Modeling of Hydrological Processes in a Coal Mining Subsidence Area with High Groundwater Levels Based on Scenario Simulations

Abstract

The Eastern Huang–Huai region of China is a representative mining area with a high groundwater level. High-intensity underground mining activities have not only induced land cover and land use changes (LUCC) but also significantly changed the watershed hydrological behavior. This study integrated the land use prediction model PLUS and the hydrological simulation model MIKE 21. Taking the Bahe River Watershed in Huaibei City, China, as an example, it simulated the hydrological response trends of the watershed in 2037 under different land use scenarios. The results demonstrate the following: (1) The land use predictions for each scenario exhibit significant variation. In the maximum subsidence scenario, the expansion of water areas is most pronounced. In the planning scenario, the increase in construction land is notable. Across all scenarios, the area of cultivated land decreases. (2) In the maximum subsidence scenario, the area of high-intensity waterlogging is the greatest, accounting for 31.35% of the total area of the watershed; in the planning scenario, the proportion of high-intensity waterlogged is the least, at 19.10%. (3) In the maximum subsidence scenario, owing to the water storage effect of the subsidence depression, the flood peak is conspicuously delayed and attains the maximum value of 192.3 m³/s. In the planning scenario, the land reclamation rate and ecological restoration rate of subsidence area are the highest, while the regional water storage capacity is the lowest. As a result, the total cumulative runoff is the greatest, and the peak flood value is reduced. The influence of different degrees of subsidence on the watershed hydrological behavior varies, and the coal mining subsidence area has the potential to regulate and store runoff and perform hydrological regulation. The results reveal the mechanism through which different land use scenarios influence hydrological processes, which provides a scientific basis for the territorial space planning and sustainable development of coal mining subsidence areas.

1. Introduction

Coal constitutes approximately 25% of the world’s primary energy consumption and occupies a significant position in the global energy system [1]. As the dominant energy source in China, coal has offered a crucial guarantee for the economic and social development of China. The energy structure in China, which is dominated by coal, will not undergo changes in the short run [2]. The surface subsidence induced by coal mining is the principal cause of the environmental deterioration in mining areas [3]. As an important coal base in China, the eastern area of Huanghuai has a high groundwater level and a large subsidence coefficient. Under the comprehensive influence of factors such as coal mining subsidence and groundwater recharge, a large area of subsidence waterlogging has emerged, constituting a typical high groundwater coal mining subsidence area [4]. A considerable number of villages and cultivated lands in the region have been devastated, posing a threat to the livelihoods of residents. Simultaneously, the subsidence and associated waterlogging have altered surface runoff and infiltration processes, leading to an imbalance in the regional hydrological cycle and consequently imposing significant impacts on the local ecological environment and social economy [5]. Land use changes in coal mining subsidence areas are dramatic, and the hydrological characteristics are complex. Accurately identifying the hydrological changes caused by coal mining subsidence under different land use scenarios not only provides scientific guidance for land spatial planning in mining areas but also offers strong support for ecological restoration, disaster prevention, and water resource management [6,7,8]. This helps facilitate the sustainable transformation of coal mining subsidence areas and promotes the coordinated development of the regional economy, ecology, and society.

The coal mining subsidence area is subject to the dual disturbances of urbanization and mining-induced subsidence, with particularly prominent ecological environment issues. It has long been the focus of academic research. Currently, research on coal mining subsidence areas predominantly emphasizes aspects such as monitoring [9] and forecasting [10] of surface subsidence, disaster risk, and safety assessments [11], resource utilization and management strategies [12], and ecological restoration in mining regions [13]. However, limited attention has been devoted to the coupling mechanisms between land use changes and hydrological processes within these subsidence areas, as well as the hydrological response characteristics under varying mining scenarios. With the development of technology, hydrological numerical simulation has emerged as an important approach [14] to reflect the relationship between the spatial heterogeneity of the study area and the variations in hydrological characteristics [15]. Hydrological simulation models have evolved from simple conceptual frameworks to sophisticated hydrodynamic systems, progressing from statistical representations to deterministic formulations. Currently, widely utilized hydrological models include SWAT [16,17], SWMM [18], TOPMODEL [19], etc. MIKE 21 stands as a two-dimensional hydrodynamic hydrological model that has undergone extensive validation and application in areas such as stormwater system evaluation and pluvial flooding analysis. Simulation outcomes offer enhanced representation of regional runoff processes—including spatial distribution, depth, and duration of inundation zones. Consequently, this model provides robust hydrological underpinning for land use planning [20].

