3D Hydrogeological Structure Modeling Based on Quantitative Correlation and Identification of Aquifer Types Within Stratigraphic Layers

Abstract

Due to the scarcity and uneven distribution of data, as well as the complexity of geological conditions, high-precision 3D hydrogeological structure modeling, especially at large scales, remains a significant challenge in the field. To address this issue, this study undertakes an in-depth analysis of the correlation between stratigraphic and hydrogeological structures. Utilizing the cumulative thickness of various aquifer types as a criterion, we establish a quantitative correlation model between stratigraphic and hydrogeological structures. This innovative approach transforms the task of 3D hydrogeological structure modeling into 3D geological structure modeling, where data are relatively abundant, thereby overcoming the data scarcity problem. To demonstrate the scientificity and feasibility of our approach, we utilize the 3D hydrogeological structures modeling of Wuhan’s metropolitan development area (MDA) as a case study. This study provides a quantitative criterion for the correlation between stratigraphic and hydrogeological structures, addressing the subjectivity and arbitrariness of previous qualitative evaluations. Additionally, it offers a scientific solution to the data scarcity issue commonly encountered in 3D hydrogeological structure modeling. Consequently, this study holds significant scientific value and practical implications.

1. Introduction

Groundwater is one of the vital types of water resources on earth [1,2]. With the rapid increase in global population and soaring socio-economic development, human reliance on groundwater has intensified [3,4,5], and so has the severity of groundwater pollution [6,7]. Consequently, the rational exploitation and protection of groundwater resources have emerged as pressing challenges for many countries [8,9]. The key to addressing these challenges lies in elucidating the recharge, transport, and discharge mechanisms of groundwater, which fundamentally requires understanding the spatial characteristics of aquifers [10,11]. 3D geological modeling is currently one of the important technical solutions for characterizing the 3D spatial structure of aquifers [12,13,14]. The 3D hydrogeological structure models can not only accurately delineate the three-dimensional spatial characteristics of aquifers but also provide detailed geological data and advanced 3D visualization support for groundwater simulation and analysis.

However, due to the high cost of hydrogeological exploration [15,16], hydrogeological data for the vast majority of cities are relatively scarce [17]. As a result, varying degrees of generalization are necessary in the process of 3D hydrogeological structure modeling [18]. The generalization of numerous aquifers and aquicludes at different scales inevitably leads to deviations or even errors in subsequent groundwater simulation and analysis, which currently restricts the application of 3D hydrogeological structure models in relevant thematic studies of groundwater [19]. In fact, relative to the information granularity and abundance required for most urban groundwater thematic studies on the 3D spatial structure of aquifers, there is always a scarcity of hydrogeological data [17,20].

Given the uniform geological genesis and evolution of stratigraphic units and their internal geological attributes, the conventional solution posits varying degrees of correlation between the spatial structure of stratigraphic units and the spatial structures of their internal geological properties, including hydrogeology [11]. Therefore, the geological data from various fields, such as basic geology, Quaternary geology, and engineering geology, can be utilized for 3D hydrogeological modeling [21]. These data sources encompass boreholes, geological profiles, thematic maps, and seismic interpretations [22]. However, the current analysis of the correlation between stratigraphic structures and hydrogeological structures primarily relies on experts’ professional experience and geological knowledge, inevitably leading to issues of subjectivity and arbitrariness.

Wuhan is located in the eastern part of the Jianghan Plain and falls within the Wuhan Platform Fold–Thrust Belt [23]. Since the Neotectonic Movement, the differential uplift and subsidence of the crust have been intense, resulting in the common development of mountain depressions or fault basins of varying sizes in the study area. Many of these are covered by Quaternary loose sediments, while others have formed numerous lakes scattered across the region. Additionally, the intense fault activities have had a significant impact on the formation and evolution of major rivers such as the Yangtze River and Han River in Wuhan. Consequently, the hydrogeological spatial structure of Wuhan is characterized by complexity and fragmentation. Macroscopically, it is interconnected through intricate faults, while locally, it exhibits distinct features due to differences in geological structures and stratigraphic genesis. Systematically understanding the spatial structure characteristics of the hydrogeological structure in the study area can effectively avoid damage to groundwater recharge, flow, and discharge conditions during underground space development and utilization. This is of great significance for the prevention and control of water inrush in engineering projects, the protection of groundwater resources, and the prevention of karst collapses, which are prominent issues in the study area [24].

The main approach of this study is exemplified by modeling the 3D hydrogeological structure of Wuhan’s MDA. It begins by analyzing the spatial correlation between stratigraphic and hydrogeological structures. From this analysis, a quantitative analysis model is established to address the randomness and subjectivity of traditional qualitative evaluation methods. Using the quantitative analysis results as criteria, the study employs a correlation-based modeling approach to convert 3D hydrogeological structure modeling into 3D geological structure modeling, thereby solving the issue of data scarcity.

