Prediction of Au-Associated Minerals in Eastern Thailand Based on Stream Sediment Geochemical Data Analysis by S-A Multifractal Model

1. Introduction

Geochemical exploration is essential in identifying undiscovered mineral deposits [1,2,3,4]. It uses multiple methods to analyze data, revealing the geographic patterns of geochemical markers, identifying related elements, and detecting geochemical anomalies due to mineralization processes [5,6,7]. To investigate these processes, a regional geochemical survey is needed. Regional geochemical surveys are crucial in both environmental and mineral exploration studies [8,9,10]. They provide baseline data about the distribution and abundance of elements and compounds in a region, serving as a reference point for future studies. In environmental studies, these surveys help assess environmental quality, monitor changes over time, and evaluate the effectiveness of mitigation measures [11]. In mineral exploration, they identify areas with economically valuable minerals by creating geochemical maps that highlight anomalous zones [12]. Overall, regional geochemical surveys provide essential information for understanding the environment, identifying mineral deposits, refining geological models, and making informed decisions about land use. These surveys also contribute to geologic mapping efforts and aid in land use planning and resource management. Geochemical anomalies have become more significant in the process of regional mineral prediction in recent decades. Geochemical element distribution patterns at the surface are the result of several primary and secondary geological processes [13,14].

In recent decades, there have been significant advancements in the processing of geochemical data, revolutionizing our understanding of Earth’s processes, environmental changes, and resource assessments. This process involves several crucial steps, including data quality control, normalization, transformation, statistical analysis, visualization, geostatistics, data integration, and computational modeling. Through these methods, data reliability is ensured, enabling effective comparison, interpretation, exploration of relationships and trends, visualization of results, analysis of spatial patterns, integration of data from multiple sources, and simulation of geochemical processes. Researchers have extensively studied state-of-the-art methods for processing geochemical exploration data over the past several decades [15]. The field has witnessed the introduction of various advanced mathematical and statistical techniques to analyze geochemical data exploration [16,17]. These methods have transitioned from traditional and multivariate statistics to more sophisticated approaches like fractal/multifractal models [18,19]and machine learning/deep learning [7]. Geochemistry relies on conventional statistical methods to decipher and interpret the properties of geochemical elements. Traditional statistics, such as classical statistics, have been instrumental in identifying geochemical anomalies using various methods [20,21,22,23,24]. Several successful traditional techniques have been applied to analyze geochemical data. While traditional methods, including multivariate statistics, are commonly employed to define thresholds and distinguish subpopulations [23,25], they often overlook the spatial information and autocorrelation structure present in geochemical data. These methods can be seen as non-spatial statistical tools. To address this limitation, spatial statistical methods, such as the moving average technique, have been developed to incorporate the spatial aspect of geochemical data analysis.

Univariate statistics have become standard in understanding the distribution attributes of geochemical elements, assisting in pinpointing pathfinder elements, and interpreting their dynamics. Meanwhile, spatial data analysis methods, such as geostatistics [26] and fractal analysis [27], have gained popularity as frequency-space-based approaches. These methods, often augmented with Geographic Information Systems (GIS), facilitate the modeling and quantification of spatial discrepancies in geochemical data. Important concepts in non-linear and complexity sciences, fractal and multifractal have found recognition in geosciences [18,26,27,28,29,30] and developed for analyzing geochemical data. These models serve as potent tools in identifying geochemical anomalies and setting geochemical baselines in diverse studies such as concentration-area (C-A) [31,32], spectrum-area model (S-A) [33,34,35], singularity analysis [3,36,37,38].

In this study, the utilization of classical statistics and multifractal properties in analyzing stream sediment geochemistry data for the region. The aim is to distinguish local anomalies from the complex geological background and summarized them to generate a geochemical map. The map provides insights into the concentration values of deposit pathfinder elements, which have multiple chemical element populations and complex spatial dispersion patterns. The primary focus lies in applying advanced GIS techniques [19,39], specifically utilizing the GeoDAS and ArcGIS platforms, to delineate promising areas with anomalies of pathfinder elements [40,41,42,43,44,45,46,47] associated with gold (As, Cu, and Zn) in eastern Thailand.