Nigussise et al. [21] used the MIKE 21 model to simulate the flood inundation risk areas in the Ayamama watershed under urbanization scenarios. Zandsalimi, et al. [22] applied the MIKE 21 model to assess the impact of different digital elevation models on urban flood modeling by generating flood inundation maps and other related analyses.

The MIKE 21 model, on the other hand, is unable to predict land use changes and has limitations in assessing hydrological variations under different land use scenarios. As a result, integrating the MIKE 21 model with land use prediction models is essential for exploring hydrological changes in coal mining subsidence areas under different land use scenarios [23,24]. At present, the development of land use prediction technologies is constantly improving, and models such as CLUES and FLUS have been developed based on the cellular automata model [25,26]. However, the above-mentioned land use prediction models perform weakly in the comprehensive simulation of multiple land types [27]. Whereas the PLUS model effectively addresses the above problems by introducing the land expansion analysis strategy and the multi-type random patch seed generation mechanism, and it has the capacity to achieve high simulation accuracy and adaptation to the complex evolution of multiple land types [28]. Luan et al. [29] conducted multi-objective land use optimization by integrating the PLUS model with the NSGA model. Li et al. [30] developed a coupled land use–landscape ecological risk model-geographical detector-PLUS framework to perform ecological risk assessment, analysis, and simulation in the arid regions of the Yellow River Basin. Research indicates that the PLUS model has good applicability in the fields of urban planning and design [31], ecological environment protection [32], water resource management [33,34], and integration with other models [35]. Integrating the MIKE 21 model with the PLUS model provides a new idea for exploring multi-scenario hydrological simulation in coal mining subsidence areas.

The study takes the Huaibei Bahe River Watershed in the eastern part of the Huang–Huai region of China as the research object. It integrates the PLUS and MIKE 21 models to simulate the hydrological processes under different development scenarios and evaluates and analyzes the hydrological process responses in the Bahe River Watershed during the process of coal mining subsidence and post-mining ecological restoration, with the expectation of providing a scientific basis for the territorial spatial ecological restoration planning of coal mining subsidence areas with high groundwater levels.

2. Overview of the Study Area and Data Sources

2.1. Overview of the Study Area

The study area spans Huaibei City and Yongcheng City in China and is one of the fourteen major coal production bases in China. This area is located in the middle of the Huaibei Plain and has a temperate semi-humid monsoon climate, with precipitation concentrated from June to September. The region is characterized by numerous river and lake systems, with severe waterlogging caused by coal mining subsidence. The Bahe River, the main river within the territory, is approximately 11.3 km. Its main tributaries include Wangying Ditch, Cao Ditch, Ding Ditch, etc., all flowing into the Bahe River from the northwest to the southeast. In this study, we used the MIKE model to extract the Bahe River Watershed as the research object, with a water inflow area of approximately 145.73 km² (Figure 1). The groundwater level in the Bahe River Watershed is relatively high, and clay aquifuges are commonly distributed underground in the mining areas within the watershed, which is conducive to the formation and storage of water accumulation. Since the construction of large-scale coal mining began in 1992, the water area has increased from 10.53 km² to 18.39 km². Over more than 30 years, the average annual transferred area of water has reached 0.262 km², forming a large area of coal mining subsidence wetlands. The subsidence water accumulation areas and the built-up areas of towns are intertwined in the spatial structure. Large areas of cultivated land have been submerged by water, the river and lake systems have changed significantly, and the landscape pattern within the watershed has changed dramatically.

