3D Geological Modeling and Characterization of Coalbed Gas Content in the Jiulongchuan Exploration Area

Abstract

Coalbed methane (CBM) is an important unconventional natural gas resource, and coal seam gas content is a key parameter for CBM resource evaluation and favorable-zone prediction. Taking the Jiulongchuan exploration area in Gansu Province as the study area, this study integrated drilling, well-logging, and measured gas content data to establish a multivariate regression model for coal seam gas content prediction. On this basis, three-dimensional geological modeling and variogram analysis were applied to characterize the spatial distribution of gas content in the main mineable coal seams (Nos. 5, 6, and 8). The results indicate that the regression model constructed using acoustic transit time, natural gamma-ray values, density logging parameters, and burial depth shows generally reasonable predictive capability for coal seam gas content. Cross-validation results suggest that the predicted gas contents are generally consistent with measured values. Spatial modeling results show that gas content in Seam No. 8 is generally higher than that in Seams No. 5 and No. 6, and gas content tends to increase with burial depth and coal seam thickness. In addition, relatively high gas contents are commonly observed along synclinal zones, whereas lower values occur near anticlinal areas. The integrated application of well-log interpretation and three-dimensional geological modeling provides a reasonable characterization of the spatial variation in coal seam gas content in the study area. The results may provide useful references for CBM resource evaluation and favorable-zone prediction in the Jiulongchuan exploration area.

1. Introduction

Coalbed methane (CBM) is an unconventional natural gas that occurs predominantly in an adsorbed state within coal seams and has emerged as a promising clean energy resource [1,2,3]. Accurate evaluation of CBM resources, prediction of development potential, and formulation of extraction strategies all rely on coalbed gas content (CBGS), which remains a critical parameter [4,5]. CBGS can be determined through either direct measurement or indirect prediction methods. Data for indirect prediction are derived from core samples, well logging, and seismic surveys. Methods for predicting gas content based on core and logging data include multivariate regression analysis [6,7,8], as well as pattern recognition techniques, such as gray system theory [9,10], neural networks [4,11], and support vector machines [12]. Isothermal adsorption experiments on coal cores and multivariate regression analysis of well-log data are widely regarded as the most effective and practical methods for predicting coalbed gas content (CBGS) [13,14,15,16]. The principal factors influencing CBGS include coal rank, burial depth, coal quality and macrolithotype, effective burial depth, seam thickness, petrographic composition, seam structure, temperature, pressure, tectonic activity, magmatic activity, roof and floor lithology, and groundwater dynamics [17,18,19,20,21]. These factors exhibit significant variability across different regions and geological periods.

Stochastic reservoir modeling, as an advanced method for reservoir characterization, is fundamentally based on the use of available data to generate multiple equiprobable, high-resolution spatial distribution models of reservoir properties [22,23,24,25,26]. During the construction of three-dimensional (3D) geological models, the development and analysis of variograms are critical for ensuring model accuracy [27,28,29]. Variograms quantitatively characterize the spatial correlation of regionalized variables, following the principle that closely spaced samples exhibit strong correlation, whereas widely separated samples show weaker correlation. This spatial correlation is typically anisotropic, necessitating variogram characterization of properties in different directions. By deriving variograms from input data and incorporating them into property models, the spatial correlation observed in experimental data can be effectively reproduced in the final model. The construction of 3D geological models not only enables detailed characterization of reservoir properties and heterogeneity but also plays a significant role in coalbed methane reservoir evaluation, particularly in gas content assessment [30,31].

