to Its Origin Determination
-
摘要:
为进一步探明我国吉林敦化意气松南山橄榄石矿的形成环境和成因机制，区分该地区橄榄石与其他产地的橄榄石，本研究使用激光拉曼光谱、扫描电子显微镜和激光剥蚀电感耦合等离子体质谱仪等测试方法对该矿床的橄榄石及其玄武岩围岩展开了一系列岩相学和地球化学分析，同时对比了不同机器学习模型对橄榄石产地判别的准确率。结果表明，该地区的玄武岩主要为尖晶石相二辉橄榄岩。通过Ca、Al、Cr-橄榄石温度计估算橄榄石的形成温度约为903~1 055 ℃。化学成分分析表明该地区橄榄石主要有地幔橄榄石(高Ni组)和斑晶橄榄石(低Ni组)两种。其中，地幔橄榄石被玄武岩浆捕获并携带上升的过程中，大颗粒的地幔橄榄石发生破碎，玄武岩浆发生分离结晶作用，晶出斑晶橄榄石和辉石，只有在岩浆搬运过程中保留下来的大颗粒的地幔橄榄石才具有成为宝石的潜力。基于橄榄石的化学成分，使用线性判别和机器学习的方法可有效的区分不同产地的宝石级橄榄石，但当样本中存在非宝石级橄榄石时，品质较低(主要影响因素是颜色)的橄榄石可能会干扰模型的准确性，应当全面地采集同一产地的各种不同样品以提高模型的准确性。
Abstract:To gain further insight into the formation environment and genesis mechanism of the peridot deposit from Yiqisong nanshan, Dunhua city, Jilin Province, China and distinguish the peridot deposit from the other origins, in this paper, a series of petrographic and geochemical analyses of peridot and its basalt were conducted using laser Raman spectroscopy, scanning electron microscopy, and laser ablation inductively coupled plasma mass spectrometer.Additionally, the accuracy of various machine learning models of peridot from differernt origins determination was also evaluated. The results suggest that the basalts in the area are predominantly spinel lherzolite.The formation temperature of the peridot was estimated to be about 903-1 055 ℃ through Ca, Al, and Cr in olivine thermometers.The peridot from this area mainly includes mantle olivine (high Ni group) and porphyritic olivine (low Ni group). The large-grained mantle olivine was captured by basaltic magma, which was fragmented during the ascent of the basaltic magma. In this process, basaltic magma underwent crystal differentiation, diopside, enstatite and porphyritic olivine precipitated. Only the large-grained mantle olivine fragments that survived the magmatic transport have the potential to be the gemstone. Geochemical data imply that the parent magma likely originated from partial melting of the asthenospheric mantle and may be a product of early Archean mantle magmatism.Based on the chemical compositions of peridot, we can use the methods of linear discriminant and machine learning to distinguish the gem-quality peridot from different origins effectively. However, when non-gem-grade olivine is present in the sample, lower quality peridot (where the main influencing factor is the colour) may interfere with the accuracy of the models. Thus, comprehensive peridot samples being various qualities from the same locality are essential to improve the the accuracy of the models.
-
Keywords:
- lherzolite /
- geochemistry /
- machine learning /
- origin /
- Yiqisong, Jilin Province
-
表 1 温度估算结果表(温度由Al、Cr、Ca-橄榄石温度计估算)Table 1. Results of the temperature estimates (temperature is estimated by Al, Cr, Ca in olivine thermometer)
Thermometer Type Average Temperature /℃ TAl in ol 926 TCr in ol 946 TCa in ol 995 表 2 基于橄榄石化学成分，使用线性判别及机器学习模型对橄榄石产地判别的准确率统计Table 2. Accuracy statistics of identification of olivine origin using linear discrimination and machine learning model based on olive petrochemical compositionsModel Accuracy/% Sample quantity：296 Sample quantity：313 B-LDA 83.4 84.3 Extra tree 93.3 93.6 Random Forest 86.5 88.3 Xgboost 92.1 88.3 Logistic Regression 79.8 85.0 lightgbm 95.5 83.0 catboost 89.9 92.6 Naive Bayes 80.9 80.9 注：产地判别选取Mn，Zn，Na，Al，Sc，V，Cr，P，Ti及REE十种元素，不同模型的准确率基本都高于80%;第一次使用296颗宝石级橄榄石样品进行判别，第二次加入17颗非宝石级橄榄石进行判别，两次判别不同模型的准确率均有所变化，表明样品数量和橄榄石品质对模型准确率有不同程度的影响 ![]()
图 1 橄榄玄武岩样品照片及试验结果：(a)橄榄石晶体及其玄武岩围岩；(b)橄榄玄武岩的BSE图像中可见橄榄石、辉石、赤铁矿-钛铁矿、角闪石、磷灰石等矿物。橄榄石和辉石呈自形-半自形，赤铁矿-钛铁矿常呈规则的六边形，角闪石大多为短柱状，平行或近垂直排列，磷灰石成近六边形穿排列在橄榄石和辉石附近；(c)橄榄玄武岩中主要矿物的拉曼光谱；(d)玄武岩微量元素(原始地幔标准化)配分模式蛛网，玄武岩中稀土元素呈右倾形分布，轻重稀土分异明显Figure 1. Photographs of the olivine basalt samples and test results: (a) peridot crystals and basalt samples; (b) BSE images of olivine basalt showing the presence of olivine, diopside, enstatite, hematite-ilmenite, hornblende, apatite, etc. Olivine and diopside exhibit idiomorphic to hypautomorphic. Hematite-ilmenite typically forms regular hexagons. Hornblende is predominantly observed as short columnar shapes, arranged parallel or nearly perpendicular. Apatite forms near-hexagonal prisms arranged in clusters adjacent to olivine and pyroxene; (c) Raman spectra of the main minerals in the olive basalts; (d) spider diagrams of trace elements (normalized by primitive mantle) in basaltic matrix. The trace elements show a right-dipping distribution, with obvious differences between light and heavy rare earths.Di: Diopside; En: Enstatite; Hem-Ilm: Hematite-Ilmenite; Pl: Plagioclase; Ol: Olivine; Ap: Apatite; Hb: Hornblende -
