【作者】 刘岳；

【导师】 陈翠华；

【作者基本信息】 成都理工大学 ， 矿物学、岩石学、矿床学， 2011， 硕士

【摘要】 矿产资源短缺、勘查成本和勘查难度剧增的新形势下,如何充分利用海量地质空间数据和现代信息技术,开展快速、高效的潜在矿产资源评价方法及应用研究,对于掌握区域矿产资源潜力,进行矿产勘查部署和选择具体的找矿靶区,从而降低矿产勘查的成本,具有重要的科学意义和实际应用价值。本论文结合前人研究成果,通过对青海东昆仑成矿带的区域构造特征、区域岩浆岩特征、区域地层特征、成矿地质特征、时空演化特征以及地球物理特征的分析和深入探讨,将青海东昆仑成矿带划分为五个Ⅳ级成矿带:昆北成矿带、昆中成矿带、昆南成矿带、都兰-鄂拉山成矿带和阿尼玛卿成矿带。通过对青海东昆仑成矿带成矿系列和典型矿床组合的深入分析,建立了矿床找矿模型,并建立了青海东昆仑成矿带金属矿床成矿系列,将金属矿床划分为两大矿床组合、7个成矿系列和14个矿床式。以数据驱动模型(证据权模型、扩展证据权模型、逻辑斯蒂回归模型)为理论基础,结合实际应用需求,开展青海东昆仑地区潜在矿产资源评价。基于数据驱动模型的矿产资源评价技术流程包括:图层的选取依据、图层的重要性评估、图层相关性处理、计算参数检验以及靶区定量化评价。本研究应用证据权模型中的邻近度分析定量刻画了线性控矿因素(断裂、破碎带、褶皱、接触带、蚀变带等)和离散数据(物探数据、化探数据等)与已知矿点之间的空间相关性,用卡方检验、Kolmogorov-Smirnov检验或NOT检验对证据图层之间的相关性进行处理,去除相关性较大的图层,避免过大圈定靶区。应用逻辑斯蒂回归模型进行潜在矿产资源评价时,是在证据权模型中邻近度分析的基础之上,选取最佳的二值证据图层,然后用逻辑斯蒂回归参数(如回归系数、Wald s统计量、回归系数的标准误等)对控矿因素的重要性排序和评价,进而应用非线性回归概率绘制潜在矿产资源预测图。以区域地质成矿特征和矿床成矿系列和典型矿床模型为理论指导,综合分析不同矿床类型下的控矿因素,建立找矿标志模型。综合多源地质空间数据(基础地质数据、地球物理数据、地球化学数据、遥感数据、已知勘探的矿床/点数据等),应用证据权模型、扩展证据权模型和逻辑斯蒂回归模型分别对青海东昆仑成矿带、野马泉成矿亚带、五龙沟成矿区三个不同空间尺度下的不同矿种开展潜在矿产资源评价。具体应用工作包括:基于证据权模型的青海东昆仑成矿带铜铅锌多金属矿、铁矿和金矿资源评价,基于证据权模型的野马泉成矿亚带铁多金属矿资源评价,基于证据权模型的五龙沟成矿区金矿资源评价;基于扩展证据权模型的青海东昆仑成矿带潜在铜铅锌多金属矿资源评价,基于扩展证据权模型的五龙沟成矿区金矿资源评价;基于逻辑斯蒂回归模型的青海东昆仑成矿带铜铅锌多金属矿和铁矿资源评价,基于逻辑斯蒂回归模型的野马泉成矿亚带铁多金属矿资源评价。各评价模型(证据权模型、扩展证据权模型、逻辑斯蒂回归模型)的实验结果具体如下:(1)基于证据权模型的青海东昆仑成矿带铁矿资源评价结果显示77%的已知矿点落入中高潜力区,其中高潜力区(总面积的11%)包含了56%的已知矿点,中潜力区(总面积的10%)包含了21%的已知矿点;青海东昆仑成矿带金矿资源评价结果显示74.5%的已知矿点落入中高潜力区,其中高潜力区(总面积的9%)包含了48.9%的已知矿点,中潜力区(总面积的11%)包含25.6%的已知矿点;青海东昆仑成矿带铜铅锌多金属矿资源评价结果显示71.2%的已知矿点落入中高潜力区,其中高潜力区(总面积的8%)包含了31.8%的已知矿点,预测率为47.5%;中潜力区(总面积的19%)包含了39.4%的已知矿点,预测率为25%;野马泉成矿亚带铁多金属矿资源评价结果显示85.7%的已知矿点落入中高潜力区,其中高潜力区(总面积的3.4%)包含了61.9%的已知矿点,中潜力区(总面积的6.2%)包含了23.8%的已知矿点;五龙沟成矿区金矿资源评价结果显示76.5%的已知矿点落入中高潜力区,其中高潜力区(总面积的5.9%)包含70.6%已知矿点,中潜力区(总面积的7.2%)包含了5.9%的已知矿点。(2)基于扩展证据权模型的青海东昆仑成矿带铜铅锌多金属矿资源评价结果显示84.9%的已知点落入中高潜力区,其中高潜力区(总面积的10%)包含了51.9%的已知矿点,中潜力区(总面积的15%)包含了33%的已知矿点;五龙沟成矿区金矿资源评价结果显示94.1%的已知矿点落入中高潜力区,其中高潜力区(总面积的2.6%)包含了76.5%的已知矿点,中潜力区(总面积的3.6%)包含了17.6%的已知矿点。(3)基于逻辑斯蒂回归模型的青海东昆仑成矿带铁多金属矿资源评价结果显示71.6%的已知矿点落入中高潜力区,其中高潜力区(总面积的8.3%)包含43.2%的已知矿点,中潜力区(总面积的15.9%)包含了28.4%的已知矿点;青海东昆仑成矿带铜铅锌多金属矿资源评价结果显示69%的已知矿点落入中高潜力区,其中高潜力区(总面积的8.5%)包含38%的已知矿点,中潜力区(总面积的16.4%)包含了31%的已知矿点;野马泉成矿亚带铁多金属矿资源评价结果显示85.7%的已知矿点落入中高潜力区,其中高潜力区(总面积的4.3%)包含76.2%已知矿点,中潜力区(总面积的5.2%)包含9.5%的已知矿点。不同预测模型的精度表明,在大区域尺度下,扩展证据权模型的较证据权模型和逻辑斯蒂回归模型高,而证据权模型高于逻辑斯蒂回归模型。随着研究区范围减小和研究程度提高,三个评价模型之间的预测精度的差别趋于减小。更多还原

【Abstract】 With the sharp increasing of the Shortage of mineral resources, prospecting costs and exploration difficulty in the new situation, how to combine the massive geological spatial data with modern information technology efficiently for the study of mineral potential resource evaluation is favorable to mastering regional mineral resource potential, prospecting deployment and selecting specific target, thereby reducing the cost of mineral exploration, which has an important scientific significance and practical value.According to tectonic features, regional magmatic characteristics, regional stratigraphy, geological characteristics, Spatio-temporal evolution characteristics and geophysical characteristics, the East Kunlun metallogenic belt can be divided into five secondary metallogenic belt:Kunbei belt, Kunzhong belt, Kunnan belt, Dulan-Elashan belt and Animaqing belt. Through analysing and summarizing typical deposits, the metallic deposits within the East Kunlun metallogenic belt can be divided into two types of deposit assemblage, 7 minerogenetic series and 14-type deposits.Based on the theory of data-driven models, such as weights of evidence model, extended weights of evidence model, and logistic regression model, integrating with practical applications, mineral potential evaluation was carried out in the East Kulun region, Qinghai province. The technical processes of the data-driven