The Jiulongchuan exploration area, located in Qingyang, Gansu Province, has been extensively studied and is characterized by abundant coal resources. The coal seams in this region occur at relatively great depths, with an average burial depth of approximately 1200 m [32]. Although local coalbed methane (CBM) enrichment has been identified in the Longdong area, indicating considerable exploration potential [33], detailed investigations of deep CBM systems remain limited. In particular, systematic studies on the logging-based interpretation of key parameters, such as gas content, and their controlling factors are still lacking. Three-dimensional (3D) geological modeling plays a critical role in characterizing the heterogeneity of gas content. This study provides a refined evaluation of gas content in the Jiulongchuan area through an integrated approach that combines 3D geological modeling, multivariate regression analysis of logging data, and detailed variogram analysis. Special emphasis is placed on the influence of geological structures, burial depth, and coal seam thickness on gas content. The results provide additional information for the characterization of deep CBM reservoirs in the Jiulongchuan exploration area.

2. Geological Setting

The Jiulongchuan exploration area is located at the southwestern terminus of the northern Shaanxi monocline within the Ordos Basin, which is adjacent to the Tianhuan Depression. Stratigraphically, it belongs to the Ordos subregion of the Shanxi–Hebei–Shandong–Henan region within the North China stratigraphic superregion. The strata form a compound monocline that gently dips northwestward, with dip angles typically ranging from 3° to 10°. Faulting is weakly developed in the area; however, several gentle folds are well developed (Figure 1).

The Middle Jurassic Yan’an Formation is the only coal-bearing stratum in the Jiulongchuan exploration area, occurring at depths of 1000–1500 m and exhibiting significant thickness variation (46.37–208.16 m; average 116.79 m). The Yan’an Formation contains eight coal seams with a total average thickness of 14.59 m, corresponding to a coal-bearing coefficient of 21.2%. Among these, the No. 5, No. 6, and No. 8 coal seams are minable or locally minable.

The Yan’an Formation is subdivided into three members, from base to top. The lower member (J₂y¹) has limited coal-bearing significance, with the No. 8 seam occurring at its top. The middle member (J₂y²) constitutes the principal coal-bearing interval, containing the No. 5 and No. 6 seams in its upper part. The upper member (J₂y³) is also of limited coal-bearing importance and contains the unminable No. 2 seam in its middle to upper part. The characteristics of the minable and locally minable seams in the Jiulongchuan exploration area are summarized in

Table 1. Characteristics of the main minable coal seams in the Jiulongchuan exploration area.

Coal Seam	Thickness/m	Average Thickness/m	Minable Thickness/m	Average Minable Thickness/m	Recovery Rate/%	Coal Seam Structure
No. 5	0.3–4.98	2.55	0.80–4.98	2.58	94.44	simple-relatively simple
No. 6	0.23–4.03	1.67	0.92–3.47	1.67	94.95	simple-relatively simple
No. 8	0.19–10.09	4.33	0.80–9.07	3	27.27	simple-complex

.

As shown in Table 1, the recovery rates of the No. 5 and No. 6 coal seams both exceed 94%, whereas the recovery rate of the No. 8 coal seam is only approximately 27%. The low recovery rate of the No. 8 coal seam is mainly attributed to the widespread development of one to two layers of gangue (mudstone/carbonaceous mudstone) within the seam. Nevertheless, the No. 8 coal seam is characterized by a large total thickness and good continuity, and its resource reserves account for more than 60% of the total CBM geological resources in the study area. Therefore, it was still selected as the primary target for 3D gas content modeling in this study.

3. Prediction of Coalbed Gas Content Based on Multivariate Regression of Well-Logging Data

The well logs in the Jiulongchuan exploration area include multiple parameters, such as long-spaced gamma–gamma (GGFR), short-spaced gamma–gamma (GGNR), natural gamma ray (GR), Laterolog 3 (LL3), spontaneous potential (SP), caliper (CAL), sonic transit time (SAT), formation temperature (TEMP), and compensated density (DEN). Following field calibration, the GGFR and GGNR curves are combined to generate a compensated density log.

To establish an optimal multivariate regression model for gas content prediction, cross-plot analyses were conducted using gas content and well-log data from coal seams in five CBM parameter wells (Figure 2). Detailed gas content test information is presented in

Table 2. Comparison between measured and predicted gas contents.