2. Geological Settings

The study area is located in the eastern region of Thailand. As illustrated in Figure 1, eastern Thailand has a varied topography, with numerous low-relief areas interrupted by linear hills and ridges. Cenozoic strike-slip faults have influenced these landforms to trend NW-SE [48]. The area is characterized by a series of sedimentary basins, volcanic rocks, and granitic intrusions that record a complex tectonic history [49] involving the Sibumasu terrane, Chanthaburi terrane, and Indochina terrane [50,51,52]. During the Permo-Triassic period, a volcanic arc system [53] existed in the Chanthaburi terrane, which is characterized by its location, sandwiched between the Indochina and Sibumasu continental blocks.

The Chanthaburi terrane is a distinct fragment of the Sukhothai Arc, as identified by Sone and Metcalfe in 2008. Noted for its I-type granitic plutons and radiometrically linked to the Triassic-Jurassic eras, the terrane is defined by its boundaries with the Indochina and Sibumasu blocks, outlined by the Sa Kaeo Suture [54] and the Klaeng tectonic line respectively. This terrane is home to a diverse range of marine sediments from the Carboniferous and late Permian-Triassic periods, all with rich fossil records, as well as deep-water Triassic sediments and younger Mesozoic continental sediments [55,56,57,58]. Additionally, the Klaeng tectonic line, a significant fault, speaks to a past of tectonic activity and collision between the Sibumasu block and the southern Sukhothai Arc. This fault line exhibits several geological characteristics, including mylonitic gneisses and low-grade metamorphic rocks [59].

The Chanthaburi terrane also includes various sedimentary deposits. Carboniferous sediments found in three major limestone hills exhibit similarities with Early Carboniferous fauna found in other parts of Thailand and Laos. The terrane also houses late Permian-Triassic sediments, particularly those discovered in the Sukpaiwan Formation. These deposits contain a diverse range of fossils that establish a stratigraphic continuity from the late Permian to the Middle Triassic [57,59]. The terrane also contains Triassic sediments from deeper marine environments, as represented by the Pong Nam Ron Formation and the Noen Po Formation, inclusive of cherts from the Late and Middle Triassic [60,61,62]. Furthermore, it hosts younger Mesozoic continental sediments, conspicuous in southeast Thailand, which have been connected to the Jurassic-Cretaceous Khorat Group [63,64,65,66,67].

Base on the granitic belt in Thailand is composed of the Western, Central, and Eastern belts [68], as shown in Figure 2. Eastern Thailand incorporates the Triassic granite found in the Central belt, referred to as the Sibumasu terrane. This type of terrane is classified as S-type granite. Additionally, the Chanthaburi terrane, classified as I-type granite, is known as the Triassic granite in the Eastern belt. Notably, gold deposits are related to the I-type granite found in the Chanthaburi Terrane [60,69] while heavy minerals are associated with the S-type granite [69,70,71].

This area has several significant mineral deposits, which align with the Central metallogenic province [69]. Some base metals including chromium, tungsten, gold, silver, and copper [72], these minerals are primarily found in association with Permo-Triassic volcanic rocks and Triassic granitic intrusions, which are believed to have been emplaced during the collision of the Indian and Eurasian Plates [73]. Also found gemstones and silica sand deposits are present in the east corner of the country, and recent exploration has also shown that gold potential is also high [74]. Finally, in the Gulf of Thailand, numerous Tertiary basins are the source of natural gas and petroleum (Figure 2).