The primary research strategy in this study involves utilizing the 3D hydrogeological structure modeling of Wuhan’s MDA as a case study. It begins with an in-depth analysis of the spatial correlations between stratigraphic and hydrogeological structures. Subsequent to this analysis, a rigorous quantitative analysis model is formulated to mitigate the inherent randomness and subjectivity associated with traditional qualitative evaluation methodologies. Utilizing the quantitative analytical outcomes as criteria, the study employs a correlation-based modeling approach to convert 3D hydrogeological structure modeling into 3D geological structure modeling, thereby effectively addressing the challenge of data scarcity.

2. Data and Methods

2.1. Study Area

Wuhan is located in the eastern part of the Jianghan Plain, at the confluence of the Yangtze River and the Han River, spanning from 113°41′ E to 115°05′ E longitude and 29°58′ N to 31°22′ N latitude [25]. The MDA of Wuhan serves as the city’s core commercial and residential district, covering an area of 3400 km². Wuhan experiences a subtropical monsoon climate, characterized by distinct seasonal variations in precipitation, which is primarily concentrated in spring and summer, with an average annual rainfall of 1204.5 mm. The MDA of Wuhan boasts a well-developed water system centered around the Yangtze River and Han River, which serve as its main trunks, and intricately interconnected, including over 50 major lakes, such as Donghu Lake, Tangxun Lake, Guanlian Lake, Nan Taizi Lake, and Zhushan Lake, covering a total water area of 429.75 km² [26]. Based on factors such as topography, elevation, material composition, and geomorphic origins, the MDA of Wuhan can be divided into five geomorphic regions: erosion low hill, erosion medium hill, fluvial plain, erosion-accumulation wave-like plain, and erosion-accumulation hillock plain [27], as illustrated in Figure 1.

2.2. Stratigraphic Structure and Lithologic Characteristics

The Wuhan MDA contains strata from the Neoproterozoic to the Cenozoic. The Neoproterozoic strata are sparsely distributed, only exposed in Qihu Village, Yangluo Street, Xinzhou District, and are primarily composed of metamorphic rocks belonging to the Wudang formation. The Paleozoic to Mesozoic strata (Silurian to Cretaceous), with a few exceptions, have developed a relatively complete set of terrestrial and marine sedimentary rocks [28]. In some local areas, magmatic intrusions and extrusions occurred during the Jurassic and Cretaceous periods. Except for the Fentou Formation of Silurian (S₁f), which is composed mainly of mudstones and shales, the rest of the Paleozoic to Mesozoic strata are predominantly carbonate rocks. These rocks are not widely exposed at the surface and are mostly concealed beneath the Cenozoic strata. Due to intense water–rock interactions and tectonic activities in the study area, fractures are well developed within the carbonate rocks. Under multiple large-scale tectonic movements, the Paleozoic to Mesozoic strata have formed a series of east–west trending fold structures, which, together with widely developed faults, have shaped the spatial pattern of bedrock structures in the study area (Figure 2). The Paleogene strata of the Cenozoic are widely distributed in faulted basins and are primarily composed of clastic rocks. Overlying these are sporadically distributed Neogene strata, which are not exposed at the surface and are buried beneath Quaternary loose sediments [29].

Since the Quaternary period, the crustal movement in Wuhan has been characterized by subsidence, resulting in the extensive coverage of Quaternary strata. The Lower Pleistocene (Qp1) consists of clayey sand formed by alluvial processes. The Middle Pleistocene (Qp2) is characterized by clayey soil formed by fluvial and alluvial processes. The Upper Pleistocene (Qp3) is dominated by silt and muddy sand formed by lacustrine and fluvial processes. The Holocene (Qh) comprises gravel, coarse sand, fine sand, and sandy silt, all formed by fluvial processes [30].

The stratigraphic sequence and lithology of the study area are detailed in Figure 3. Overall, the hydrogeological characteristics within the Quaternary unconsolidated sedimentary strata are primarily determined by the lithofacies features, including lithology and porosity. Significant differences in lithofacies and sedimentary origins among Quaternary strata from different periods lead to substantial variations in their hydrogeological characteristics. Specifically, the Holocene (Qh) and Late Pleistocene (Qp3) strata, dominated by sandy layers, form the phreatic aquifer in the study area. The Middle Pleistocene (Qp2) strata, predominantly composed of clayey soil, constitute the aquitard. The sporadically distributed Early Pleistocene (Qp1) strata, mainly consisting of clayey sandy soil, underlie and form the Quaternary unconsolidated porous confined aquifer [31]. However, for bedrock strata, the situation is different; they are primarily characterized by fractured aquifers. Due to the interconnection facilitated by fault structures in the study area, the fractures in various bedrock strata are well-connected, forming a large-scale bedrock fracture-confined aquifer system.

