Geological, geophysical, geochemical and remote sensing geological anomalies of ore deposits and their surrounding areas constitute a basis for understanding the ore-forming conditions and evaluating mineral resource potential (Chen and Liu, 2001; Zhao et al., 2001, 2000, 1999, 1996; Zhao and Chen, 1999, 1998; Zhao and Meng, 1993; Zhao and Chi, 1991). Digital mineral exploration is realized by establishing digital mineral exploration patterns (Zhao, 2002; Chen and Liu, 2001). Digital mineral exploration firstly involves the extraction of single-disciplinary ore-finding information such as geology, geochemistry, geophysics and remote sensing data based on the establishment of a conceptual mineral exploration model with comprehensive information (Zhao and Chen, 2011; Wang et al., 2010, 1996, 1992). Then, the multi-disciplinary ore-finding anomaly information is integrated and used to establish a digital mineral exploration model. Finally, the ore-finding targets are quantitatively delineated and the resource potential evaluated by means of a digital mineral exploration pattern (Chen and Liu, 2001). This digital mineral exploration pattern is a quantitative expression of the ore-controlling factors and ore-finding evidence (Chen and Zhao, 2009).
The main applications of digital geology include digital characterization of geological bodies, their genetic analysis, evaluation and prediction of ore bodies (Zhao and Chen, 1999, 1998). Quantitative geoscience is the application of data science in geoscience, and digital mineral exploration is an important part of quantitative geoscience. It constitutes the application of data science in mineral exploration.
The superiority and necessity of digital mineral exploration include the following: (a) quantitative characterization of a geological body includes its structural, spatial, statistical, geometric, evolutionary characteristics; (Zhao, 1982); (b) quantification and digitization form the basis and are the prerequisite for fully extraction of the ore-finding information; (c) data analysis is necessary for the extraction of various kinds of ore-finding information, identification of deep geological structures and ore bodies, and quantitative evaluation of both ore-finding targets and establishing uncertainty of mineral exploration (Zhao, 2007); and (d) quantitative description of the history of pertinent geological events (Zhao, 1992).
The Gejiu super-large Sn-Cu polymetallic deposits are located at the western margin of the South China Block near the meeting point of the Yangtze, Indo-China, and Cathaysian blocks (Fig. 1a). Outcropping strata in the Gejiu area belong mainly to the Triassic Sedimentary Formation which consists of carbonates in the lower part, and clastic and carbonate rocks with the intercalated basic lavas in the upper part. The most significant fault in the Gejiu tin polymetallic ore field is the NS trending Gejiu fault, which separates the Gejiu field into eastern and western districts (Fig. 1b). Other, secondary faults are NE-trending, NW-trending and EW-trending. Intrusions in the Gejiu district include: (1) porphyritic granite and equigranular granite; (2) alkaline rocks; and (3) gabbro-monzonite. These intrusions are well exposed in the western part, but buried deeply in the eastern part (Fig. 1b). It has been shown that these intrusions are buried deeply in the eastern part, including the Malage-Songshujiao, Laochang-Kafang plutons are predominant S-granites which had experienced high degrees of fractionation. The intrusions exposed in the western part, including Jiasha gabbro-monzonite, Baiyunshan-Changlinggang alkali granites, as well as Longchahe porphyritic granites are predominantly I-granites. Sn-Cu polymetallic deposits are associated with the S-granites in the eastern Gejiu region, and U-Th-Nb-Ta-REE mineralization is related to the I-granites in the western Gejiu region (Chen et al., 2020).
Figure 1. (a) Geological outline map of southeastern Asia, showing major tectonic units and location of where the Gejiu tin-copper polymetallic ore field is situated and (b) simplified geological map showing the spatial distribution of intrusions and the various types of ore deposits in the Gejiu Sn-Cu polymetallic ore field and surrounding areas. CB. Cathaysia Block; YB. Yangtze Block; SB. Sibumasu Block; TP. Tibet Plate; ICB. Indo-China Block; IP. India Plate (revised after Chen et al., 2017).
Complicated geo-processes including evolution-differentiation of the magmas and hydrothermal systems resulted in the diversity of mineralization mentioned above (Chen et al., 2020; Zhang et al., 2008).
On the basis of the above-mentioned analyses of ore-forming features, an ore-finding conceptual model was established for the study area, including the following: (a) SN and NW, NE and EW trending fracture systems, and their secondary linear and ring structures, which control the spatial distribution of various types of Sn-Cu polymetallic deposits; (b) the Triassic (Gejiu area) and the Cambrian-Dubian Carbonate (Bozhushan area) are the main ore-bearing strata; (c) Late Cretaceous tectonic magmatism was a necessary factor for the formation of the granitic complex and related hydrothermal ore deposits; (d) Pb-Cd-Ag-Zn-Sn-Mn-Cu-As element combination anomalies with elemental zonation are developed in some areas; and (e) U-Th-Nb and Au-Sb element anomaly associations are important markers for identifying ore anomalies (Chen et al., 2020; Huang et al., 2018; Zhang et al., 2008). This comprehensive ore-finding conceptual model forms a basis for both selecting ore-finding target variables and establishing quantitative ore-occurrence patterns (Zhao and Chen, 1999, 1998).
The establishment of a quantitative evaluation pattern of ore-finding targets in the conceptual model includes the following: (a) define statistical units; (b) extract ore-finding target variables based on the conceptual model; (c) assign values to ore-finding target variables; and (d) choose the mathematical model to construct a favorability function.