3. Geochemical Data

The total collected samples were 5,376 stream sediment samples (Figure 3) from stream systems in eastern Thailand, with a density of a sample per 5 km². The data were collected under the Department of Mineral Resources in Thailand (DMR), covering 34,380 km² of territory with multi-element analyses of 12 elements, with elements As (arsenic), Ba (barium), Co (cobalt), Cr (chromium), Cu (copper), Fe [75], Mg (magnesium), Mn (manganese), Ni (nickel), Pb (lead), V (vanadium), and Zn (zinc). The samples were collected from active stream sediment [76] consisting of clay to sand-sized particles along a 30-50 m stretch of the stream. The sediment sample was prepared by sieving it to provide 2-3 kg of the 80-mesh fraction for analysis. Large samples were necessary to minimize the “nugget effect,” which refers to small-scale variations in the element concentration within a sample. A multi-element analytical methodology and a unified data quality monitoring the procedure was used to analyze the sediment samples, each sample was digested in a modified aqua regia solution and analyzed by inductively coupled plasma optical emission spectrometry (ICP-OES) analysis [77] for 12 elements with numbers in parentheses denoting detection limits: As (2 ppm), Ba (3 ppm), Co (1 ppm), Cr (1 ppm), Cu (2 ppm), Fe (30 ppm), Mg (50 ppm), Mn (1 ppm), Ni (1 ppm), Pb (4 ppm), V (1 ppm), and Zn (7 ppm).

4. Methodology

4.1. Descriptive Statistics

Various measures are used in statistical analysis, including the maximum value (Max), minimum value (Min), average value, mean value (X), standard deviation (SD), and coefficient of variation (CV). These measures are essential for understanding the behavior of geochemical elements in the study area and comparing datasets. A high SD indicates that the data points are widely spread out from the mean, while a low SD indicates that the data points are tightly clustered around the mean. The SD is useful for comparing the variability of two or more datasets or determining whether a particular data point is statistically significant or unusual relative to the rest of the data [78]. Elements concentrations within these samples are considered because these elements represent the mineralization composition in this region [33].

4.2. The Spectrum-Area (S-A) Multifractal Model

The S-A model was developed by Cheng et al. in 1999, 2012 and is a helpful method for identifying mixed geochemical patterns in complex geological environments. This model separates geochemical anomalies from the background based on distinct anisotropic scaling properties in the frequency domain, which has been a powerful tool to decompose mixed geochemical patterns [31,79]. The S-A model employs spectral analysis. The spatial distribution of orebodies can be better understood using these novel models. To create a new method that can not only distinguish between anomalies and background but also reveal information on the anisotropic scale invariance of geochemical patterns, Cao and Cheng (2012) merged the S-A model and generalized scale invariance. Using a power-law function that describes the relationship between the spectral energy density value (S) and the area in units (A) of the wave number with a threshold above (≥S) with spectral energy density values above S on the power density plane [80].

The equation of the S-A model:

A (≥S) α S^-2d/β

(1)

Where S is the spectral energy density value, A (≥S) is the area with a threshold larger than or equal to S, β is an anisotropic scaling exponent, d is a parameter representing the degree of overall concentration, and ∝ denotes proportionality.

The following are the steps for applying the S-A model: (1) To turn the gridded data into the frequency domain, use the fast Fourier transformation. (2) Establish the power spectrum of the data. (3) Compute A(S) to find the region with a threshold greater than or equal to S. (4) Make an S-A log-log plot. (5) Before creating the filter, choose the S threshold value (breakpoints in the log-log plot). (6) An inverse fast Fourier technique is used to transform the resulting power spectra. In step 5, there can be two filters. The low- and high-pass filters result in the creation of two maps, the first of which is the anomaly map and the second of which is the background map.