2.3. Hydrogeological Structural Characteristics of the Strata

Since the Paleozoic era, the MDA of Wuhan has evolved under the combined influence of regional geological tectonics and Quaternary geology, geomorphology, and meteorology, leading to the current regional hydrogeological configuration. These factors have formed the current hydrogeological pattern, which is primarily composed of three aquifer groups: the carbonate rocks series, the clastic rocks series in faulted basins, and the loose sedimentary layers dominated by fluvial alluvial and proluvial deposits.

2.3.1. Analysis of Hydrogeological Structural Characteristics of Quaternary Strata

The tectonic movements since the late Quaternary have significantly influenced the distribution of river networks and sediments. These movements have led to the formation of three distinct terraces in the region, primarily due to episodic crustal uplift and subsidence. The first and second-order river terraces are respectively covered by tens of meters thick Holocene (Qh) and Upper Pleistocene (Qp3) alluvial and fluvial deposits. The Holocene (Qh) is primarily composed of gravel and sand layers with varying particle sizes. The Upper Pleistocene (Qp3) consists mainly of finer-grained sandy soil, locally interspersed with minor amounts of sandy clayey soil and gravel layers. These formations create two unconfined aquifers with distinct hydraulic properties but close hydraulic connections. The Middle Pleistocene (Qp2) is characterized by extensive coverage of reticulated red clay (alluvial–proluvial facies), which exhibits exceptionally poor permeability, serving as an aquiclude for the Quaternary unconsolidated strata across the entire study area. The Lower Pleistocene (Qp1) strata are relatively scarce and scattered within the study area, and their lithology is dominated by clayey sandy soil and cohesive soil, forming the unconsolidated porous confined aquifer of the study area.

Figure 4 provides a comparative analysis of the stratigraphic structures and their internal hydrogeological characteristics during different Quaternary periods in the study area. Overall, although the frequent episodic uplift and subsidence movements of the crust due to neotectonic activities in the study area, the thickness of unconsolidated strata in different Quaternary periods is generally thin. However, the uplift and subsidence trends of the crust within the same geological period are generally consistent, leading to relatively stable depositional environments within each period. This consistency results in stable provenance and hydrodynamic conditions within each stratum, such that although there are slight differences in lithology within strata from different Quaternary periods (Figure 4a), their hydrogeological characteristics are similar. Macroscopically, this is manifested in the high degree of unity between aquifer types and strata from different Quaternary periods (Figure 4b).

2.3.2. Analysis of Hydrogeological Structural Characteristics of Bedrock

During the Yanshanian orogeny, the study area underwent the formation of a series of densely folded structures trending northwest–west, composed primarily of carbonate rocks spanning the Devonian, Carboniferous, Permian, and Triassic periods. These folds are intricately interwoven with faults, giving rise to the fracture–karst aquifer within the study area. Under the infiltration of precipitation or recharge from overlying aquifers, this aquifer serves as pathways and reservoirs for groundwater transport and storage, extending linearly along the syncline axis. In contrast, the Silurian shale and mudstone in the anticline core have weak water-bearing capacity, becoming the largest aquiclude in the study area.

The faulted basins and uplifted blocks along the eastern margin of the second subsidence zone of the Neocathaysian System are interspaced and adjacent to each other. The faulted basins, built upon the foundation of Paleozoic or older folds, have accumulated relatively thick Cretaceous-Paleogene strata, consisting primarily of clastic rocks with well-developed fractures, forming a fracture-confined aquifer (K-E). Additionally, in the western Donghu and Xihu District and the northern Huangpi District of the study area, Neogene strata are also present, comprising a suite of gray-green claystones and clastic rocks in a semi-consolidated state. These strata have formed a fracture-pore confined aquifer.

Figure 5 provides a comparative analysis of the stratigraphic structures and internal hydrogeological characteristics of bedrocks from different geological ages within the study area. In general, the spatial structure of the bedrock strata governs the variability of its internal hydrogeological characteristics. Due to the cumulative effects of the Indosinian, Yanshanian, Himalayan, and subsequent neotectonic movements, the geological structures in the study area are complex and variable. Consequently, the strata formed during different geological ages are generally thin, and the lithology and hydrogeological characteristics of each stratum exhibit distinct properties. Nevertheless, within each stratum, the lithology and hydrogeological characteristics are relatively uniform and highly correlated.