2.2. Data Sources and Processing

(1) Land use data. The main data used in this research are the Landsat series remote sensing images of 1992, 2007, and 2022, which were obtained through the Resource and Environmental Science Data Platform of the Chinese Academy of Sciences, with a spatial resolution of 30 m [36]. ENVI 6.1 software [37] was utilized for preprocessing, such as atmospheric correction and radiometric calibration, and the remote sensing images of the study area were subjected to supervised classification [38]. The supervised classification method employed in this research is the Maximum Likelihood Classification (MLC). This method is grounded in the Bayesian decision theory [39]. The likelihood of a given pixel belonging to a particular training sample is computed. Eventually, the pixel is grouped into the category with the highest likelihood value [40]. Based on the land use classification method of the Chinese Academy of Sciences [41] and the characteristics of land use in the study area, the land use in the study area was classified into six types: cultivated land, forest land, grassland, water area, construction land, and unused land. Integrating the physical geographical characteristics and actual management requirements of Huaibei City, the unused land within the study area mainly encompasses industrial and mining derelict land, sandy land, bare land, bare rocky and gravelly land, and seasonal waterlogged marshland formed as a result of coal mining subsidence. Subsequently, the classification results were verified and corrected based on the high-precision historical images of Google Earth and on-site mapping. Eventually, the land use classification results of the three periods were obtained. A total of 900 samples were used for training in the study, and 200 randomly generated samples were used to test the classification accuracy. The test results show that the overall accuracy of land use data in each year is above 89%. The accuracy assessment of each year is presented in

Table 1. Accuracy assessment of land use in each year.

	1992		2007		2022
Land Use Type	PA (%)	UA (%)	PA (%)	UA (%)	PA (%)	UA (%)
Cultivated land	87.2	85.3	89.1	87.6	91.2	88.5
Forest land	92.5	90.8	93.7	91.4	94.3	92.1
Grassland	83.4	81.7	85.2	83.9	88.6	86.3
Water areas	95.8	96.2	96.1	96.5	97.3	97.0
Construction Land	82.1	80.5	85.6	84.2	89.8	87.4
Unused Land	80.3	78.6	83.4	81.9	86.7	84.5
Overall accuracy (%)	89.5		90.8		92.3
Kappa coefficient	0.86		0.88		0.90

. The interpretation results of the remote sensing images met the research requirements.

(2) Socio-economic data. The extent of coal mining subsidence and the scope of basic farmland are sourced from the territorial spatial planning of Huaibei City. The data of major roads, railways, night lights, population density, locations of town government headquarters, and GDP in the study area were all acquired through the Resources and Environmental Science Data Center of the Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences.

(3) Climatic and environmental data. River system data originates from OpenStreetMap. Soil data are derived from the Harmonized World Soil Database.

3. Research Methods

The PLUS model and the MIKE model were integrated for the research to evaluate the variations in hydrological characteristics in the Bahe River Watershed under different future development scenarios. Among them, the PLUS model was employed to predict the land use in the Bahe River Watershed under various development scenarios, and the MIKE model was utilized to simulate the hydrological responses under different land use changes. Finally, the results were analyzed in terms of hydrological elements and hydrological characteristics. The flow chart is presented in Figure 2.

3.1. Establishment of a Multi-Scenario Land Use Simulation Model

We adopted the PLUS model to conduct multi-scenario simulations of land use changes in the Bahe River Watershed. This model is a new patch-level refined land use prediction model that can consider the driving effect of spatial policies. It integrates the Land Expansion Analysis Strategy (LEAS) module and the Cellular Automata Model based on Multiple-Type Random Patch Seeds (CARS) module, which can better simulate the changes in land use at the patch scale with higher accuracy [42]. The LEAS module extracts the land use patches that have changed by comparing the land use data of different periods and obtains the driving factors and development probabilities of the expansion of each land use type using the random forest algorithm. Combined with the CARS module, the land use demand, transfer cost matrix, and neighborhood weights are set to simulate the evolution of land use patches and predict the future land use spatial pattern [32]. In the research, the neighbourhood weights were obtained based on the expansion of each land use type from 1992 to 2022. Neighbourhood weights were calculated using Formula (1):

$$W_i={\displaystyle \frac{S_i-S_{min}}{S_{max}-S_{min}}}$$

(1)

In the formula, $W_i$ is the neighborhood weight for land use type $i$; $S_i$ is the change area of land use type $i$; $S_{max}$ is the maximum land class change area; and $S_{min}$ is the minimum land class change area.