5. Result and Discussion

5.1. Descriptive Statistics

Descriptive statistics of pathfinder elements for gold deposits (As, Cu, and Zn) are shown in Table 1. High coefficients of variation (CV) values for As, Cu, and Zn concentration (183.61%, 159.75%, and 302.55% respectively) indicate the extremely variable concentrations of these elements. The CV values were used for the description of global variability. These statistical findings suggest that eastern Thailand comprises a variety of geochemical environments and has undergone a variety of different geological events over geological time that resulted in complex oreforming processes. Some of the values, particularly the incredibly high values, might fit fractal or multifractal distributions. Afterward, based on SPSS and using the fractal distribution histograms and a quantile-quantile (Q-Q) plot for visualizing the distribution of geochemical data. The distribution histograms (log 10) 4 elements of As, Cu, and Zn have high skewness values, indicating that these data are not normally distributed and contain outliers and show clearly positively skewed distributions indicating that the majority of the data points are concentrated toward the low of the distribution. The skewness of As, Cu, and Zn are 16.74, 31.25, and 6.86 respectively. Based on their Q-Q plots indicate that were not normally distributed [2] because the point dots depart from the normal straight lines (Figure 4).

5.2. Spatial Distribution of As, Cu, and Zn

Anomaly areas with high stream sediment As concentrations (>24.00 ppm, 95th percentile), represented by red shading, are delineated in Figure 5. In total, anomalous can be divided into three zones in NW-SE orientations. The first zone (I) is located in the western of the study area part of the Sibumasu Terrane underlain mostly by Quaternary sediment (Q) and is associated with Granitic rocks (Trgr) and the sedimentary rocks Carboniferous-Permian (CP). The second zone (II) is located in the middle of the study area part of Chanthaburi Terrane and is associated with the granitic rocks (Trgr) intrusion in the sedimentary rocks Permian (PE), Upper Carboniferous (C2), Permian-Triassic mixed marine (PTr), Triassic Noen Phuyaiyeu formation (Trn), and Jurassic-Cretaceous Khorat group (JK). The third zone (III) represents the eastern part of the study area including the Chanthaburi Terrane and Indochina Terrane associated with the Granitic rocks (Trgr), Permian-Triassic volcanic rock (PTrv), and the sedimentary rocks Early-Middle Permian limestone Sa Kaeo Suture melange (Ps-2).

High stream sediment anomaly zones In Figure 5, Cu concentrations (>50.50 ppm, 95th percentile) are indicated by red shading. Anomaly can be split into three zones overall in NW-SE orientations. The first zone (I) is located in the western part of Sibumasu Terrane and is associated with Granitic rocks (Trgr) and the sedimentary rocks Carboniferous-Permian (CP). The second zone (II) is located in the middle of the study area part of Chanthaburi Terrane and is associated with the sedimentary rocks Permian (PE) The third zone (III) is located in the eastern of the study area part of the Indochina Terrane associated with the Intrusive rocks Permian volcanic [81].

Anomaly regions with high stream sediment Zn (Figure 5) contents (>100 ppm, 95th percentile) are indicated by red stippling. In total, anomalous can be divided into one zones (I) in NW-SE orientations is located in the western part of Sibumasu Terrane and is associated with Granitic rocks (Trgr) and the sedimentary rocks Permian (PE), and Carboniferous-Permian (CP). The second zone (II) is located in the middle of the study area part of Chanthaburi Terrane underlain mostly by Quaternary sediment (Q).

The elements were distributed statistically across the geological units, including Quaternary (Q), Cretaceous Sau Khua formation (Ksk), Jurassic-Cretaceous Khorat group (JK), Jurassic-Cretaceous Lam Thup formation (JKl), Jurassic-Cretaceous Phra Wihan formation (JKpw), Jurassic Phu Kradung formation (Jpk), Triassic Pong Nam Ron formation (Trpn), Triassic Noen Phuyaiyeu formation (Trn), Permian-Triassic mixed marine (PTr), Early-Middle Permian limestone Sa Kaeo Suture melange (Ps-2), Early-Middle Permian limestone (Ps-1), Early-Middle Permian Saraburi group [54], Carboniferous-Permian (CP), Upper Carboniferous (C2), Lower Carboniferous (C1), Devonian- Carboniferous (DC), Silurian-Devonian (SD), Permian (PE), Basalt deposits (Qbs), Granitic rock (Trgr), Permian-Triassic volcanic (PTru, PTrv), and Permian volcanic [81]. Table 2 summarizes these units, and their distribution characteristics were analyzed to determine the distribution of elements.