2.4. Data

2.4.1. Hydrogeological Borehole Data

This study collected a total of 159 hydrogeological borehole data, as illustrated in Figure 1. Of these, 46 boreholes specifically targeted the aquifer exploration within the Quaternary unconsolidated strata without penetrating into the bedrock aquifer. Meanwhile, an additional 113 boreholes successfully exposed both the unconsolidated and bedrock aquifers. In total, these 159 hydrogeological boreholes encompass 1006 cataloged information for various aquifer types and 4105 stratigraphy logging information, as shown in

Table 1. Summary of aquifer and stratigraphic logging information from hydrogeological boreholes.

Objets	Statistical Categories	Number of Samples
Aquifer types	Quaternary unconsolidated aquifer	443
	Bedrock aquifer	563
	Total	1006
Stratigraphy Logging	Quaternary	2340
	Bedrock	1765
	Total	4105

. This study primarily utilizes these data in conjunction with the 3D geological structural model of the study area to analyze the hydrogeological characteristics of each stratum.

2.4.2. 3D Geological Structural Model

The 3D stratigraphic structural model of Wuhan’s urban development area (Figure 6) covers an area of approximately 3400 km² with a vertical depth of 200 m. The model comprises 26 strata (Table 1), including 4 Quaternary loose sedimentary strata with relatively thin layers, none exceeding a maximum thickness of 20 m, and 22 bedrock strata with a maximum thickness not exceeding 60 m. Additionally, the model incorporates 195 faults and 37 folds. The construction of this model fully utilized the rich multi-source and heterogeneous geological data, including drill holes, profiles, and thematic maps from various specialties such as geological structures, bedrock geology, and Quaternary geology. It uses more than 236,000 geological boreholes and 12 comprehensive geological profiles as primary modeling data, with GOCAD 17 as the modeling software. GOCAD is currently one of the most widely used three-dimensional geological modeling software internationally. It incorporates extensive auxiliary processing functions for geological data and various 3D geological modeling methods, making it particularly suitable for research on 3D geological modeling in complex geological conditions [32,33]. Our study employs the grid-based modeling scheme, with a grid standard of 50 × 50 × 0.5 m³, and the model boasts a grid scale of 573 million and a volume of 63.4 GB. The overall accuracy of the model is 87.6%, enabling a detailed representation of the 3D spatial structure of each stratum within the study area. This study primarily utilizes this model as the analytical subject, establishes associations with the stratigraphic information from hydrogeological boreholes, and employs inferential modeling techniques to construct a 3D hydrogeological structural model of the study area.

2.5. Method

The fundamental methodology of this research endeavor involves the correlation analysis between stratigraphic structure and hydrogeological structure. Utilizing the aquifer logs and stratigraphic logs from hydrogeological boreholes as sample datasets, we establish a quantitative correlation model between different aquifer types and stratigraphic types. Based on the model, we analyze and evaluate the aquifer type of each stratigraphic layer. Subsequently, we employ a correlation-based modeling approach to construct a 3D hydrogeological structure model for the study area. The detailed research scheme is illustrated in Figure 7.

2.5.1. Delineation of Aquifer Types Through Basic Sample Statistical Analysis

In this research, we evaluated the aquifer type of each stratum using 159 hydrogeological borehole datasets based on the strong correlation between the stratigraphic and hydrogeological properties within the stratum. Initially, these 159 borehole datasets were partitioned into 127 experimental datasets and 32 validation datasets in an 8:2 ratio. The former was predominantly employed to delve into the hydrogeological characteristics of the stratigraphic units within the study area, whereas the latter served as a validation tool to ensure the correctness of the analytical outcomes. The detailed allocation scheme is presented in

Table 2. The allocation scheme of hydrogeological borehole data.

Categories	Types of Hydrogeological Boreholes	Number of Samples
Experimental datasets	Quaternary unconsolidated aquifer	37
	Bedrock aquifer	90
	Total	127
Validation datasets	Quaternary unconsolidated aquifer	9
	Bedrock aquifer	23
	Total	32

.

Subsequently, the hydrogeological borehole data were mapped into the 3D stratigraphic structural model of the study area based on their coordinate positions, enabling the statistical analysis of aquifer types and their corresponding sample sizes within each stratigraphic layer in the model. However, due to the differing standardization schemes between the hydrogeological boreholes and those used for 3D geostructural modeling, the stratigraphic information from the hydrogeological boreholes may not perfectly match the corresponding locations in the model. Therefore, it was crucial to validate the validity of the hydrogeological borehole data. This was done by comparing the lithological logging data from the hydrogeological boreholes with the standardized data used for 3D geological modeling. The results show that 96.57% of the lithological information is consistent between the two datasets. Consequently, the aquifer types within each stratum in the model can be inferred based on the statistical analysis of the aquifer types and their sample sizes from the hydrogeological boreholes. The statistical results of the aquifer types for each stratum in the model are shown in Figure 8.