Well	Number	Depth/m	Measured Gas Content/cm3/g	Predicted Gas Content/cm3/g
NZ306	1	1137.54–1137.84	0.33	0.28
	2	1146.56–1146.86	0.14	0.11
	3	1219.08–1219.38	0.58	0.53
NZ311	1	1154.05–1154.30	0.39	0.53
	2	1159.62–1159.87	0.13	0.16
	3	1209.70–1209.95	4.98	5.98
NZ415	1	1143.51–1143.77	0.37	0.43
	2	1151.86–1152.06	0.09	0.06
	3	1191.09–1191.39	4.52	5.23
NZ614	1	1169.34–1169.64	0.34	0.27
	2	1199.81–1200.11	1.06	1.17
	3	1241.10–1241.40	5.42	5.21
NZ619	1	1097.81–1098.09	0.58	0.48
	2	1133.57–1133.87	1.15	0.97
	3	1157.20–1157.52	0.66	0.72

. The results indicate that the correlations between gas content and individual logging parameters are weak. In general, coalbed methane content increases with burial depth. Moreover, increasing gas content reduces the acoustic velocity in coal seams, resulting in higher sonic transit time values, indicating a positive correlation. In contrast, increasing ash content reduces the organic carbon content, thereby diminishing the hydrocarbon generation and preservation capacity of coal seams. Ash content shows strong positive correlations with both natural gamma-ray and density log values. Based on these observations, a composite parameter P was constructed (Equation (1)) to better characterize coalbed methane content.

P = SAT/(GR × DEN)

(1)

Regression analysis was performed using the constructed composite parameter P and coalbed gas content (CBGS) (Figure 3a), showing a relatively good correlation between the two variables.

Considering the strong relationship between gas content and burial depth, a multivariate regression approach was adopted by integrating the parameter P with burial depth to establish a gas content prediction model based on logging data and burial depth (Equation (2)), with a coefficient of determination (R²) of 0.98 (Figure 3b).

Vgas = −65.9157 + 0.0339P + 0.056Depth

(2)

To evaluate the multicollinearity between burial depth and the composite parameter (P), the variance inflation factor (VIF) was calculated. Taking P as the dependent variable and burial depth as the independent variable, a univariate linear regression was performed, yielding an R² value of 0.6956. Accordingly, the VIF was calculated as VIF = 1/(1 − 0.6956) = 3.29. The VIF values for both P and burial depth are 3.29, which are lower than the critical threshold of five and far below the stricter threshold of 10. This indicates that although a moderate correlation exists between the two independent variables ((Pearson r = 0.83)), no severe multicollinearity is present. The standard errors of the regression coefficients did not exhibit excessive inflation, and the explanatory effects of each variable on gas content can still be robustly estimated. Therefore, the regression equation with (R² = 0.9483) reflects the genuine combined explanatory power of P and burial depth on gas content, rather than an artificial effect caused by multicollinearity. The parameter P provides gas content information from the perspectives of logging elastic response and lithology correction, whereas burial depth serves as a comprehensive proxy variable for reservoir pressure, temperature, and preservation conditions. Together, they contribute complementary predictive information. Therefore, this formula can effectively reflect the gas content characteristics of the study area.

To verify the reliability of the statistical model for predicting coal seam gas content, a cross-validation approach based on the comparison between predicted and measured values was adopted in this study to evaluate the prediction accuracy using sample data from five CBM parameter wells in the Jiulongchuan exploration area. The measured gas content of each well was compared with the gas content predicted by the statistical model. The model’s prediction accuracy was evaluated using the mean absolute error (MAE), root mean square error (RMSE), mean absolute percentage error (MAPE), and coefficient of determination (R²). The corresponding calculation formulas are as follows (Equations (3)–(5)):

$$MAE={\displaystyle \frac{1}{n}}{\displaystyle {\sum }_{i=1}^n}\left|y_i-{\hat{y}}_i\right|$$

