| Citation: |
Yong-sheng Li, Chong Peng, Xiang-jin Ran, Lin-Fu Xue, She-li Chai, 2022. Soil geochemical prospecting prediction method based on deep convolutional neural networks-Taking Daqiao Gold Deposit in Gansu Province, China as an example, China Geology, 5, 71-83. doi: 10.31035/cg2021044
|
Soil geochemical prospecting prediction method based on deep convolutional neural networks-Taking Daqiao Gold Deposit in Gansu Province, China as an example
-
Abstract
A method is proposed for the prospecting prediction of subsurface mineral deposits based on soil geochemistry data and a deep convolutional neural network model. This method uses three techniques (window offset, scaling, and rotation) to enhance the number of training data for the model. A window area is used to extract the spatial distribution characteristics of soil geochemistry and measure their correspondence with the occurrence of known subsurface deposits. Prospecting prediction is achieved by matching the characteristics of the window area of an unknown area with the relationships established in the known area. This method can efficiently predict mineral prospective areas where there are few ore deposits used for generating the training dataset, meaning that the deep-learning method can be effectively used for deposit prospecting prediction. Using soil active geochemical measurement data, this method was applied in the Daqiao area, Gansu Province, for which seven favorable gold prospecting target areas were predicted. The Daqiao orogenic gold deposit of latest Jurassic and Early Jurassic age in the southern domain has more than 105 t of gold resources at an average grade of 3‒4 g/t. In 2020, the project team drilled and verified the K prediction area, and found 66 m gold mineralized bodies. The new method should be applicable to prospecting prediction using conventional geochemical data in other areas.
-
-
References
Bi ZW, Wang YK, Sun X, Zhang JZ, Zhang T, Liu TF. 2020. Geological characteristics, magmatic geochemistry and petrogenic chronology of the Bagong Pb polymetallic ore spot in Ritu County, Tibet. Geology in China, 47(2), 497–515 (in Chinese with English abstract). doi: 10.12029/gc20200216. Cameron EM, Hamilton SM, Leybourne MI, Leybourne ML, Hall GEM. 2004. Finding deeply buried deposits using geochemistry. Geochemistry: Exploration, Environment, Analysis, 4(1), 7–32. doi: 10.1144/1467-7873/03-019. Caritat PD, Main PT, Grunsky EC, Mann A. 2017. Recognition of geochemical footprints of mineral systems in the regolith at regional to continental scales. Australian Journal of Earth Sciences, 64, 1033–1043. doi: 10.1080/08120099.2017.1259184. Carranza EJM. 2010. Geochemical Anomaly and Mineral Prospectivity Mapping in GIS. In: Handbook of Exploration and Environmental Geochemistry, vol. 11. Amsterdam, Elsevier, 145‒151. Chawla NV, Bowyer KW, Hall LO, Kegelmeyer WP. 2002. SMOTE: Synthetic minority over-sampling technique. Journal of Artificial Intelligence Research, 16(1), 321–357. doi: 10.1613/jair.953. Chelsea F, Pieter A, Sergey L. 2017. Model-agnostic meta-learning for fast adaptation of deep networks. arXiv: 1703.03400v3. Chen LR, Guan QF, Xiong YH, Liang JY, Wang Y, Xu YQ. 2019. A spatially constrained multi- autoencoder approach for multivariate geochemical anomaly recognition. Computers and Geosciences, 125, 43–54. doi: 10.1016/j.cageo.2019.01.016. Chen YL, Lu LJ, Li XB. 2014. Application of continuous restricted Boltzmann machine to identify multivariate geochemical anomaly. Journal of Geochemical Exploration, 140, 56–63. doi: 10.1016/j.gexplo.2014.02.013. Chen YL. 2015. Mineral potential mapping with a restricted Boltzmann machine. Ore Geology Reviews, 71, 749–760. doi: 10.1016/j.oregeorev.2014.08.012. Chen YL, Wu W. 2017. Application of one-class support vector machine to quickly identify multivariate anomalies from geochemical exploration data. Geochemistry: Exploration, Environment, Analysis, 17(3), 231–238. doi: 10.1144/geochem2016-024. Cohen DR, Kelley DL, Anand R, Coker WB. 2010. Major advances in exploration geochemistry, 1998–2007. Geochemistry: Exploration, Environment Analysis, 10, 3–16. doi: 