三维成矿预测数据整合过程不确定性研究——以宁芜盆地钟姑矿田为例

基于多元地学大数据的三维成矿预测方法是开展深部找矿预测的新方法和新手段,也是当前成矿预测领域的研究热点之一。然而,大数据具有高维、混杂、非精确等特点,其分析处理过程面临多重不确定性。多元地学大数据整合是三维成矿预测的最终环节,其存在的不确定性将直接作用于预测结果,影响进一步的找矿应用和风险评估。本文以宁芜盆地钟姑矿田为例,从大数据思维出发,定量分析和度量预测要素和数学模型在数据整合过程中存在的不确定性及对三维成矿预测结果的影响。结果显示,断裂构造、背斜轴部等预测要素的不确定性对三维成矿预测结果的影响最为强烈;数据整合模型中,较之Logistic回归模型和证据权重模型,神经网络模型可能具有更高的不确定性程度。进一步工作可通过增强上述预测要素的可靠性和有效性、采用更多的数据整合模型进行更为全面的不确定性分析和评价,以获得更为可靠的三维成矿预测成果,从而降低成矿预测和找矿勘探风险。

Uncertainty analysis of data combination within 3D prospectivity modelling: A case study in Zhonggu orefield, Ningwu basin, China

Based on the multivariate geological big data, the three-dimensional prospectivity modeling has become one of the new methods and approaches to predict the concealed ore deposit in depth. However, multiple uncertainties exist in big data processing because of some inevitable shortages in big data, such as high dimension, mixed, imprecision, and so on. Data combination is the final step of the 3D prospectivity modeling. The uncertainty within this final step would directly affect the modeling results and influence the further exploration and risk evaluation. This paper takes Zhonggu orefield which is located in Ningwu basin, the Middle and Lower Yangtze River metallogenic belt as an example, to measure the uncertainty caused by exploration criteria and mathematic model on big data and detect its effect on 3D prosepctivity modeling results quantitatively based on the big data thinking. The results show that the uncertainty about the fault, anticlinal axis and other exploration criteria will affect the results of 3D prospectivity modelling strongly, and compared with other data combination models, Logistic regression model may has a higher uncertainty. The further work can focus on the enhancement of the reliable and efficiency of the exploration criteria above, and using more data combination models to evaluate the uncertainty to get more reliable and efficient 3D prosepctivity modeling results to provide support and help for reducing the risk of deep exploration.