(3)

$$RMSE=\sqrt{{\displaystyle \frac{1}{n}}{\displaystyle \sum _{i=1}^n}{\left(y_i-{\hat{y}}_i\right)}^2}$$

(4)

$$MAPE={\displaystyle \frac{1}{n}}{\displaystyle \sum _{i=1}^n}\left|{\displaystyle \frac{y_i-{\hat{y}}_i}{y_i}}\right|\times 100\%$$

(5)

The results indicate that the predicted gas contents are generally consistent with the measured values. The mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE) are 0.19 cm³/g, 0.26 cm³/g, and 18.74%, respectively(

Table 3. Cross-validation results between statistical model predictions and measured gas contents.

Well	MAE/cm3/g	RMSE/cm3/g	MAPE/%
NZ306	0.043	0.044	18.19
NZ311	0.39	0.583	20.63
NZ415	0.267	0.412	25.84
NZ614	0.13	0.143	13.12
NZ619	0.113	0.124	15.91

). The coefficient of determination (R²) and correlation coefficient are 0.973 and 0.992, respectively. Validation results for individual wells show that the prediction errors are generally within an acceptable range, suggesting that the established statistical model can reasonably reflect the variation characteristics of coal seam gas content in the study area and can provide support for subsequent three-dimensional gas content modeling and favorable area prediction.

4. 3D Geological Modeling of CBGS

Detailed static analyses of geological structure, stratigraphy, and CBM reservoirs were integrated with refined logging interpretations and implemented in Petrel^® using stochastic modeling to construct a three-dimensional (3D) property model of the coal-bearing strata in the Jiulongchuan exploration area, enabling a quantitative description of the spatial distribution of CBM occurrence.

The 3D geological modeling workflow comprises three key stages: grid generation, property modeling, and model refinement. The horizontal grid cell size was set to 100 m × 100 m, and the formation was vertically discretized into 70 layers, resulting in a total of 121 × 115 × 70 = 974,050 grid cells. Property modeling involves geometric model construction, well-log discretization, integrated data analysis, and petrophysical property modeling, among which the latter two are particularly critical.

4.1. Variogram Analysis of CBGS

Data analysis aims to optimize the fit between variogram models and the spatial characteristics of the data by adjusting variograms in the major, minor, and vertical directions. This process provides a fundamental basis for subsequent stochastic modeling. In stochastic reservoir modeling, the construction of spatial variograms is essential for characterizing reservoir heterogeneity. This involves the selection of variogram models (i.e., transition models) and key parameters, including the nugget, sill, and range, which directly influence simulation results and CBM development potential.

The anisotropy of three-dimensional (3D) space requires the definition of variogram models and parameters in each of the three principal directions. The major and minor directions should be orthogonal within the horizontal plane, while the variogram model type and nugget value are typically kept consistent across directions; in contrast, the range varies with direction. In this study, a spherical variogram model was adopted, with the nugget set to zero to enhance the representation of spatial correlation among data points. The spherical variogram model was selected because it can effectively characterize the finite spatial correlation range and moderate continuity of coal seam gas content in the study area, which is consistent with the variation pattern of the experimental variograms. The nugget value was set to zero because the data used in this study were derived from calibrated well-log interpretation and measured gas content data with relatively limited random noise. In addition, the spatial variation in coal seam gas content exhibits relatively continuous geological characteristics at the modeling scale.

The variograms of gas content for the No. 5 coal seam in the major, minor, and vertical directions are presented in Figure 4.

The variogram analysis of Seam No. 5 shows that the sill values in both the major and minor directions are close to one, indicating evident spatial variability of gas content in the horizontal directions. The vertical sill value is 0.4, suggesting a relatively smaller fluctuation amplitude in the vertical direction. The ranges in the major, minor, and vertical directions are 3097 m, 2309 m, and 1.8 m, respectively, indicating strong spatial continuity in the horizontal directions but relatively limited vertical correlation. It should be noted that the vertical range mainly reflects the effective spatial correlation scale of gas content samples rather than the actual coal seam thickness itself. Owing to the limited vertical sampling interval and internal heterogeneity within the coal seam, the fitted vertical range is smaller than the seam thickness. Overall, gas content in the study area exhibits stronger lateral continuity and relatively faster variation in the vertical direction.

