A statistical analysis of 592 surficial samples in fracture zones shows that the raw element data set are characterized by positive values of skewness (>8), kurtosis (>80), and coefficient of variation (>3)(Table 1). As a measure of symmetry, the high skewness values were due to the extremely values in the data set, indicating non-normal data distributions. Kurtosis is a measure of whether the data are tailed relative to a normal distribution, and high kurtosis of this geochemical data tend to have heavy tails or outliers. Coefficient of variation (CV) is an important parameter to evaluate the degree of element differentiation. The element concentrations in fracture zones are characterized by high CV values, indicating inhomogeneous distribution and the possibility of local enrichment of elements, thus they are more likely to be concentrated proximal to ore according to the sample with high abnormal ore-forming elements. Also listed are the element concentrations of Gejiu Formation derived from 12 carbonate wall rocks and the concentration factor (CF). The latter represents the ratio of the mean value to the background values of Gejiu Formation, which can be used to indicate metallic element concentrations (Zhao et al., 2015). The CF values illustrate that rocks in fracture zones are characterized by much larger concentrations of ore-forming elements compared to the wall rocks. Combining the multiple filled veins (e.g., calcite and quartz veins) and alteration minerals in the fault zone, it implies that rocks in fracture zones might occur mass transfer during fluid-rock interactions and contain much ore-forming solutions.
Ele. Mean Lower quartile Upper Lower quartile Std. deviation Skewness Kurtosis CV Carbonate rocks of Gejiu Formation CF Ag 0.8 0.1 0.4 2.7 10.1 140.9 3.5 0.2 3.4 Sn 13.1 1.4 4.1 90.3 19.0 403.0 6.9 2.1 6.4 Bi 0.7 0.1 0.1 12.9 24.4 595.7 17.5 0.1 8.7 Hg 78.0 11.0 48.2 300.8 10.2 123.8 3.9 12.4 6.3 Mn 1 673.9 121.2 1 023.5 7 821.2 12.2 161.6 4.7 108.8 15.4 Cu 29.4 3.6 10.0 362.5 24.0 581.1 12.3 4.3 6.9 Zn 361.5 42.6 268.0 1 199.6 12.6 217.4 3.3 17.9 20.2 Mo 2.4 0.7 0.9 12.8 12.1 162.5 5.3 1.0 2.3 Cd 9.3 0.0 1.9 95.4 21.2 484.3 10.3 0.3 33.0 W 2.5 1.0 1.7 8.2 10.5 122.5 3.3 0.4 6.2 Pb 467.6 31.7 192.6 2 573.4 12.5 173.9 5.5 7.1 66.0 As 180.8 3.8 14.5 3 815.6 24.4 592.3 21.1 1.6 115.1 CF. Concentration factor, which is the ratio of the mean value to the average concentration of Gejiu Formation; CV. coefficient of variation, which is defined as the ratio of the standard deviation to the mean.
Table 1. Descriptive statistics of raw fault tectono-geochemical data in Laochang (units10-6, except Hg)
Histograms and Q-Q plots indicate that the most of variables are symmetrically distributed (Fig. 7). These distributions may be interpreted as mixed origins of the elements due to influence by secondary geological processes such as surface leaching or weathering in fracture zones (Zhao et al., 2015). The concentrations and distributions of metal elements may demonstrate the multiphase hydrothermal fluid activities in the fracture zones, which is in agreement with the occurrences of mineralization and alteration minerals.
Another statistical analysis displays the distribution characteristics of Sn and Cu in different fracture zones. The faults were divided into four groups, which are NE-, E-W-, NW-, and N-S-trending faults and their corresponding intervals of strike are defined as 15°-75°, 75°-105°, 105°-165°, and 165°-195°. The result of box-plot reveals that the main ore controlling faults may be NE and EW directions (Fig. 8). The areas surrounding the fault intersections also enrich Sn and Cu, indicating that intersection part could be appropriate for the penetration of ore-forming fluids to form mineralization because of the higher permeability.