![]()
图 1 橄榄玄武岩样品照片及试验结果：(a)橄榄石晶体及其玄武岩围岩；(b)橄榄玄武岩的BSE图像中可见橄榄石、辉石、赤铁矿-钛铁矿、角闪石、磷灰石等矿物。橄榄石和辉石呈自形-半自形，赤铁矿-钛铁矿常呈规则的六边形，角闪石大多为短柱状，平行或近垂直排列，磷灰石成近六边形穿排列在橄榄石和辉石附近；(c)橄榄玄武岩中主要矿物的拉曼光谱；(d)玄武岩微量元素(原始地幔标准化)配分模式蛛网，玄武岩中稀土元素呈右倾形分布，轻重稀土分异明显
Figure 1. Photographs of the olivine basalt samples and test results: (a) peridot crystals and basalt samples; (b) BSE images of olivine basalt showing the presence of olivine, diopside, enstatite, hematite-ilmenite, hornblende, apatite, etc. Olivine and diopside exhibit idiomorphic to hypautomorphic. Hematite-ilmenite typically forms regular hexagons. Hornblende is predominantly observed as short columnar shapes, arranged parallel or nearly perpendicular. Apatite forms near-hexagonal prisms arranged in clusters adjacent to olivine and pyroxene; (c) Raman spectra of the main minerals in the olive basalts; (d) spider diagrams of trace elements (normalized by primitive mantle) in basaltic matrix. The trace elements show a right-dipping distribution, with obvious differences between light and heavy rare earths.Di: Diopside; En: Enstatite; Hem-Ilm: Hematite-Ilmenite; Pl: Plagioclase; Ol: Olivine; Ap: Apatite; Hb: Hornblende
表 1 温度估算结果表(温度由Al、Cr、Ca-橄榄石温度计估算)
Table 1 Results of the temperature estimates (temperature is estimated by Al, Cr, Ca in olivine thermometer)
Thermometer Type Average Temperature /℃ TAl in ol 926 TCr in ol 946 TCa in ol 995 
下载: 导出CSV
表 2 基于橄榄石化学成分，使用线性判别及机器学习模型对橄榄石产地判别的准确率统计
Table 2 Accuracy statistics of identification of olivine origin using linear discrimination and machine learning model based on olive petrochemical compositions
Model Accuracy/% Sample quantity：296 Sample quantity：313 B-LDA 83.4 84.3 Extra tree 93.3 93.6 Random Forest 86.5 88.3 Xgboost 92.1 88.3 Logistic Regression 79.8 85.0 lightgbm 95.5 83.0 catboost 89.9 92.6 Naive Bayes 80.9 80.9 注：产地判别选取Mn，Zn，Na，Al，Sc，V，Cr，P，Ti及REE十种元素，不同模型的准确率基本都高于80%;第一次使用296颗宝石级橄榄石样品进行判别，第二次加入17颗非宝石级橄榄石进行判别，两次判别不同模型的准确率均有所变化，表明样品数量和橄榄石品质对模型准确率有不同程度的影响
下载: 导出CSV
-
[1] De Hoog J C M, Gall L, Cornell D H. Trace-element geochemistry of mantle olivine and application to mantle petrogenesis and geothermobarometry[J]. Chemical Geology, 2010 (270): 196-215.
[2] Adamo I, Bocchio R, Pavese A, et al. Characterization of peridot from Sardinia, Italy[J]. Gems & Gemology, 2009, 45(2): 130-133.
[3] Shen A, Blodgett T A, Shigley J E. Country-of-origin determination of modern gem peridots from LA-ICP-MS trace-element chemistry and linear discriminant analysis (LDA)[C]// Geological Society of America, 2013.
[4] Shen A H, Koivula J I, Shigley J E. Identification of extraterrestrial peridot by trace elements[J]. Gems & Gemology, 2011, 47 (3): 208-213.
[5] Zhang Z Q, Ye M, Shen A H. Characterisation of peridot from China's Jilin Province and from North Korea[J]. The Journal of Gemmology, 2019, 36 (5): 436-446. doi: 10.15506/JoG.2019.36.5.436
[6] 汪连成. 吉林省大石河橄榄石矿床的产出特征[J]. 长春地质学院学报, 1996(1): 43-46. Wang L C. The occurrence of Dashihe olivine deposit in Jilin Province, China[J]. Journal of Changchun University of Earth Sciences, 1996, 26 (1): 43-46. (in Chinese)
[7] 仲源, 沈锡田, 张志清, 等. 机器学习在橄榄石产地溯源中的应用[J]. 宝石和宝石学杂志(中英文), 2023, 25(6): 65-75. doi: 10.15964/j.cnki.027jgg.2023.06.006 Zhong Y, Shen A H, Zhang Z Q, et al. 2023. Application of machine learning algorithms in the geographical origin determination of peridot[J]. Journal of Gems & Gemmology, 2023, 25(6): 65-75. (in Chinese) doi: 10.15964/j.cnki.027jgg.2023.06.006
计量
- 文章访问数: 38
- HTML全文浏览量: 8
- PDF下载量: 21