The variograms of gas content for the No. 6 coal seam in the major, minor, and vertical directions are presented in Figure 5.

The variogram analysis of Seam No. 6 indicates that the sill values in both the major and minor directions are close to one, suggesting evident spatial variability of gas content in the horizontal directions. The vertical sill value is 0.4, indicating a relatively smaller fluctuation amplitude in the vertical direction. The ranges in the major, minor, and vertical directions are 1382 m, 1370 m, and 1.6 m, respectively, indicating relatively good spatial continuity in the horizontal directions but limited vertical correlation. Similar to Seam No. 5, the relatively small vertical range of Seam No. 6 suggests limited vertical spatial correlation. Overall, the gas content of Seam No. 6 shows relatively stable variation in the horizontal plane, with similar continuity in the major and minor directions, while the vertical spatial correlation remains relatively limited.

The variograms of gas content for the No. 8 coal seam in the major, minor, and vertical directions are presented in Figure 6.

The variogram analysis of Seam No. 8 indicates that the sill values in both the major and minor directions are close to one, suggesting evident spatial variability of gas content in the horizontal directions. The vertical sill value is 0.4, indicating a relatively smaller fluctuation amplitude in the vertical direction. The ranges in the major, minor, and vertical directions are 2010 m, 2010.7 m, and 1.9 m, respectively, indicating relatively good spatial continuity in the horizontal directions but limited vertical correlation. Seam No. 8 also exhibits relatively stable lateral continuity, whereas relatively rapid local variation is observed in the vertical direction.

4.2. 3D Geological Model of CBGS

Three-dimensional (3D) geological models and gas content cross-sections were constructed using sequential Gaussian simulation (SGS).

Since the gas content data do not conform to a normal distribution (Figure 7a), the original data were first transformed into a normal distribution (Figure 7b). The specific calculation formulas are as follows:

$${\mathrm{P}}_{\mathrm{i}}={\displaystyle \frac{\mathrm{i}-0.5}{\mathrm{n}}}$$

(6)

$${\mathrm{Z}}_{\mathrm{i}}={\mathrm{\varnothing }}^{-1}\left({\mathrm{P}}_{\mathrm{i}}\right)$$

(7)

where i is the ascending order index of the original data; ${\mathrm{P}}_{\mathrm{i}}$ is the cumulative probability; n is the total number of data points; ${\mathrm{\varnothing }}^{-1}$ is the inverse function of the standard normal distribution; and ${\mathrm{Z}}_{\mathrm{i}}$ is the value after normal transformation.

The simulation results are presented in Figure 8, Figure 9 and Figure 10.

The gas content model reveals significant vertical heterogeneity in the No. 8 coal seam, with generally higher gas content compared to other seams. The north–south (S–N) cross-section indicates relatively high gas content in the southern and central parts of the study area, gradually decreasing toward the north. Along the syncline axis, higher gas content is observed in the upper part of the No. 8 coal seam. The west–east (W–E) cross-section shows that gas content in the west-central area is higher than that in the eastern region. In the western area, coalbed methane content decreases in the updip direction, whereas no clear spatial trend is observed in the eastern part. Additionally, along the syncline axis, the upper portion of the No. 8 coal seam exhibits higher gas content than the lower portion. Well cross-section analysis further demonstrates that the No. 8 coal seam is thicker and more laterally continuous than other seams and generally exhibits higher gas content along the syncline axis. In particular, the deeply buried segment of the syncline axis near Well NZ311 shows significantly higher gas content than other areas. In contrast, within the monocline structure, coalbed methane content decreases in the updip direction.