According to the statistical results, apart from the lack of sample data for volcanic rock (K₂-E₁) and intrusive rock (J) in the model, the other stratigraphic units such as Early Pleistocene (Qp1), Wanglongtan (T₃-J₁w), Longtan (P₃l), Maokou (P₂m), Dapu (C_1-2d), Hezhou (C₁h), and Wudang (Nh₁W) also have relatively scarce sample data. The primary reasons for this include not only the scarcity and uneven distribution of the collected hydrogeological borehole data but also the inherently limited distribution ranges of these stratigraphic units.

Further analysis reveals that each stratigraphic unit exhibits a pronounced dominance of a specific aquifer type in terms of sample data volume, underscoring the stability of hydrogeological characteristics within individual stratigraphic units. Specifically, the Holocene (Qh) and Late Pleistocene strata (Qp3) have the highest number of samples for unconsolidated porous phreatic aquifers, with 118 and 102 samples, respectively. The Middle Pleistocene (Qp2) strata have the highest number of clayey aquiclude samples, with 98. The Early Pleistocene (Qp1) strata have the most samples for unconsolidated porous confined aquifers, totaling 7. Among the bedrock strata, with exceptions such as Guanghuasi (N1g) (9 samples of clastic rock pore-fracture-confined aquifer), Gonganzhai (K₂-E₁g) (72 samples of clastic rock fracture-confined aquifer), Fentou (S₁f) (68 samples of sandy shale–mudstone aquiclude), and Wudang (Nh₁W) (1 sample of metamorphic rock aquiclude), the majority of samples in other strata belong to karst fractured aquifers. To elaborate on the distribution of sample data for different aquifer types within each stratigraphic unit, normalization was performed according to Equation (1).

$$k_{ij}={\displaystyle \frac{n_{ij}}{{\displaystyle \sum _1^j{n}_{ij}}}}$$

(1)

where $k$ represents the proportion of the sample size of an aquifer type within a stratum, $i$ denotes the type of stratigraphic unit ($i\le 26$), $j$ denotes a specific aquifer type within a particular stratigraphic unit, and $n$ represents the sample size of a certain aquifer type within a stratum.

After normalizing the sample data, the proportion of the samples for each aquifer type within every stratum is presented in Figure 9. Notably, there are no sample data for volcanic rock (K₂-E₁) and intrusive rock (J). Additionally, the sample sizes of all aquifer types in the Dalong Formation (P₃d) exhibit a relatively even distribution. The remaining 23 strata each have a dominant aquifer type with a sample size proportion generally above 0.8. This indicates the absolute control and high correlation of the stratigraphic structure, internal lithology, and structure within the study area on its internal hydrogeological characteristics.

2.5.2. Revised Analysis of Aquifer Types Based on Cumulative Thickness of Aquifer

Reliance solely on the sample data of different aquifer types in hydrogeological boreholes to determine aquifer types within the corresponding strata ignores two crucial factors: the variability of borehole depths and the repeated occurrence of the same aquifer type at different depths within the same borehole. This represents a pervasive issue in 3D geological modeling, which can introduce errors into the analysis results [17,20].

As illustrated in Figure 10, while boreholes I, II, IV, and V have successfully intersected all aquifer types within strata such as Qh, Qp3, and Qp2, borehole III fails to uncover the colluvial clay aquiclude in the Qp3 stratum. This leads to an undercount of colluvial clay aquiclude samples when compiling statistics for all aquifer types in the Qp3 stratum. Additionally, the repeated occurrence of the sand pebble aquifer in boreholes I, IV, and V artificially inflates its sample count compared to colluvial clay aquiclude in the Qp3 stratum. Based on these skewed statistics, one might erroneously infer that the Qp3 stratum is primarily composed of sand pebble aquifer, whereas in reality, colluvial clay aquiclude dominates the Qp3 stratum. Therefore, directly determining the aquifer types within strata based solely on the sample size of different aquifer types encountered in hydrogeological boreholes may result in biases and inaccuracies.

To address this issue, this study further refines the aforementioned analytical approach by incorporating the cumulative thickness of aquifers. The main approach involves utilizing the cumulative thickness of each aquifer type within every stratum (as shown in

Table 3. Statistics on the maximum samples and largest cumulative thickness of aquifer types within each stratum.

