A Dual-Branch Geochemical Prospecting Anomaly Detection Model with Spectral-Spatial and Spatial Correlation Fusion
-
摘要:
建立能兼顾多元素地球化学空谱特征、有效拟合数据复杂分布的检测模型，是识别异常区域的关键.针对新疆东昆仑高海拔深切割浅覆盖地区地球化学找矿异常提取难题，本研究提出一种空谱特征-空间关联双分支模型（Spatial-Spectral Feature and Global Spatial Correlation Network，SSGSNet），空谱特征分支基于ResNet残差块，融入双重注意力模块提取局部空谱特征；空间关联分支通过patch嵌入和自注意力机制挖掘全局空间关联特征.融入构造数据提高了地球化学综合异常找矿的准度，SHAP值也解释了模型中断裂的关键作用.实验结果表明，SSGSNet模型的AUC值达0.945 3，显著优于ResNet、ViT单模型和普通的空谱双分支模型.野外查证显示，遥西、巴什干克等4处高异常区均发现不同程度金矿化现象，证实该模型可有效解决复杂背景下地球化学异常信息提取难题，为覆盖区矿产勘探提供了可靠的技术支撑与靶区指导.
Abstract:Establishing a detection model that can take into account the multi-element geochemical spatial-spectral characteristics and effectively fit the complex distribution of data is the key to identification of abnormal areas. In response to the challenge of extracting geochemical prospecting anomalies in the high-altitude, deep-cutting, and shallow-coverage areas of the Eastern Kunlun Mountains in Xinjiang, this study proposes a Spatial-Spectral Feature and Global Spatial Correlation Network (SSGSNet). Based on ResNet residual blocks, the spatial-spectral feature branch is integrates a dual-attention module to extract local spatial-spectral features, with the spatial correlation branch using patch embedding and self-attention mechanisms to mine global spatial correlation features. Incorporating tectonic data improves the accuracy of geochemical prospecting, and SHAP values explain the critical role of faults within the model. Experimental results show that the AUC value of the SSGSNet model reaches 0.945 3, significantly outperforming the ResNet and ViT single models as well as the conventional spatial-spectral dual-branch model. Field verification shows that gold mineralization phenomena of varying degrees were found in four high-anomaly areas, including Yaoxi and Bashiganike, which confirms that the model can effectively solve the problem of extracting complex background geochemical anomaly information, providing reliable technical support and target area guidance for mineral exploration in covered areas.
-
图 3 研究区构造单元划分及采样点分布
底图据郭利（2020）；大地构造单元划分：Ⅰ-1.塔里木陆块；Ⅱ-1.阿尔金弧盆系；Ⅱ-1-1.阿中地块；Ⅱ-1-2.阿南蛇绿构造混杂岩；Ⅱ-2-1.西昆仑北带；Ⅱ-2-2.库地‒其曼于特蛇绿混杂岩带；Ⅱ-2-4.西昆仑南带；Ⅱ-2-5.东昆仑柴南（祁曼塔格弧后）弧盆系；Ⅱ-2-6.东昆仑南带；Ⅱ-2-7.康西瓦‒木孜塔格结合带；Ⅲ-1.巴颜喀拉地块
Fig. 3. Tectonic unit subdivision and sampling site distribution of the study area
图 4 多元素地球化学浓度分布（a~n）和断裂影响域（o）
Fig. 4. Multi-element geochemical concentration distribution (a-n) and fracture influence zones(o)