10.1144/1467-7873/09-215. Ding Y, Du QZ, Yasin Q, Zhang Q, Liu LH. 2020. Fracture prediction based on deep learning: Application to a buried hill carbonate reservoir in the S area. Geophysical Prospecting for Petroleum, 59(2), 267–275 (in Chinese with English abstract). doi: 10.3969/j.issn.1000-1441.2020.02.013. Fang GC, Chen YC, Chen ZH, Zeng ZL, Liu CH, Tong QQ, Sun J, Zhu GH. 2016. Petrology and geochemistry of granite in the Pangushan tungsten deposit, south Jiangxi Province. Geology in China, 43(5), 1558–1568 (in Chinese with English abstract). doi: 10.12029/gc20160506. Fu C, Lin NT, Zhang D, Wen B, Wei QQ, Zhang K. 2018. Prediction of reservoirs using multi-component seismic data and the deep learning method. Chinese Jouranl of Geophysics, 61(1), 293–303 (in Chinese with English abstract). doi: 10.6038/cjg2018L0193. Goswami S, Upadhyay PK, Saravanan B, Natarajan V, Verma MB. 2019. Two types of uranium mineralization in Gulcheru quartzite: Fracture-controlled in Ambakapalle area and litho-controlled in Tummalapalle area, Cuddapah Basin, Andhra Pradesh, India. China Geology, 2(2), 142–156. doi: 10.31035/cg2018099. Inoue H. 2018. Data augmentation by pairing samples for images classification. arXiv: 1801.02929v1. Ke CH, Wang XX, Nie ZR, Yang Y, Lü XQ, Wang SA, Li JB. 2020. Age, geochemistry, Nd-Hf isotopes and relationship between granite and gold mineralization of Zhongchuan granitoid pluton in West Qinling. Geology in China, 47(4), 1127–1154 (in Chinese with English abstract). doi: 10.12029/gc20200415. Liu YG, Lü XB, Zhang ZJ, You GJ, Cao XF, Wang YQ, Liu G. 2011. Genesis of Daqiao gold deposit in Xihe County, Gansu Province. Mineral Deposit, 30(6), 1085–1099 (in Chinese with English abstract). doi: 10.16111/j.0258-7106.2011.06.009. Mann AW, Birrell RD, Mann AT, Humphreys DB, Perdrix JL. 1998. Application of the mobile metal ion technique to routine geochemical exploration. Geochemical Exploration, 61(1–3), 87–102. doi: 10.1016/S0375-6742(97)00037-X. McLaughlin N, Rincon JMD, Miller P. 2015. Data-augmentation for reducing dataset bias in person re-identification. In: 12th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS), Karlsruhe, Germany, 1‒6. doi: 10.1109/AVSS.2015.7301739. Peng C, Pan BZ, Xue LF, Liu HY. 2019. Geophysical survey of geothermal energy potential in the Liaoji Belt, northeastern China. Geotherm Energy, 7, 14. doi: 10.1186/s40517-019-0130-y. Peng C, Xue LF, Zhu M, Chai Y, Liu WY. 2016. The location and evolution of the tectonic boundary between the Paleoproterozoic Jiao-Liao-Ji Belt and the Longgang Block, northeast China. Precambrian Research, 272, 18–38. doi: 10.1016/j.precamres.2015.10.016. Peng C, Pan BZ, Xue LF, Ma YN, Liu WY, Li TL, Liu HY, Zhu K, Xu CH. 2020a. Location of the tectonic boundary between the Paleoproterozoic Jiao-Liao-Ji Belt and the Archean Longgang Block in the Lianshanguan area, northeastern China. Precambrian Research, 344, 105762. doi: 10.1016/j.precamres.2020.105762. Peng C, Pan BZ, Xue LF, Dai CQ, Liu HY, Zhu K, Xu CH. 2020b. Three-dimensional structural modeling and deformation analysis of Archean magnetite quartzite from the Anshan–Benxi area, northeastern China. Geomechanics and Geophysics for Geo-Energy and Geo-Resources, 6(4), 1–23. doi: 10.1007/s40948-020-00185-7. Ran XJ, Xue LF, Zhang YY, Liu ZY, Sang XJ, He JX. 2019. Rock classification from field image patches, analyzed using a deep convolutional neural network. Mathematics, 7(8), 755. doi: 10.3390/math7080755. Rodriguez-Galiano V, Sanchez-Castillo M, Chica-Olmo M, Chica-Rivas M. 2015. Machine learning predictive models for mineral prospectivity: An evaluation of neural networks, random forest, regression trees and support vector machines. Ore Geology Reviews, 71, 804–818. doi: 10.1016/j.oregeorev.2015.01.001. Snell J, Swersky K, Zemel RS. 2017. Prototypical networks for few-shot learning. arXiv: 1703.05175v2. Smee BW. 1998. A new theory to explain the formation of soil geochemical responses overly deeply covered gold mineralization in arid environments. Journal of Geochemical Exploration, 61(1–3), 149–172. doi: 10.1016/S0375-6742(98)00007-7. Wang XQ. 1998. Leaching