The average value (X) of As element in Upper Carboniferous (C2), Permian-Triassic mixed marine (PTr), Early-Middle Permian limestone Sa Kaeo Suture melange (Ps-2) (22.33, 19.74, and 19.74 respectively) are higher than the average value of the whole unit but the coefficient (Cv) are low showing weak activity and strong stability. The coefficient of variation (Cv) in Permian (PE), Jurassic Phu Kradung formation (Jpk), and Early-Middle Permian limestone (Ps-1) are 1.84, 1.82, and 1.78 respectively and the Quaternary (Q), in the eastern part is the highest value of 2.15, which shows that the active migration ability of As element is strong.

Average value (X) of Cu element in Permian volcanic [81], Permian-Triassic volcanic (PTru, PTrv), Early-Middle Permian limestone Sa Kaeo Suture melange (Ps-2), and Devonian- Carboniferous (DC), (108.50, 41.54, 31.12 and 30.39 respectively) are higher than the average value of the whole unit and the coefficient (Cv) of are low and Permian volcanic [81] is lowest showing very weak activity and strong stability. The coefficient of variation (Cv) in Permian (PE) and Silurian-Devonian (SD), are 1.60 and 1.12 respectively and the Quaternary (Q), in the eastern part is the highest value of 2.63, which shows that the active migration ability of Cu element is strong.

Zn element’s average value (X) in Permian volcanic [81], Basalt deposits (Qbs), and the Quaternary (Q) (69.83, 61.04, and 57.38 respectively) are higher than the average value of the whole unit and the coefficient (Cv) of are low and Permian volcanic [81] is lowest showing very weak activity and strong stability. The coefficient of variation (Cv) in the Quaternary (Q) and Permian (PE) are 3.60 and 3.03 respectively and which shows that the active migration ability of the Zn element is strong.

The elements were distributed statistically across the geological units, including Quaternary (Q), Cretaceous Sau Khua formation (Ksk), Jurassic-Cretaceous Khorat group (JK), Jurassic-Cretaceous Lam Thup formation (JKl), Jurassic-Cretaceous Phra Wihan formation (JKpw), Jurassic Phu Kradung formation (Jpk), Triassic Pong Nam Ron formation (Trpn), Triassic Noen Phuyaiyeu formation (Trn), Permian-Triassic mixed marine (PTr), Early-Middle Permian limestone Sa Kaeo Suture melange (Ps-2), Early-Middle Permian limestone (Ps-1), Early-Middle Permian Saraburi group [54], Carboniferous-Permian (CP), Upper Carboniferous (C2), Lower Carboniferous (C1), Devonian- Carboniferous (DC), Silurian-Devonian (SD), Permian (PE), Basalt deposits (Qbs), Granitic rock (Trgr), Permian-Triassic volcanic (PTru, PTrv), and Permian volcanic [81] [81]. Table 2 summarizes these units, and their distribution characteristics were analyzed to determine the distribution of elements.

The average value (X) of As element in Upper Carboniferous (C2), Permian-Triassic mixed marine (PTr), Early-Middle Permian limestone Sa Kaeo Suture melange (Ps-2) (22.33, 19.74, and 19.74 respectively) are higher than the average value of the whole unit but the coefficient (Cv) are low showing weak activity and strong stability. The coefficient of variation (Cv) in Permian (PE), Jurassic Phu Kradung formation (Jpk), and Early-Middle Permian limestone (Ps-1) are 1.84, 1.82, and 1.78 respectively and the Quaternary (Q), in the eastern part is the highest value of 2.15, which shows that the active migration ability of As element is strong.