图 5 Au元素K-means聚类及正负样本分布
Fig. 5. K-means clustering of Au elements and distribution of positive and negative samples
图 7 不同特征输入的双分支模型异常识别结果
a.化探+构造；b.化探
Fig. 7. Anomaly recognition results of the dual-branch model with different feature inputs
图 11 不同模型异常识别结果对比
a.ResNet结果；b.ViT结果；c.DualBranchNet结果
Fig. 11. Comparison of anomaly recognition results of different models
表 1 改进型残差块的结构
Table 1. Structure of the improved residual block
单元 残差块 层 内核大小 步长 输入/输出 填充 Layer1 ResidualBlock1 Conv1+BN+ReLU 3×3 1 64/64 1 Conv2+BN 3×3 1 64/64 1 CA 64/64 SA 64/64 Shortcut(Identity)+ReLU ResidualBlock2 Conv1+BN+ReLU 3×3 1 64/64 1 Conv2+BN 3×3 1 64/64 1 CA 64/64 SA 64/64 Shortcut(Identity)+ReLU Layer2 ResidualBlock3 Conv1+BN+ReLU 3×3 2 64/128 1 Conv2+BN 3×3 1 128/128 1 CA 128/128 SA 128/128 Shortcut(Conv3+BN)+ReLU 1×1 2 64/128 ResidualBlock4 Conv1+BN+ReLU 3×3 1 64/128 1 Conv2+BN 3×3 1 128/128 1 CA 128/128 SA 128/128 Shortcut(Identity)+ReLU Layer3 ResidualBlock5 Conv1+BN+ReLU 3×3 2 128/256 1 Conv2+BN 3×3 1 256/256 1 CA 256/256 SA 256/256 Shortcut(Conv3+BN)+ReLU 1×1 2 128/256 ResidualBlock6 Conv1+BN+ReLU 3×3 1 256/256 1 Conv2+BN 3×3 1 256/256 1 CA 256/256 SA 256/256 Shortcut(Identity)+ReLU Layer4 ResidualBlock7 Conv1+BN+ReLU 3×3 2 256/512 1 Conv2+BN 3×3 1 512/512 1 CA 512/512 SA 512/512 Shortcut(Conv3+BN)+ReLU 1×1 2 256/512 ResidualBlock8 Conv1+BN+ReLU 3×3 1 512/512 1 Conv2+BN 3×3 1 512/512 1 CA 512/512 SA 512/512 Shortcut(Identity)+ReLU 
下载: 导出CSV
表 2 典型金（锑）矿床主要地质特征
Table 2. Main geological characteristics of typical gold (antimony) deposits
地质特征 屈库勒克东 巴什干克 赋矿地层 上石炭统哈拉米兰河群上亚群（C2H2）为主要的含矿地层，赋矿岩性主要有粉砂岩、砂岩、断层破碎蚀变岩、灰岩、闪长玢岩及石英脉等. 下奥陶统库拉甫河岩组第三段（O1K3）地层的灰岩、白云岩、玄武岩、安山岩等岩石及其碎裂岩、构造角砾岩中. 控矿构造 矿体分布主要受近EW向F2断裂控制，在挤压揉皱构造作用较发育地段，一般Au、Sb含量较高. F1、F2（阿尔金南缘断裂衍生次级断裂）控制赋金的韧‒脆性剪切带分布，韧‒脆性剪切带中的次级构造破碎蚀变带控制着主矿体的空间位置. 岩体岩脉 主要为中酸性脉岩，包括闪长玢岩脉、石英脉、石英方解石脉.闪长玢岩脉呈细‒大脉状、透镜状产出，碎裂岩化较强. 基性‒中酸性岩体均有出露，金矿（化）主要分布于闪长玢岩和花岗闪长岩岩体边部及其周围的地层中. 围岩蚀变 断层破碎带、硅化、褐铁矿化、黄钾铁钒化、黄铁矿化是寻找矿（化）体的重要标志，而辉锑矿化、锑华是寻找金锑的直接标志. 蚀变类型主要以黄铁矿化、毒砂化、硅化、碳酸盐化、绢云母化为主，在石英脉、碳酸盐脉充填贯入的部位，矿体含金品位高. 矿体特征 近EW向展布，产状总体为340°~15°∠53°~75°.矿体形态较规则，呈似层状、脉状，局部具有分支复合现象.矿石主要金属矿物为辉锑矿、自然金、黄铁矿，氧化矿物有锑华、褐铁矿、孔雀石等.平均品位Au为7.68×10⁻⁶、Sb为3.05%.金资源量为18 023.32 kg，锑资源量为71 685.82 t. 总体走向为NE-SW向，倾向NW，平均倾角45°~75°.空间上呈叠瓦式或雁列式排列展布，多呈脉状、似层状，少量呈透镜状产出.载金矿物主要为黄铁矿、毒砂、辉锑矿、钛铁矿、赤（褐铁矿）等.平均品位3.9×10⁻⁶，金资源量70.78 t. 地球化学特征 1∶5万水系沉积物测量圈定异常元素组合为Au、Sb、As、Pb元素，主成矿元素为Au、Sb.钻孔原生晕显示各指示元素的分带序列从上到下依次为As-Sb-Pb-Zn-Au-Ag-W. Cu、Au以及部分Zn、Pb、Ag的高背景异常区，可作为形成Cu-Au（多金属）矿床的矿源层，地化剖面显示Au与As、Sb、Hg等元素套合较好.