model including: evaluating the importance of the layers, processing correlation among layers, calculating test’s parameters and target’s quantitative evaluation. In this paper, weights of evidence model was used to quantitatively characterize the spatial correlation between ore-controlling factors of linear (faults, fracture zones, folds, the contact zone, alteration, etc.) or discrete data (geophysical data, geochemical data, etc.) and discovered deposits/occurrences. The methods used are chi-square test, Kolmogorov-Smirnov test, or "NOT" test, which can remove high correlation among map layers, and avoid excessive delineating targets. Using the Logistic regression model for mineral potential evaluation, the optimal binary evidential maps were derived from proximity analysis of weights of evidence model, then logistic regression parameters (such as the regression coefficient Wald’s statistic, standard error of regression coefficients, etc.) were calculated to rank ore-controlling factors based on the importance of some parameters, last mapping mineral potential according to the nonlinear regression probability.Regional metallogenic characteristics, minerogenetic series and typical deposit model as theoretical guidances, combining with multi-source geological data, such as geophysical data, geochemical data, remote sensing data with discovered deposits /occurrences, comprehensive analysis of different deposit types under the ore-con- trolling factors, constructing prospecting model. The weight of evidence model, the extended weight of evidence model, and the logistic regression model were used to map mineral potential for three study areas, the East Kunlun metallogenic belt, the Yemaquan secondary metallogenic belt and the Wulonggou metallogenic district with different spatial scales. The work has been done as follows: mineral resource evaluation in the East Kunlun region belt based on weights of evidence model for copper-lead-zinc polymetallic resources, gold resources and iron resources, mineral resource evaluation for iron polymetallic resources in the Yemaquan secondary metallogenic belt based on weights of evidence model, mineral resource evaluation for gold resources in the Wulonggou metallogenic district based on weights of evidence model; mineral resource evaluation for copper-lead-zinc polymetallic resources in the East Kunlun region belt based on extended weights of evidence model, mineral resource evaluation for gold resources in the Wulonggou metallogenic district based on extended weights-of-evidence model; mineral resource evaluation in the East Kunlun region belt based on logistic regression model for copper-lead-zinc polymetallic resources and iron resources, mineral resource evaluation for iron polymetallic resources in the Yemaquan secondary metallogenic belt based on logistic regression model.The experimental results of the evaluation models (weights of evidence model, extended weights of evidence model, logistic regression model) indicate that: (1) The results derived from the evaluation of iron resources in the East Kunlun metallogenic belt based on weights of evidence model indicate that high and moderate potential area contains 77% of the total deposits, among which the high potential area occupies 11% of the total area, containing 56% deposits, and the moderate potential area occupies 10%, containing 21% deposits.The results derived from the evaluation of gold resources in the East Kunlun metallogenic belt based on weights of evidence model indicate that high and moderate potential area contains 74.5% of the total deposits, among which the high potential area occupies 9% of the total area, containing 48.9% deposits, and the moderate potential area occupies 