3.1.1. Driving Factors

We comprehensively considered the natural, social, and economic conditions of the Bahe River Watershed, as well as the data availability, and selected 15 driving factors and 3 limiting factors for land use scenario prediction, as shown in

Table 2. Descriptions of driving factor data.

Data Types	Data Name	Data Description	References and Data Sources
Natural driving factors	DEM	Raster data, 30 m resolution, combining surface subsidence	The resource and environmental science data platform of the Chinese Academy of Sciences [41]; Huai Bei City’s Ecological Restoration Plan for Territorial Space (2021–2035) [43].
	Slope	Extracted based on DEM, 30 m resolution	Based on DEM data, it was obtained using GIS 10.3 software [44,45].
	Aspect	Extracted based on DEM, 30 m resolution
	Soil type	Raster data, 1 km resolution, resample from 1 km resolution to 30 m resolution	Harmonized World Soil Database (HWSD) [46,47].
	Distance to river	Euclidean distance, 30 m resolution	The resource and environmental science data platform of the Chinese Academy of Sciences [41,47].
	Groundwater level	Kriging interpolation method for water level monitoring station data, 30 m resolution	National Meteorological Science Data Center [48].
	Precipitation	Kriging interpolation method for 2022 mean precipitation raster data, 30 m resolution	The resource and environmental science data platform of the Chinese Academy of Sciences [41].
Socioeconomic Driving Factors	Nighttime lights	2022 Nighttime lights raster data, 30 m resolution
	GDP	2022 GDP raster data, 30 m resolution
	Population	2022 Population raster data, 30 m resolution
	Distance to town	Euclidean distance, 30 m resolution	The resource and environmental science data platform of the Chinese Academy of Sciences [41,49].
	Distance to road	Euclidean distance, 30 m resolution
	Distance to railway	Euclidean distance, 30 m resolution
	Impervious surface	Remote sensing image extraction, 30 m resolution	Based on remote sensing image data, extraction is carried out in GIS [45,50].
	Coal mining subsidence area	Huaibei City Land Spatial Ecological Restoration Plan 2022–2035, 30 m resolution	According to the document [51], process it in GIS [45].
Limiting factors	Ecological protection area	Remote sensing image extraction, 30 m resolution	Based on remote sensing images, extract in GIS [44,45].
	Cultivated land protection area	Huaibei City Land Spatial Master Plan 2021–2035, 30 m resolution	According to the document [52], process it in GIS [45].
	River protection zone	Remote sensing image extraction, 30 m resolution	Based on remote sensing images, extract in GIS [44,45].

.

3.1.2. Scenario Setup

To meet different development needs, comprehensively considering the land use transfer matrix of the Bahe River Watershed from 1992 to 2022 and relevant policies such as the “Overall Land and Space Planning of Huaibei City (2021–2035)” on the regulation of land development and the development of coal resources, three scenarios, namely the baseline scenario, the maximum subsidence scenario, and the planning scenario, were set to simulate the land use change trend of the Bahe River Watershed in 2027. In the setting of the research cost matrix, the cost transfer matrix for each scenario was set in combination with the current land use status and ecological environment policies of the Bahe River Watershed (

Table 3. Transition cost matrix for each scenario.

	Baseline Scenario						Maximum Subsidence Scenario						Planning Scenarios
	a	b	c	d	e	f	a	b	c	d	e	f	a	b	c	d	e	f
a	1	1	1	1	1	1	1	1	1	1	1	0	1	0	0	1	1	0
b	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1
c	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1
d	1	1	1	1	1	1	0	0	0	0	0	0	1	1	1	1	0	1
e	0	0	0	1	1	1	0	0	0	0	1	0	0	0	0	0	1	0
f	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1

a to f denote cultivated land, forest land, grassland, water areas, construction land, and unused land. A value of 1 indicates that transition can occur between two land use types, while 0 indicates no transition between two land use types.