Strata	Aquifer Type with Maximum Number of Samples	Number of Samples	Cumulative Thickness of Each Aquifer Type(m)
Qh	Unconsolidated porous unconfined aquifer	118	1069.3
Qp3		102	784.2
Qp2	Clayey aquiclude	98	826.7
Qp1	Unconsolidated porous confined aquifer	7	25.8
Guanghuasi (N1g)	Clastic rock pore-fracture-confined aquifer	9	102.8
Volcanic rock (K2-E1)	Volcanic rock aquiclude	0	0
Gonganzhai (K2-E1g)	Clastic rock fracture-confined aquifer	72	687.6
Intrusive rock (J)	Intrusive rock aquiclude	0	0
Wanglongtan (T3-J1w)	Fracture–karst aquifer	1	5.6
PuQi (T2p)		3	11.2
Jialingjiang (T1-2j)		7	65.3
Daye (T1d)		45	397.5
Dalong (P3d)	Fracture–karst weak aquifer	5	10.3
	Fracture–karst aquifer	3	43.1
Longtan (P3l)	Fracture–karst aquifer	3	30.8
Gufeng (P2g)		23	138.5
Maokou (P2m)		3	21.6
Qixia (P2q)		32	208.7
Liangshan–Chuanshan (P2l-P1c)		4	27.4
Huanglong (C2h)		39	295.6
Dapu (C1-2d)		3	22.3
Hezhou (C1h)		1	7.6
Hezhou–Gaolishan (C1h-g)		40	305.4
Huangjiadeng (D3h)		8	38.6
Yuntaiguan (D3y)		36	269.7
Fentou (S1f)	Sandy shale–mudstone aquiclude	68	526.7
Wudang (Nh1W)	Metamorphic rock aquiclude	1	6.8

Note: The volcanic rock aquiclude (K₂-E₁) and the intrusive rock aquiclude (J) were determined based on relevant hydrogeological investigation reports in the study area.

) as the basis for evaluation. Following normalization, the proportion of each aquifer type within the stratum is determined, which then serves as a criterion to infer and identify the aquifer types of the strata.

After normalization according to Formula (2), the proportion of each aquifer type within the stratum is determined, which is then used as the basis for inferring and discriminating the aquifer types within that stratum [34].

$$K_{ij}={\displaystyle \frac{M_{ij}}{{\displaystyle \sum _1^j{M}_{ij}}}}={\displaystyle \frac{{\displaystyle \sum _{a=1}{m}_{ij}^a}}{{\displaystyle \sum _1^j{\displaystyle \sum _{a=1}{m}_{ij}^a}}}}$$

(2)

where $K$ represents the proportion of thickness for a specific aquifer type within a stratigraphic unit, $i$ denotes the type of stratigraphic unit ($i\le 26$), $j$ represents the aquifer type within a particular stratigraphic unit, $M$ signifies the cumulative thickness of a specific aquifer type within the stratigraphic unit, $a$ represents the number of occurrences of a specific aquifer type within a stratigraphic unit, and $m$ indicates the thickness of a particular segment of an aquifer type within the stratigraphic unit.

In Table 3, within the Dalong (P₃d) stratigraphic unit, there are five samples of fracture–karst weak aquifer with a proportion of 0.625 and a cumulative thickness of 10.3 m, while the fracture–karst aquifer has three samples with a cumulative thickness of 43.1 m and a proportion of 0.807. The inconsistency in results based on these two criteria arises from the fact that the hydrogeological boreholes encountered two repeated occurrences of fracture–karst weak aquifer in the Dalong (P₃d) unit, leading to a bias in the former’s analysis. This indicates that using the cumulative thickness of aquifer types as a criterion can overcome analytical biases caused by variations in borehole depths and repeated occurrences of the same aquifer type within the same stratigraphic unit.

Figure 11 summarizes the cumulative thickness proportions of the primary aquifer types in each stratigraphic unit, all exceeding 0.8, demonstrating their absolute dominance within those units. Additionally, the proportion of sample counts for the primary aquifer types in each stratigraphic unit is smaller than their cumulative thickness proportions, suggesting that the criterion based on cumulative aquifer thickness optimizes the discrimination process for stratigraphic aquifer types.

2.5.3. Classification of Aquifer Types Within the Stratigraphic Units

Utilizing the cumulative thickness of each aquifer type as a criterion, we statistically analyzed the cumulative thickness of all aquifer types within each stratum. These values were then normalized to determine the degree of association between each stratum and its internal aquifer types. This is specifically illustrated in Figure 12.