Average value (X) of Cu element in Permian volcanic [81], Permian-Triassic volcanic (PTru, PTrv), Early-Middle Permian limestone Sa Kaeo Suture melange (Ps-2), and Devonian- Carboniferous (DC), (108.50, 41.54, 31.12 and 30.39 respectively) are higher than the average value of the whole unit and the coefficient (Cv) of are low and Permian volcanic [81] is lowest showing very weak activity and strong stability. The coefficient of variation (Cv) in Permian (PE) and Silurian-Devonian (SD), are 1.60 and 1.12 respectively and the Quaternary (Q), in the eastern part is the highest value of 2.63, which shows that the active migration ability of Cu element is strong.

Zn element’s average value (X) in Permian volcanic [81], Basalt deposits (Qbs), and the Quaternary (Q) (69.83, 61.04, and 57.38 respectively) are higher than the average value of the whole unit and the coefficient (Cv) of are low and Permian volcanic [81] is lowest showing very weak activity and strong stability. The coefficient of variation (Cv) in the Quaternary (Q) and Permian (PE) are 3.60 and 3.03 respectively and which shows that the active migration ability of the Zn element is strong.

5.3. Determining Thresholds Using the S-A Model

The decomposed geochemical maps were produced by IDW using the S-A method, which was applied to each raster map: anomalies and backgrounds of the distributions for As, Cu, and Zn were built up using the inverse distance weighting (IDW) interpolation method [82] by ArcGIS software (Figure 6). One of the most widely used moving-average interpolation methods is IDW, which works under the concept that within observations’ values have more effect on interpolated values than distant observations’ values. The IDW approach has the benefit of being easy to use and understand, and it is also simple to apply. The main issue is from the weights being determined solely based on location and ignoring the variance of the results [83]. The failure to take into account the local features of data is a typical flaw in moving average interpolation techniques like IDW. Then, the power spectrum densities (S) were calculated for the elemental distributions. The S-A multifractal model was used to decompose complex geochemical patterns into constituent anomalous and background patterns [82] and was implemented with the aid of GeoDAS software [34]. The power spectrum density (S) and the area (A) that is greater than or equal to S could be fitted with two straight lines using the least square method.

The S-A plot consisting of S versus the number of cells greater than or equal to S could be fitted by three straight lines using the least square method. The left lines of the cutoff values represented the geochemical noise, anomaly, and background, respectively. Then these three components were converted to the spatial domain by an inverse Fourier transformation. The decomposed maps show the spatial distribution of the geochemical anomaly and the geochemical background of As, Cu, and Zn.

Using the S-A approach, the geochemical map of As concentration (Figure 6) was divided into two spatial components. The As geochemical map was converted into the frequency domain using a Fourier transform. The relationship between A (≥S) and S is shown on an As log-log scale (Figure 6a). The data can be fitted with three straight-line segments using the Least Squares [84] method. These segments represent three power energy spectrum S ranges, each distinguished by a unique scaling property of the S-A relation. There are three ranges of S: left (S ≤ 4,138), middle (4,138 < S ≤ 20,606), and right (S ≥ 20,606). The last segment displays a short-range spectrum of high frequencies. Two filters were defined using two distinct ranges of S separated by the threshold value S = 20,606: an anomaly filter with S < 20,606 (Figure 6b) and a background filter with S > 20,606 (Figure 6c). These filters are irregularly shaped to preserve the anisotropy and interior structure of the two-dimensional power energy spectrum, which they represent. The two maps are the result of applying the two filters obtained from the S-A plot to the Fourier-transformed functions and then converting them back into the spatial domain.

The geochemical map of Cu concentration (Figure 7) was decomposed into two spatial components using the S-A method. The Cu geochemical map was converted into the frequency domain using a Fourier transform. The relationship between A (≥S) and S is shown on an As log-log scale (Figure 7a). There are three ranges of S: left (S ≤ 3,488), middle (3,488 < S ≤ 114,777), and right (S ≥ 114,777). The last segment displays a short-range spectrum of high frequencies. Two filters were defined using two distinct ranges of S separated by the threshold value S = 114,777: an anomaly filter with S < 114,777 (Figure 7b) and a background filter with S > 114,777 (Figure 7c).