下载: 导出CSV
表 3 筛选的水系沉积物地球化学元素及依据
Table 3. Geochemical elements of water system sediments screened and their basis
单元素 地质意义 统计支持 Au 目标矿种，高偏度/峰度指示局部富集，成矿因子核心元素 描述性统计、PC3 Sb 目标矿种，与Au共生，成矿因子核心元素，高偏度反映断裂控矿特征 描述性统计、PC3、相关系数 Hg 热液前缘晕指示元素，高峰度反映深部矿体挥发性组分迁移，与Au、Sb正相关 描述性统计、地质类比 Bi、Mo、Sn 热液型矿床伴生元素，高偏度指示隐伏岩体接触带 描述性统计、PC3 W、Co 高熔点热液元素，常见于花岗岩相关热液矿床（如钨金共生），Co与基性岩背景相关 PC3、地质类比 Zn 中低温热液元素，可能指示多金属硫化物阶段 PC3、相关系数 As 低温热液标型元素，与Sb密切相关（如辉锑矿‒毒砂组合） PC3、聚类分析 Pb、Cu 多金属硫化物组合成员，反映热液成矿多阶段特征 PC3、相关系数 Li 隐伏花岗岩体接触带，热液蚀变 PC3 Ag 亲硫性元素，在热液活动中易于富集 聚类分析 注：PC3为PCA分析中第3因子，体现主要成矿元素.
下载: 导出CSV
-
Chen, G. X., Cheng, Q. M., 2018. Fractal-Based Wavelet Filter for Separating Geophysical or Geochemical Anomalies from Background. Mathematical Geosciences, 50(3): 249-272. https://doi.org/10.1007/s11004-017-9707-9 Cheng, Q. M., Agterberg, F. P., 1996. Multifractal Modeling and Spatial Statistics. Mathematical Geology, 28(1): 1-16. https://doi.org/10.1007/BF02273520 Cheng, Q. M., Agterberg, F. P., Ballantyne, S. B., 1994. The Separation of Geochemical Anomalies from Background by Fractal Methods. Journal of Geochemical Exploration, 51(2): 109-130. https://doi.org/10.1016/0375-6742(94)90013-2 Cheng, Q. M., Xu, Y. G., Grunsky, E., 2000. Integrated Spatial and Spectrum Method for Geochemical Anomaly Separation. Natural Resources Research, 9(1): 43-52. https://doi.org/10.1023/A:1010109829861 Ding, K., Xue, L. F., Ran, X. J., et al., 2023. CNN2D-SENet-Based Prospecting Prediction Method: A Case Study from the Cu Deposits in the Zhunuo Mineral Concentrate Area in Tibet. Minerals, 13(6): 730. https://doi.org/10.3390/min13060730 Dosovitskiy, A., Beyer, L., Kolesnikov, A., et al., 2020. An Image Is Worth 16×16 Words: Transformers for Image Recognition at Scale. arXiv, 2010.11929. https://arxiv.org/abs/2010.11929 Guo, L., Bai, H., Xing, L., et al., 2020. Metallogenic Regularity and Metallogenic Model of Late Paleozoic Hydrothermal Gold-Antimony Deposits in the Muztag Area, East Kunlun, Xinjiang. West-China Exploration Engineering, 32(5): 114-117 (in Chinese with English abstract). Harris, R. J., 2001. A Primer of Multivariate Statistics (3rd Edition). Psychology Press, New York. https://doi.org/10.4324/9781410600455 Hu, J., Shen, L., Sun, G., 2018. Squeeze- and -Excitation Networks. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City. https://doi.org/10.1109/CVPR.2018.00745 Li, Z. T., Xue, L. F., Ran, X. J., et al., 2022. Intelligent Prospect Prediction Method Based on Convolutional Neural Network: A Case Study of Copper Deposits in Longshoushan Area, Gansu Province. Journal of Jilin University (Earth Science Edition), 52(2): 418-433 (in Chinese with English abstract). Liu, Y. P., Zhu, L. X., Zhou, Y. Z., 2020. Experimental Research on Big Data Mining and Intelligent Prediction of Prospecting Target Area—Application of Convolutional Neural Network Model. Geotectonica et Metallogenia, 44(2): 192-202 (in Chinese with English abstract). Luo, Z. J., Xiong, Y. H., Zuo, R. G., 2020. Recognition of Geochemical Anomalies Using a