The Quaternary loose sedimentary strata in the study area are primarily composed of phreatic aquifers, which are distributed mainly in the relatively soft clay layers (Qh) and the clayey sandy soil and gravel layers (Qp3). The hydrogeological characteristics of these strata are primarily influenced by the lithology and geomorphological features. Specifically, the water-bearing capacity of the clay aquifers (Qh) is relatively poor, while that of the gravel aquifers (Qp3) is relatively better. Spatially, these aquifers are isolated by the reticulated red clay aquiclude (Qp2), and the sporadic clayey sand layer (Qp1) beneath them exhibits more pronounced confined aquifer characteristics [30,31]. The hydrogeological characteristics of the bedrock strata are primarily influenced by the lithology and internal fracture structures. The aquifers are primarily found in carbonate rocks or in clastic rocks with abundant fractures and are separated by the widespread shale or mudstone aquiclude (S₁f) within the area.

Based on the maximum accumulated thickness proportion of aquifers within each stratum in the study area, and in combination with the 3D geological structural model, the aquifer types of the strata in the study area are classified, as detailed in

Table 4. Classification of aquifer types in stratigraphic layers of the study area.

Groundwater Types	Aquifer Types	Corresponding Strata
Unconsolidated porous water	Unconsolidated porous phreatic aquifer	Qh
		Qp3
	Unconsolidated porous confined aquifer	Qp1
-	Clayey aquiclude	Qp2
Pore–fracture water	Clastic rock pore–fracture-confined aquifer	Guanghuasi (N1g)
Fracture water	Clastic rock fracture-confined aquifer	Gonganzhai (K2-E1g)
Karst water	Fracture–karst aquifer	All Jurassic(J)-Devonian(D)
Bedrock aquiclude	Volcanic rock aquiclude	K2-E1
	Intrusive rock aquiclude	J
	Sandy shale–mudstone aquiclude	Fentou (S1f)
	Metamorphic rock aquiclude	Wudang (Nh1W)

Note: The volcanic rock aquiclude (K₂-E₁) and the intrusive rock aquiclude (J) were determined based on relevant hydrogeological investigation reports in the study area.

.

3. Results

This study employs the cumulative thickness of different aquifer types from hydrogeological boreholes within each stratum as the criterion to quantitatively analyze their correlation with aquifer types (Figure 10). Based on the maximum cumulative thickness of aquifers, the aquifer type for each stratum is determined. Utilizing the 3D geological structural model as a model foundation, a 3D hydrogeological structural model of the study area is constructed through correlation modeling.

The hydrogeological structural model covers an area of 3400 km² and extends to a depth of 200 m, encompassing a total of 9 different aquifer types. The 3D spatial structural characteristics of each aquifer are shown in Figure 13.

The aquifers within the Quaternary unconsolidated sedimentary strata are dominated by the Holocene (Qh) and Late Pleistocene (Qp3) alluvial–proluvial phreatic aquifers, which spatially cover the entire research area. The Middle Pleistocene (Qp2) strata, primarily composed of clayey soil, serves as a crucial Quaternary aquiclude in the research area, relatively isolating the Quaternary phreatic aquifers from the underlying Early Pleistocene (Qp1) confined aquifers, which are dominated by clayey sandy soil. The sporadically distributed Early Pleistocene (Qp1) confined aquifers in the central part of the study area are mainly controlled by the overlying Middle Pleistocene (Qp2) confining layer, while those in the northern mountainous region are primarily influenced by the poor permeability of their own red clay. The 3D structural model of the Quaternary unconsolidated porous aquifer system in the study area is illustrated in Figure 14.

The bedrock aquifer types in the study area are primarily comprised of fracture–karst aquifers, clastic rock fractured confined aquifers, and locally, fissured-porous detrital rock confined aquifers [28,29]. The former are primarily distributed within the fold tectonic zones. Intense and continuous tectonic movements have resulted in complex variations in the 3D spatial structure of the aquifer rock formations. Most aquifer formations have dip angles greater than 65°, and they are fragmented by faults from different geological periods, leading to frequent hydraulic connections with nearby surface water systems such as the Yangtze River and Han River, resulting in frequent karst collapses in the study area. The 3D structural model of the bedrock aquifer system in the study area is illustrated in Figure 15.

The clastic rock confined aquifers are a group of aquifer formations overlying the faulted depression regions, which are significantly controlled by the differential uplift and subsidence movements of the strata in the study area. Among them, the fractured confined aquifer of the Gonganzhai formation (K₂-E₁g) is the most widely distributed, with its boundaries primarily controlled by faults, serving as important channels for hydraulic connections between this aquifer and other bedrock aquifer formations. On the other hand, the pore-fractured confined aquifer formations of the Guanghuasi formation (N₁g) are sporadically overlying the fractured confined aquifer of the Gonganzhai formation (K₂-E₁g). Their boundaries are controlled by smaller secondary faults, and the recharge, runoff, and discharge characteristics are closely related to those of the fractured confined aquifer formations of the Gonganzhai formation (K₂-E₁g).