4.3. Gas Content Distribution Model of the No. 8 Coal Seam

For the No. 8 coal seam with relatively high gas content, the gas content attribute map and the 3D spatial distribution model are shown in Figure 11. For better visualization, the vertical (Z-axis) scale in Figure 11 is exaggerated by a factor of 10.

The gas content model of the No. 8 coal seam shows a significant increase in gas content in the vicinity of Well NZ311 and toward the northwest, where burial depths are greater. A high-gas-content zone is also identified in the southwestern part of the study area. Overall, gas content is higher along the syncline axis and lower along the anticline axis. These observations suggest that gas content in the No. 8 coal seam is closely correlated with geological structure, burial depth, and seam thickness.

4.4. Main Controlling Factors of Coal Seam Gas Content

4.4.1. Geological Structure

The coal seam structures in the study area are characterized by a series of short-axis anticline–syncline assemblages distributed in a northeastward direction. The fold amplitudes are relatively gentle, with coal seam dip angles generally ranging from 3° to 8°, and no obvious fault structures have been identified. Therefore, the coal seams largely preserve their primary structural characteristics. The coal-bearing strata, together with their overlying and underlying formations, are mainly composed of interbedded mudstones and medium- to fine-grained clastic rocks, which also generally retain their original sedimentary and structural features without significant fault-related destruction. Such geological conditions are favorable for the preservation of coalbed methane. The three-dimensional geological model indicates that gas content in the synclinal axis is generally higher than that in the anticlinal axis, and gas content gradually decreases toward the uplifted flanks of the syncline. It should be particularly noted that although Well NZ306 has a relatively large burial depth, its coal seam gas content, especially in Seam No. 8, remains relatively low. This is mainly because the well is located near the junction of two anticlines, where tectonic deformation is relatively intense, resulting in the development of fractures that facilitate gas escape and consequently reduce gas preservation capacity. In addition, in the Longdong area, the gas content of Seam No. 6 is generally lower than that of Seam No. 5, which may be related to sedimentary conditions.

Specifically, the major variogram direction (NW–SE) is approximately parallel to the syncline axis in the study area. Along this direction, the coal seam floor undulates relatively gently, and the gradients of burial depth and seam thickness are relatively small, allowing gas content to maintain stronger spatial continuity over a larger distance. This corresponds to the relatively large variogram ranges in the major direction (3097 m for Seam No. 5, 1382 m for Seam No. 6, and 2010 m for Seam No. 8). The minor direction (NE–SW) is orthogonal to the major direction and approximately perpendicular to the syncline axis. Along this direction, the strata cross multiple secondary fold limbs, where burial depth gradients increase significantly. In addition, the development of local fractures and cleats further enhances lateral reservoir heterogeneity, resulting in reduced spatial continuity of gas content and relatively smaller variogram ranges in the minor direction (2309 m for Seam No. 5, 1370 m for Seam No. 6, and 2011 m for Seam No. 8). The relatively small difference between the major and minor ranges of Seam No. 8 may be related to its greater seam thickness, better continuity, and stronger intrinsic stability. In thick coal seams, the anisotropic modification of the spatial gas content structure by tectonic deformation may be partially buffered. In contrast, the major-direction range of Seam No. 5 (3097 m) is much larger than that of Seam No. 6 (1382 m), indicating stronger spatial continuity of gas content along the syncline axis in Seam No. 5, which may be associated with differences in depositional patterns and the intensity of later tectonic modification. These results suggest that the spatial anisotropy of gas content is not a random phenomenon but rather a direct manifestation of directional structural control exerted by the regional tectonic framework on reservoir properties.