Taking a 10×10 km grid as sampling unit for defining the target variables, there are 288 sample units within the study region. Eleven target variables were selected according to the ore finding target conceptual model. These are Sn-Cu polymetallic mineralization (X1), Late Cretaceous granites (X2), faults (X3), Triassic/Devonian/Cambrian carbonates (X4); Pb-Cd-Ag-Zn-Sn-Mn-Cu-As anomalies (X5), W-Be-Bi-Cu-As-Sn anomalies (X6), U-Th-Nb anomalies (X7), Au-Sb anomalies (X8); IMF1 negative gravity anomalies (showing concealed intrusions) and circular positive gravity anomalies (showing skarn alteration zones) (X9), IMF2 negative gravity anomaly (showing concealed intrusions)(X10), IMF3 negative gravity anomalies (showing concealed intrusions) (X11).
The variables defined for each sample unit are quantified in binary format with the values 1 or 0, representing favorable or unfavorable for mineralization, respectively (Zhao et al., 1999; Zhao et al., 1983). The resulting 24×12 data matrix will be used for estimating the ore-forming favorability. When appearing in an unit the ore-finding target variable is assigned to the value 1, otherwise it is assigned the value 0. Element association anomaly assignment satisfies the following rules: (a) element association anomaly with three grades zonations (inner, central, and outer zone) is assigned the value 3; (b) element association anomaly with two grade zonations (middle and outer zone) is assigned the value 2; element association anomaly with one zonation (outer zone) is assigned the value 1. The mathematical model of quantitative evaluation of ore-finding targets can be established on the basis of the ore-finding target variables characterizing the geological distribution of resources in the study area.
The weights of the 11 variables were calculated by principal-component analysis applied to the resulting matrix using characteristic analysis (McCammon et al., 1983), the function for calculating favorability of a given sample unit is as follows,
Equation 1 was applied to calculate favorabilities for all 228 sample units, which were grouped into nineteen groups within the interval 0.1. The resulting frequency distribution and cumulative frequency distribution are shown in Table 1 and on normal probability paper in Fig. 2, respectively.
Intervals 0.1–0.2 0.2–0.3 0.3–0.4 0.4–0.5 0.5–0.6 0.6–0.7 0.7–0.8 0.8–0.9 0.9–1.0 1.0–1.1 Frequency 2 10 17 60 51 37 28 23 16 9 Probability (%) 0.69 3.47 5.90 20.83 17.71 12.85 9.72 7.99 5.56 3.13 Cumulative probability (%) 0.69 4.17 10.07 30.90 48.61 61.46 71.18 79.17 84.72 87.85 Intervals 1.1–1.2 1.2–1.3 1.3–1.4 1.4–1.5 1.5–1.6 1.6–1.7 1.7–1.8 1.8–1.9 1.9–2.0 Frequency 12 6 4 3 4 3 0 2 1 Probability (%) 4.17 2.08 1.39 1.04 1.39 1.04 0.00 0.69 0.35 Cumulative probability (%) 92.01 94.10 95.49 96.53 97.92 98.96 98.96 99.65 100.00
Table 1. Frequency and cumulative frequency of sampling units in the study area
Figure 2. Cumulative frequency distribution of metallogenic favourable sample units in the study area.
The normal probability plot shows two distinct populations—one with favorability less than 0.8, on the left, representing background population of ore-forming favorability of units, and the other with values greater than 0.8 on the right representing the anomalous population. The favorability value 0.8, therefore, can be used as the threshold for delineating ore-forming anomaly districts. Thus, an ore-forming anomaly district refers to an area, which consists of a cluster of units with favorability value greater than 0.8.
Eight ore-finding targets with F > 0. 80 were delineated and selected as the targets for further exploration (Fig. 3). Figure 3 shows that the eight ore-finding targets cover all known large Sn-Cu and Ag-Pb-Zn polymetallic deposits as well as more than 90 per cent of medium-small deposits in the study area.
The ore-finding targets can be ranked further by an index defined as the product of the mean value of favourabilities of the ore-finding targets and the target area (Chen et al., 2001),
where Poi represents the ore-finding priority of ith target and ave (Fi) and Si the mean value of F, and the area of the ith target, respectively.
The value Poican be normalized as below (Chen et al., 2001)
where Pi is a normalized index standing for the priority of further exploration and POmax and POmin represent the maximum and the minimum values of POi, respectively. The ranks of the 8 targets are shown in Table 2. The Pivalue of the target Ⅴ (the Gejiu district) containing five known large Sn-Cu polymetallic deposits is 100%, and the Pi value of target Ⅶ (the Bozhushan district) containing one known large Ag-Pb-Zn-W polymetallic deposit is 5.43%. Thus, it is possible for the targets Ⅱ and Ⅲ in the Gejiu district that each of them would have resource potential of one large Sn-Cu polymetallic ore deposit, respectively, compared to the target Ⅴ. Similarly, in the Bozhushan district the target Ⅷ should contain resource potential with two large Ag-Pb-Zn-W polymetallic ore deposits compared to the target Ⅶ.
Number Ⅰ Ⅱ Ⅲ Ⅳ Ⅴ Ⅵ Ⅶ Ⅷ Si (km2) 7 11 12 8 39 3 6 9 Fi 0.965 7 1.10 1.05 0.86 1.26 0.86 0.85 0.91 Poi 5.32 12.1 12.6 6.88 49.14 2.58 5.11 8.19 Pi (%) 5.88 20.45 21.52 9.24 1 0 5.43 12.69 Predicted number of large deposits 0.27 1.02 1.08 0.46 5 0 1 2
Table 2. Optimality and probability of prospecting in target areas