The S-A approach was used to divide the geochemical map of Zn concentration (Figure 8) into two spatial components. Using a Fourier transform, the Zn geochemical map was transformed into the frequency domain. The relationship between A (≥S) and S is shown on an As log-log scale (Figure 8a). There are three ranges of S: left (S ≤ 12,889), middle (12,889 < S ≤ 850,339), and right (S ≥ 850,339). The last segment displays a short-range spectrum of high frequencies. Two filters were defined using two distinct ranges of S separated by the threshold value S = 850,339: an anomaly filter with S < 850,339 (Figure 8b) and a background filter with S > 850,339 (Figure 8c).

The decomposed pattern obtained from the concentration map of arsenic (As) using the background filter reveals variations in As concentration values in the background, which could indicate favorable geological conditions for the occurrence of gold mineralization. Overlaying the known gold deposits on a background map shows that these deposits are located within areas characterized by high background As values found around intrusive and sedimentary rocks in the central (zone II), and eastern parts (zone III). The anomaly map suggests high mineral deposit potential in areas with NW-SE trending faults and mapped intrusions [85]. The most of As anomaly map shows that high values follow around intrusive rocks in the middle zone Chanthaburi terrane and along NW-SE trending fault/fractures, which are favorable areas for mineral deposits. These results of the As element anomaly assessment of pathfinder elements for gold deposits may indicate the potential relationship to the gold deposits in the study area. The integrated anomalies of As occur around the intrusions and in the vicinity of faults in the center. However, the As element anomalies occurring in the western part of the study area are not relative to gold deposits and may be caused by the high background concentration values of the intrusive rock of the Sibumasu terrain, which is divided by the tectonic line NW-SE trend, As element anomaly assessment and pathfinder elements for gold deposits support this potential [86,87], further investigation of gold anomalies around intrusions and faults is recommended.

Similar maps with distinctive features were generated using copper (Cu) and zinc (Zn) data. These two maps are also displayed in Figure 7 and Figure 8, depicting the NE-SW distribution characteristics of the area. Interestingly, the high concentrations observed in the geochemical maps of the study area most of the anomalies do not align with the presence of the known gold deposits in the western part (zone I). Instead, they are primarily distributed in the western part around the intrusive rocks and sedimentary rocks of the Sibumasu terrane. The central (zone II) shows that the anomaly high value follows around the Permian rocks (PE).

6. Conclusions

1. Stream sediment geochemical data analysis provides a robust means to identify potential mineral deposits across a large region. Traditional statistical methods may fail to account for spatial variability in complex geological settings, but incorporating the S-A multifractal model helps to better interpret the data. This approach provides an enhanced understanding of the geological processes, furthering our ability to pinpoint exploration targets. Through statistical analysis, the study uncovers a distinct distribution of elements such as As, Cu, and Zn across various geological units. These findings offer crucial insight into the dynamic interactions of elements within different geological contexts, revealing the migration abilities of these elements in specific units. Overall, the study demonstrates that the S-A multifractal model provides a powerful tool for analyzing stream sediment geochemical data. This approach helps to identify specific geological processes that contribute to patterns in the data, improving the understanding of statistical results and the regularity of changes in these properties in spatial and temporal domains. This information proves valuable in identifying precise exploration targets for mineral deposits.

2. By applying the S-A multifractal model and the Inverse Distance Weighting (IDW) method, the research has successfully decomposed geochemical maps into anomalous and background patterns. The anomaly map for the As element, in particular, has shown a strong correlation with the gold deposits in the study area, indicating its potential as a reliable indicator for identifying gold deposits. While geochemical maps derived from stream sediment data offer an initial understanding of element distribution, it is important to exercise caution during interpretation due to potential influences from regional geological factors.

3. The study has created prediction maps, pinpointing areas of interest associated with pathfinder elements within the gold deposit zones and beyond. These maps can guide mineral exploration by narrowing down target areas for further investigation. It underlines the importance of utilizing a variety of analytical methods for accurate results, acknowledging that regional geological factors can contribute to elevated values unrelated to mineralization.