11%, containing 25.6% deposits.The results derived from the evaluation of copper-lead-zinc polymetallic resources in the East Kunlun metallogenic belt based on weights of evidence model indicate that high and moderate potential area contains 71.2% of the total deposits, among which the high potential area occupies 8% of the total area, containing 31.8% deposits, and predicting 47.5% deposits, the moderate potential area occupies 19%, containing 39.4% deposits, and predicting 25% deposits.The results derived from the evaluation of iron polymetallic resources in the Yemaquan secondary metallogenic belt based on weights of evidence model indicate that high and moderate potential area contains 85.7% of the total deposits, among which the high potential area occupies 3.4% of the total area, containing 61.9% deposits, and the moderate potential area occupies 6.2%, containing 23.8% deposits. The results derived from the evaluation of gold resources in the Wulonggou metallogenic district based on weights of evidence model indicate that high and moderate potential area contains 76.5% of the total deposits, among which the high potential area occupies 5.9% of the total area, containing 70.6% deposits, and the moderate potential area occupies 7.2%, containing 5.9% deposits. (2) The results derived from the evaluation of copper-lead-zinc polymetallic resources in the East Kunlun metallogenic belt based on extended weights of evidence model indicate that high and moderate potential area contains 84.9% of the total deposits, among which the high potential area occupies 10% of the total area, containing 51.9% deposits, and the moderate potential area occupies 15%, containing33% deposits. The results derived from the evaluation of gold resources in the Wulonggou metallogenic district based on extended weights of evidence model indicate that high and moderate potential area contains 94.1% of the total deposits, among which the high potential area occupies 2.6% of the total area, containing 76.5% deposits, and the moderate potential area occupies 3.6%, containing 17.6% deposits. (3) The results derived from the evaluation of iron resources in the East Kunlun metallogenic belt based on logistic regression model indicate that high and moderate potential area contains 71.6% of the total deposits, among which the high potential area occupies 8.3% of the total area, containing 43.2% deposits, and the moderate potential area occupies 15.9%, containing 28.4% deposits. The results derived from the evaluation of copper-lead-zinc polymetallic resources in the East Kunlun metallogenic belt based on logistic regression model indicate that high and moderate potential area contains 69% of the total deposits, among which the high potential area occupies 8.5% of the total area, containing 38% deposits, and the moderate potential area occupies 16.4%, containing 31% deposits. The results derived from the evaluation of iron polymetallic resources in the Yemaquan secondary metallogenic belt based on logistic regression model indicate that high and moderate potential area contains 85.7% of the total deposits, among which the high potential area occupies 4.3% of the total area, containing 76.2% deposits, and the moderate potential area occupies 5.2%, containing 9.5% deposits. The prediction accuracy of different data-driven models indicate that the extended weight of evidence model is higher than the weight of evidence model and the logistic regression model, whereas the weight of evidence model is higher than the logistic regression model at a large regional scale. With the scope decreasing and the research increasing of the study area, the difference of the prediction accuracy among data-driven models tends to decrease.更多还原