of mobile forms of metals in overburden: Development and applications. Geochemical Exploration, 61(1–3), 39–55. doi: 10.1016/S0375-6742(97)00039-3. Wei SG, Song Y, Tang JX, Gao K, Feng J, Li YB, Hou L. 2016. Geochronology, geochemistry and petrogenesis of quartz diorite porphyrite from the Sena copper (gold) deposit, Tibet. Geology in China, 43(6), 1894–1912 (in Chinese with English abstract). doi: 10.12029/gc20160604. Wu TT, Zhao DF, Shao J, Bao QZ, Wang HB. 2014. Geological and geochemical characteristics and genesis of the Biliyagu lead-zinc-silver deposit, Inner Mongolia. Geology in China, 41(4), 1242–1252 (in Chinese with English abstract). doi: 10.3969/j.issn.1000‒3657.2014.04.017. Xiong YH, Zuo RG. 2020. Recognizing multivariate geochemical anomalies for mineral exploration by combining deep learning and one-class support vector machine. Computers and Geosciences, 140, 104484. doi: 10.1016/j.cageo.2020.104484. Xu ST, Zhou YZ. 2018. Artificial intelligence identification of ore minerals under microscope based on deep learning algorithm. Acta Petrologica Sinica, 34(11), 3244–3252 (in Chinese with English abstract). Yang PQ, Liu JD, Zhang YF, Liang S, Zhao Y, Liu SM. 2017. Ore geochemical characteristics and metallogenic epoch of typical graphite deposits in Jiamusi Massif, Heilongjiang Province. Geology in China, 44(2), 301–315 (in Chinese with English abstract). doi: 10.12029/gc20170207. Zuo RG, Xiong YH, Wang J, Carranza EJM. 2019. Deep learning and its application in geochemical mapping. Earth-Science reviews, 192, 1–14. doi: 10.1016/j.earscirev.2019.02.023. -
Access History
Figures(9)
Tables(2)
Export File
Citation
Yong-sheng Li, Chong Peng, Xiang-jin Ran, Lin-Fu Xue, She-li Chai, 2022. Soil geochemical prospecting prediction method based on deep convolutional neural networks-Taking Daqiao Gold Deposit in Gansu Province, China as an example, China Geology, 5, 71-83. doi: 10.31035/cg2021044
Format
Content
DownLoad:
-
Figure 1.
a‒Tectonic subdivision of China (modified from Peng C et al. 2016, 2019, 2020a); b‒major geological facies of the region containing the study area. c‒geological map of the study area, showing the location of the Daqiao gold deposit in Gansu Province. 1‒Holocene alluvial sand and gravel; 2‒Pleistocene loess; 3‒section 2 of the Gansu Group; 4‒section 1 of the Gansu Group; 5‒first lithological section of the Middle Triassic Daheba Formation; 6‒first lithological section of the Middle Triassic Taoshiguan Formation; 7‒second lithological section of the Middle Triassic Taoshiguan Formation; 8‒second lithological section of the Late Carboniferous Minhe Formation; 9‒third lithological section of the Middle Devonian Huangjiagou Formation; 10‒second lithological section of the Middle Devonian Huangjiagou Formation; 11‒first lithological section of the Middle Devonian Huangjiagou Formation; 12‒section 3 of the latest Silurian Zhuowukuo Formation; 13‒section 2 of the latest Silurian Zhuowukuo Formation; 14‒granite veins; 15‒granodiorite veins; 16‒Huashiguan Formation; 17‒fault; 18‒gold deposit; 19‒range of geochemical data collection; 20‒place name.
-
Figure 2.
Soil geochemical map of the Daqiao area in Gansu Province, China.
-
Figure 3.
Flow chart for soil geochemistry measurements and the proposed method of prediction for deposit prospecting.
-
Figure 4.
Method for generation of the training dataset.
-
Figure 5.
Diagram of the prospecting prediction method used for matching soil geochemistry characteristics with subsurface mineralization (deposit) distribution.
-
Figure 6.
Accuracy and loss entropy curves of the model training and verification used.
-
Figure 7.
A comparison of different network models and the prediction results obtained. a‒vgg model, grid size 23 m × 23 m; b‒vgg model, grid size 10 m × 10 m; c‒Googlenet Inception v1 model, grid size 23 m × 23 m; d‒Googlenet Inception v1 model, grid size 10 m × 10 m.
-
Figure 8.
The prospecting map of the Daqiao mining area was predicted using the proposed method.
-
Figure 9.
Comparison of the predicted prospecting area delineated using the proposed method and the manually determined prediction area.