). Among them, a-f represent cultivated land, forest land, grassland, water areas, construction land, and unused land in sequence. Based on the damage characteristics of land use caused by coal mining subsidence, the conversion of construction land to water area is allowed.

(1) Baseline scenario. The baseline scenario, as the reference scenario, continues the trend of land use change in the Bahe River Watershed from 2007 to 2022. Based on the characteristics of land use conversion obtained from the two-period land use data, combined with the Markov chain demand prediction module, the future land use is predicted. Only the river protection restricted area is set, without considering the influence of development plans and policy requirements, and the parameters are not adjusted [53]. In the setting of the cost matrix, except that construction land is prohibited from being converted into cultivated land, forest land, and grassland, other land types can be converted into one another to simulate the land use situation under the natural growth scenario in 2037, and this state is used as the reference scenario.

(2) Maximum subsidence scenario. The maximum subsidence scenario is built upon the baseline scenario. Based on the projected coal mining subsidence in the coming 15 years and the implementation of comprehensive governance in the coal mining subsidence area over the past 10 years, the conversion of cultivated land, forest land, grassland, construction land, and unused land to water areas within the future subsidence area is increased by 30%. Simultaneously, in the cost transfer matrix, the conversion of water areas with overly deep coal mining subsidence to other land types is restricted.

(3) Planning scenarios. The planning scenario needs to take into account ecological protection, cultivated land protection, and economic development simultaneously. According to the territorial spatial planning policies of the study area, the intensity of ecological restoration and land reclamation in the coal mining subsidence area is strengthened. Compared with the baseline development scenario, in this scenario, the conversion probability of cultivated land to forest land is increased by 20%, the conversion probability to construction land is slowed down by 30%, and ecological restricted areas and basic farmland restricted areas are added as new limiting factors. Considering that the waterlogged area of coal mining subsidence is converted into other land types through land reclamation, the conversion of water areas to other land types is allowed in the cost transfer matrix.

3.2. Establishment of the Hydrological Process Simulation Model

3.2.1. Parameter Setting

The MIKE 21 model can simulate water level and flow changes caused by various forces and any two-dimensional free surface flow, ignoring stratification [54]. The establishment of the model mainly comprises steps such as mesh division, roughness setting, and dry–wet water depth setting [55,56]. The mesh generator of the MIKE 21 model adopts unstructured irregular triangular meshes to divide the Bahe River Watershed into meshes. To meet the accuracy requirements, the area of the largest mesh in the study does not exceed 0.005 km². The study employs a flood simulation model within the local area, and the model boundary allows for free water flow; thus, an open boundary is adopted. In the specific parameter setting of the model, empirical values are adopted for parameters that have no obvious influence on the calculation results. Among them, the CFL number takes the default value of 0.08; in the dry-–wet boundary setting, the dry water depth is set at 0.001 m, the wet water depth is set at 0.002 m, and the inundation water depth is set at 0.0015 m; the vortex viscosity parameter is calculated using the Satoransky formula and takes a fixed value of 0.28 m²/s. Meanwhile, based on the “Hydraulic Calculation Manual” and comprehensively considering the land use types in the study area and the on-site investigation situation, the Manning coefficient values of roughness for each land type are determined: cultivated land 0.03, forest land 0.05, grassland 0.04, water areas 0.01, construction land 0.0125, and unused land 0.02.