In this research, principal components analysis (PCA) is utilized to transform the original fault tectono-geochemical data set of 592 surficial samples into a smaller set of linear combinations that represent most of the variance. To open the geochemical data and address the closure problem prior to PCA (Zuo et al., 2013; Carranza, 2011; Aitchison, 1986), the clr-transformation is applied to process the raw data by the following formula (Zuo et al., 2013; Carranza, 2011; Aitchison, 1986),
Four principal components with eigenvalues of 1.0 and above are: PC1-PC4 and these combined, accounted for 72.6% variability in the clr-transformed data (Fig. 9a). Figure 9b shows the bar graph of the loadings for the first principal component accounting for 37.2% variance. Sn, Cu, W, Mo, and Bi are in association of PC1, representing granite-related hydrothermal elements associated with Sn-Cu ploymetallic mineralization in Laochang district (Zhao et al., 2015; Zhuang et al., 1996; Geological Survey Team 308, 1984)). Group PC2, dominated by Pb and Zn, could be interpreted to represent a group of epithermal elements associated with Pb-Zn mineralization. Group PC3, dominated by Mn and Cd, likely presents element association which associate with carbonate sources. Group PC4, dominated by Ag, Hg, and As, likely presents low temperature epithermal element association which reflect granite-related hydrothermal processes.
Figure 9. The results of PCA showing: (a) eigenvalues of principal components; (b) component loading of PC1.
For generating reliable primary anomaly that can be used for further prospecting of the deep mineralization, inverse distance weighting was used as an interpolation approach for mapping Sn, Cu, W, Mo, and Bi. In order to obtain different levels of geochemical anomalies, the C-A plots consisting of the concentrations(c) versus the number of cells were obtained with the concentrations greater than or equal to c (Zhao et al., 2017). According to Fig. 10, there were more than two enrichment steps and the threshold values for each element were obtained. Then the anomaly maps with different levels for Sn, Cu, W, Mo, Bi were generated in Fig. 11, which also showed the score map of PC1. The Most of the high anomalies located in the area along the Mengzimiao fault and some secondary faults, presenting NW, NE, and EW direction.
Mineralizations can be regarded as products of a singular process characterized by power law models (Cheng, 2007). For identify the geochemical anomalies associated with the granite-related Sn-Cu ploymetallic mineralization in the study area, further analysis was required to recognize more well-defined features that are directly related to mineralization. In our research, based on PCA results, the MSVD method was used to process the score map of PC1 presenting the main ore-forming element combination for Sn-Cu mineralization. Equation 10 represents a power law plot of λ-E modeled by line segments having different slopes to determine p and q. The curve can be divided into three parts based on their different slopes, two break points, p=4, q=12 (Fig. 12). The left segment is made of ranking from λ13 to λ78, and the percentage of its energy only takes up about 0.008 6% of total energy, which might be present data errors without any geological significance (Chen et al., 2015). The percentage of the energy of the right segment from λ1 to λ4 is about 94.96% of total energy. The corresponding reconstructed map can be regarded as low-pass filtered image, which usually indicate the regional ore forming background (Fig. 13a). The reconstructed image from the λ1 to λ4 probably demonstrates the buried deeply geological bodies. Taking -0.2 as threshold value, the positive anomaly area is mainly distributed in the east and south of the study area, while the negative anomaly is mainly distributed in the west part. The junction of positive and negative anomalies possibly indicates the boundary of deep geological bodies, such as the granite and carbonate formation.
Figure 13. Reconstructed geochemical component images from (a) 1st to 4th eigenvalues and (b) 5th to 12th eigenvalues in the Laochang orefield.
The middle segment consists of ranking from λ5 to λ12, and the percentage of its energy is about 0.042% of total energy. The reconstructed map with the sum from 5th to 12th eigenvalues can be regarded as high-pass filtered image (Chen et al., 2015), which usually reflect the ore-forming elements of locally enrichment (Fig. 13b). The results show that most of the anomaly zones are distributed in NE zonal direction and are spatially coincident with the mineralization points. Consequently, the anomaly zones of the reconstructed map may be as the prospecting potential area, which have been confirmed by drilling and tunnel engineering. As shown in Fig. 12b, two ore-discovery drills are located in the anomaly area of the south of the Mengzimiao fault. Some ores are found in the core at depths of 800-1 000 m, frequently spatially localized along the fracture zone or contact zone of granite and carbonate rocks. Mineralization associated with different rock types, mostly as vein and fracture infillings and as disseminated forms. The average contents of Cu and Sn in those ores reach about 2.69% and 0.50% respectively. Therefore, these anomalies may provide prospecting guidance for the discovery of concealed orebodies.