Deep Variational Autoencoder Network. Applied Geochemistry, 122: 104710. https://doi.org/10.1016/j.apgeochem.2020.104710 Luo, Z. J., Zuo, R. G., 2025. Causal Discovery and Deep Learning Algorithms for Detecting Geochemical Patterns Associated with Gold-Polymetallic Mineralization: A Case Study of the Edongnan Region. Mathematical Geosciences, 57(1): 193-220. https://doi.org/10.1007/s11004-024-10153-6 Masci, J., Meier, U., Cireşan, D., et al., 2011. Stacked Convolutional Auto-Encoders for Hierarchical Feature Extraction. Springer, Berlin. https://doi.org/10.1007/978-3-642-21735-7_7 Reimann, C., Filzmoser, P., Garrett, R. G., et al., 2008. Statistical Data Analysis Explained: Applied Environmental Statistics with R. Wiley, Hoboken. https://doi.org/10.1002/9780470987605 Wang, J., Zuo, R. G., Caers, J., 2017. Discovering Geochemical Patterns by Factor-Based Cluster Analysis. Journal of Geochemical Exploration, 181: 106-115. https://doi.org/10.1016/j.gexplo.2017.07.006 Wang, Z. Y., Li, T., Zuo, R. G., 2024. Leucogranite Mapping via Convolutional Recurrent Neural Networks and Geochemical Survey Data in the Himalayan Orogen. Geoscience Frontiers, 15(1): 101715. https://doi.org/10.1016/j.gsf.2023.101715 Xin, L., 2023. Genesis of the Large Qukulekedong Au-Sb Deposit, East Kunlun Orogen, China (Dissertation). Chengdu University of Technology, Chengdu (in Chinese with English abstract). Xiong, Y. H., Zuo, R. G., Luo, Z. J., et al., 2022. A Physically Constrained Variational Autoencoder for Geochemical Pattern Recognition. Mathematical Geosciences, 54(4): 783-806. https://doi.org/10.1007/s11004-021-09979-1 Xiong, Y. H., Zuo, R. G., Wang, K. X., et al., 2018. Identification of Geochemical Anomalies via Local RX Anomaly Detector. Journal of Geochemical Exploration, 189: 64-71. https://doi.org/10.1016/j.gexplo.2017.06.021 Xu, Y., Shi, L. Y., Zuo, R. G., 2024. Geologically Constrained Unsupervised Dual-Branch Deep Learning Algorithm for Geochemical Anomalies Identification. Applied Geochemistry, 174: 106137. https://doi.org/10.1016/j.apgeochem.2024.106137 Xu, Y., Zuo, R. G., 2024. An Interpretable Graph Attention Network for Mineral Prospectivity Mapping. Mathematical Geosciences, 56(2): 169-190. https://doi.org/10.1007/s11004-023-10076-8 Xu, Y., Zuo, R. G., Zhang, G. B., 2023. The Graph Attention Network and Its Post-Hoc Explanation for Recognizing Mineralization-Related Geochemical Anomalies. Applied Geochemistry, 155: 105722. https://doi.org/10.1016/j.apgeochem.2023.105722 Yang, J. X., Zhao, Y. Q., Chan, J. C., 2017. Learning and Transferring Deep Joint Spectral-Spatial Features for Hyperspectral Classification. IEEE Transactions on Geoscience and Remote Sensing, 55(8): 4729-4742. https://doi.org/10.1109/TGRS.2017.2698503 Yin, B. J., Zuo, R. G., Xiong, Y. H., et al., 2021. Knowledge Discovery of Geochemical Patterns from a Data-Driven Perspective. Journal of Geochemical Exploration, 231: 106872. https://doi.org/10.1016/j.gexplo.2021.106872 Yousefi, M., Kamkar-Rouhani, A., Carranza, E. J. M., 