4. Model Quality Evaluation

To validate the quality of the model, this study randomly and uniformly selected 20% of the hydrogeological boreholes (32 in total) within the study area as verification data. Given the significant variation in sample sizes across different aquifer types within the boreholes, to comprehensively assess the model’s accuracy, this study relies on the cumulative thickness of each aquifer type in the verification boreholes within the corresponding aquifer model. Firstly, the accuracy of each aquifer model is analyzed individually, and finally, the overall accuracy of the hydrogeological structural model is determined using a weighted average approach. The specific details are shown in

Table 5. Model accuracy assessment scheme.

Aquifer Types	Theoretical Number of Aquifer Samples	Actual Number of Aquifer Samples	Accuracy
Unconsolidated porous phreatic aquifer	495.5	465.9	0.94
Clayey aquiclude	168.6	156.8	0.93
Unconsolidated porous confined aquifer	28.6	26.1	0.91
Accuracy of the 3D structural model of Quaternary unconsolidated aquifer system			0.927
Clastic rock pore–fracture-confined aquifer	18.6	16.8	0.90
Clastic rock fracture-confined aquifer	162.7	141.5	0.87
Fracture–karst aquifer	526.2	478.8	0.91
Volcanic rock aquiclude	-
Intrusive rock aquiclude	-
Sandy shale–mudstone aquiclude	175.3	152.5	0.87
Metamorphic rock aquiclude	-
Accuracy of the 3D structural model of the bedrock aquifer system			0.888
Overall accuracy of the 3D hydrogeological structural model			0.908
Comprehensive accuracy after considering the uncertainty of 3D geological structure model			0.795

Note: the “Theoretical number of aquifer samples” refers to the cumulative thickness of samples of each aquifer type in the verification boreholes, assuming that the aquifer model has a 100% accuracy, under which all samples from each aquifer type in the verification boreholes would be accurately placed within their respective aquifer models. The “Actual number of aquifer samples” represents the cumulative thickness of samples from each aquifer type in the verification boreholes that are actually located within their corresponding aquifer models under real conditions. The corresponding accuracy is calculated as the ratio of these two values.

.

Except for the volcanic rock aquiclude, intrusive rock aquiclude, and metamorphic rock aquiclude, which have no verification data due to their limited distribution ranges, the verification accuracy of the other aquifer models exceeded 0.87. Specifically, the accuracy of the 3D structural model of the Quaternary unconsolidated aquifer system is 0.927, and that of the 3D structural model of the bedrock aquifer system is 0.888. The overall accuracy of the 3D hydrogeological structural model of the study area is 0.908, and the comprehensive accuracy, taking into account the uncertainty of 0.876 of the 3D geological structure model, is 0.795. These verification results demonstrate that the model accurately depicts the 3D spatial structural variations of the aquifers in the study area, thereby validating the rationality and scientific nature of the modeling approach based on the correlation between the hydrogeological structural features and stratigraphic structural features.

5. Conclusions

The hydrogeological structural characteristics of Wuhan’s MDA are determined by the stratigraphic structure, as well as the lithological and fracture characteristics within it. Notably, there are pronounced differences in hydrogeological features among distinct strata. However, within the same stratum, due to the relative homogeneity of its internal lithology and fractures, the hydrogeological characteristics remain relatively stable, manifesting in each stratum having a dominant aquifer type that prevails overwhelmingly.

This study quantitatively analyzed the correlation between stratigraphic structures and hydrogeological structures by using the aquifer types and stratigraphic logs from hydrogeological boreholes to identify the dominant aquifer type within each stratum. To address the limitations of using sample size of aquifer types as the criterion, a revised approach was proposed that utilizes the cumulative thickness of each aquifer type as the criterion to refine the quantitative analysis results. Based on this revised criterion and supported by the 3D geological structural model of the study area, the 3D hydrogeological structural model for Wuhan MDA was constructed using correlation modeling. The comprehensive accuracy of the model is 0.795.

The 3D hydrogeological structure model can accurately reveal the spatial structure variations of aquifers in real 3D space. It not only guides the avoidance of aquifer damage during the urban underground space development and utilization but also provides 3D model support for groundwater monitoring and simulation. Consequently, this model holds significant importance for the management and protection of groundwater resources. The 3D hydrogeological structure model should be recognized as one of the professional support platforms for smart city construction and management. Offering specialized information services and technical support to urban management can drive the self-upgrade and improvement of the model by continuously integrating new geological data.