4.4.2. Burial Depth of Coal Seams

The coal seams in the study area are deeply buried. The burial depths of Seam No. 5 in the five parameter wells range from 1097.55 to 1155.16 m, while those of Seam No. 6 range from 1132.78 to 1200.43 m, and those of Seam No. 8 range from 1152.10 to 1246.12 m. Overall, coal seam gas content tends to increase with increasing burial depth (Figure 12). According to statistical results from 88 sampling points in Seam No. 8, the average gas content is 0.96 cm³/g, indicating that burial depth alone is not the dominant geological factor controlling gas content. Due to their greater burial depths, Seam No. 8 and Seam No. 8-1 generally exhibit gas contents greater than 3 cm³/g, representing relatively high values. In contrast, the gas contents of Seam No. 2, Seam No. 5, and Seam No. 6 are generally lower, mostly ranging from 0 to 3 cm³/g. From the perspective of individual wells, relatively high gas contents in Seam No. 8 are mainly observed in Wells NZ311, NZ415, and NZ614, whereas the remaining wells generally exhibit lower gas contents.

4.4.3. Coal Rank

The variation in maximum vitrinite reflectance (R_o) with depth in the study area is shown in Figure 13a. As illustrated in the figure, R_o generally increases with increasing depth, indicating that the degree of coal metamorphism increases with burial depth. The relationship between R_o and gas content in the study area is presented in Figure 13b, the cross-plot between gas content and R_o indicates that the correlation between gas content and R_o in the study area is not significant. This may be attributed to the fact that the maximum vitrinite reflectance of coal seams in the study area ranges from 0.69% to 0.75%, corresponding to the second stage of bituminous coal metamorphism, during which the methane-generation capacity remains relatively weak. Therefore, the influence of coal rank on coal seam gas content in the study area is considered to be limited.

4.4.4. Coal Seam Thickness

Relatively speaking, thicker peat layers within coal-bearing sequences generally generate greater amounts of coalbed methane during the coalification process. In the study area, Seam No. 8 is classified as a medium-thick to extra-thick coal seam and therefore generally exhibits relatively high gas content, whereas Seam No. 5-1 and Seam No. 6 are thinner and show relatively lower gas contents. As shown in Figure 14, there is a certain correlation between coal seam thickness and gas content in the study area, with a fitting coefficient of 0.5801. Overall, gas content tends to increase with increasing coal seam thickness. In particular, the relatively thick No. 8 coal seam generally exhibits comparatively high gas content.

These observations suggest that gas content in the No. 8 coal seam is closely correlated with geological structure, burial depth, and seam thickness. The relatively high-gas-content zones are mainly distributed along synclinal areas and deeper burial sections of Seam No. 8, where preservation conditions and coal seam thickness are comparatively favorable. These results may provide useful references for favorable-zone prediction and future drilling deployment in the Jiulongchuan exploration area.

5. Conclusions

(1)

A multivariate regression model based on acoustic transit time, natural gamma-ray values, density logging parameters, and burial depth was established for coal seam gas content prediction. Cross-validation results suggest that the model predictions are generally consistent with measured gas content data.

(2)

Three-dimensional geological modeling and variogram analysis indicate that coal seam gas content exhibits spatial heterogeneity and anisotropy in the study area. The gas content in Seam No. 8 is generally higher than that in Seams No. 5 and No. 6.

(3)

Gas content tends to increase with burial depth and coal seam thickness. Relatively high gas contents are commonly distributed along synclinal zones, whereas lower gas contents are observed near anticlinal areas, suggesting that geological structure and preservation conditions may influence gas accumulation.

This study provides a preliminary spatial modeling and analysis of coalbed methane gas content based on limited drilling, logging, and measured gas content data. Due to the limited well control and dataset size, the current model mainly reflects the overall spatial distribution trend of gas content, while detailed multiple-realization SGS uncertainty analysis and structurally constrained modeling were not further conducted. In addition, reservoir pressure, temperature, and Langmuir adsorption parameters were not incorporated into the model; therefore, the results mainly represent statistical spatial variation characteristics rather than rigorous adsorption thermodynamic processes. Future studies should integrate additional drilling, logging, and adsorption experimental data to further conduct multiple SGS realizations, quantitative uncertainty analysis, and structurally constrained modeling in order to improve the reliability and accuracy of coalbed methane gas content prediction.