2014. Application of Staged Factor Analysis and Logistic Function to Create a Fuzzy Stream Sediment Geochemical Evidence Layer for Mineral Prospectivity Mapping. Geochemistry: Exploration, Environment, Analysis, 14(1): 45-58. https://doi.org/10.1144/geochem2012-144 Yu, S. Y., Deng, H., Liu, Z. K., et al., 2024. Identification of Geochemical Anomalies Using an End-to-End Transformer. Natural Resources Research, 33(3): 973-994. https://doi.org/10.1007/s11053-024-10334-4 Yu, X. L., Li, H., Wei, X. L., et al., 2025. Zircon U-Pb Age and Geochemical Characteristics of Granitic Magmatic Rocks in Mailong Gold Deposit, East Kunlun, and Their Geological Significance. Earth Science, 50(6): 2107-2123 (in Chinese with English abstract). Zhang, C. J., Zuo, R. G., Xiong, Y. H., 2021. Detection of the Multivariate Geochemical Anomalies Associated with Mineralization Using a Deep Convolutional Neural Network and a Pixel-Pair Feature Method. Applied Geochemistry, 130: 104994. https://doi.org/10.1016/j.apgeochem.2021.104994 Zuo, R. G., 2019. Exploration Geochemical Data Mining and Weak Geochemical Anomalies Identification. Earth Science Frontiers, 26(4): 67-75 (in Chinese with English abstract). Zuo, R. G., Carranza, E. J. M., Wang, J., 2016. Spatial Analysis and Visualization of Exploration Geochemical Data. Earth-Science Reviews, 158: 9-18. https://doi.org/10.1016/j.earscirev.2016.04.006 Zuo, R. G., Cheng, Q. M., Xu, Y., et al., 2024. Explainable Artificial Intelligence Models for Mineral Prospectivity Mapping. Scientia Sinica (Terrae), 54(9): 2917-2928 (in Chinese with English abstract). doi: 10.1360/N072024-0018 Zuo, R. G., Wang, J., Chen, G. X., et al., 2015. Identification of Weak Anomalies: A Multifractal Perspective. Journal of Geochemical Exploration, 148: 12-24. https://doi.org/10.1016/j.gexplo.2014.05.005 Zuo, R. G., Xiong, Y. H., Wang, J., et al., 2019. Deep Learning and Its Application in Geochemical Mapping. Earth-Science Reviews, 192: 1-14. https://doi.org/10.1016/j.earscirev.2019.02.023 Zuo, R. G., Xu, Y., 2023. Graph Deep Learning Model for Mapping Mineral Prospectivity. Mathematical Geosciences, 55(1): 1-21. https://doi.org/10.1007/s11004-022-10015-z Zuo, R. G., Xu, Y., 2024. A Physically Constrained Hybrid Deep Learning Model to Mine a Geochemical Data Cube in Support of Mineral Exploration. Computers & Geosciences, 182: 105490. https://doi.org/10.1016/j.cageo.2023.105490 郭利, 柏辉, 邢令, 等, 2020. 新疆东昆仑木孜塔格地区晚古生代热液型金锑矿成矿规律及成矿模式. 西部探矿工程, 32(5): 114-117. 李忠潭, 薛林福, 冉祥金, 等, 2022. 基于卷积神经网络的智能找矿预测方法: 以甘肃龙首山地区铜矿为例. 吉林大学学报(地球科学版), 52(2): 418-433. 刘艳鹏, 朱立新, 周永章, 2020. 大数据挖掘与智能预测找矿靶区实验研究: 卷积神经网络模型的应用. 大地构造与成矿学, 44(2): 192-202. 邢令, 2023. 东昆仑屈库勒克东大型Au-Sb矿床成因研究(博士学位论文). 成都: 成都理工大学. 于小亮, 李华, 魏小林, 等, 2025. 东昆仑迈龙金矿区花岗质岩浆岩锆石U-Pb定年和地球化学特征及其地质意义. 地球科学, 50(6): 2107-2123. doi: 10.3799/dqkx.2025.046 左仁广, 2019. 勘查地球化学数据挖掘与弱异常识别. 地学前缘, 26(4): 67-75. 左仁广, 成秋明, 许莹, 等, 2024. 可解释性矿产预测人工智能模型. 中国科学: 地球科学, 54(9): 2917-2928. -
点击查看大图
计量
- 文章访问数: 96
- HTML全文浏览量: 48
- PDF下载量: 34
- 被引次数: 0