3.2.2. Design Rainfall

In the study area, single-peak heavy rainfall is prevalent. Therefore, the Chicago rainfall pattern was chosen as the rain pattern for model design [57,58]. With reference to the “Notice on the Publication of the Rainstorm Intensity Formula of Huaibei City” revised by the Natural Resources and Planning Bureau of Huaibei City, the revised rainstorm intensity formula of Huaibei City is presented in Equation (2). These data are based on the complete rainfall observation sequence from 1951 to 2019 at the main meteorological station in Huaibei. They are derived from the Annual Maximum Series (AMS) through Log-Pearson Type III frequency analysis. Considering the distinctive nature of the study area as a coal mining subsidence area, the 100-year storm can mirror the hydrological response characteristics of the watershed under extreme conditions and better expose the potential rainstorm flood risks in this special area. Meanwhile, the study area is located in northern China, where rainfall is concentrated in summer and is of high intensity. Using the 100-year storm is more in line with the actual situation of the study area. A lower rainfall intensity might make the simulation results of the study overly optimistic and fail to fully reflect the risks the study area faces under extreme conditions. Thus, a 100-year storm scenario is established to simulate the rainfall situation.

$$q={\displaystyle \frac{1104.984\times \left(1+0.620lgP\right)}{{\left(t+4.203\right)}^{0.542}}}$$

(2)

In the formula, $q$ represents the design storm intensity, $L/\left(S\cdot h{m}^2\right)$; $P$ represents the design return period, $a$; and $t$ represents the duration of rainfall, $min$.

3.2.3. Evaluation of the Applicability of the PLUS-MIKE21 Model

The applicability of the PLUS model in land use prediction was evaluated using the Kappa coefficient (>0.75) and the overall accuracy. After calculation, the Kappa coefficient was 0.781 and the overall accuracy was 0.874. Due to the lack of necessary measured data in the study area for the calibration and validation of the MIKE 21 model, the study will initially select the corresponding parameters based on relevant data and experience. Then, through on-site investigation and inquiry to collect historical flood mark data and compare the simulation results, the rationality of the parameters was verified. The comparison between the calculation results and the measured flood levels is shown in

Table 4. Comparison between model simulation and the actual flood mark [59].

Serial Number	Location	Flood Mark Water Level (m)	Simulated Water Level (m)	Difference (m)
1	Qiangu Bridge	2.74	2.62	−0.12
2	Xiaoxin Village	1.61	1.43	−0.18
3	Xinxing Village	3.22	3.59	0.37

. After verification, the accuracies of both models met the requirements. Therefore, they were applicable to the analysis of rain and flood inundation under different land use scenarios in the Bahe River Watershed.

4. Results and Analysis

4.1. Land Use Analysis

It can be seen from the land use data from 1992 to 2022 (Figure 3) that, during this period, cultivated land has always been the most dominant land use type in the study area (

Table 5. Land use structure of the Bahe River Watershed from 1992 to 2022.

Land Use Type	1992		2007		2022
	Area (km2)	Proportion (%)	Area (km2)	Proportion (%)	Area (km2)	Proportion (%)
Cultivated land	102.85	70.58	94.09	64.56	87.38	59.96
Forest land	3.92	2.69	2.07	1.42	6.72	4.61
Grassland	0.31	0.22	1.45	0.99	1.35	0.93
Water areas	10.53	7.23	15.90	10.91	18.39	12.62
Construction land	27.84	19.10	31.94	21.92	30.69	21.06
Unused land	0.27	0.18	0.29	0.20	1.21	0.83
Total	145.73	100	145.73	100	145.73	100

). Affected by surface subsidence, the proportion of cultivated land area decreased from 70.58% in 1992 to 59.96% in 2022, over 30 years. For construction land, the proportion of its area increased by 1.96% over 30 years. Although the area of urban built-up areas expanded, large areas of village construction land sank into the water. The proportion of water areas increased from 7.23% in 1992 to 12.62% in 2022, with a cumulative increase of 7.85 km², an increase of up to 74.53%, and the growth total ranked first among land types. The area of forest land showed a trend of first decreasing and then increasing, with the proportion decreasing from 2.69% in 1992 to 1.42% in 2007. In 2022, due to the implementation of the ecological restoration policy in the coal mining subsidence area, the proportion of forest land recovered to 4.61%. The areas of grassland and unused land accounted for a relatively small proportion, and during this period, the areas of both slightly increased. Overall, the land use change in the Bahe River Watershed from 1992 to 2022 was relatively intense, and the water area continued to increase. With the continuous intensification of the urbanization process and coal mining subsidence, the area of cultivated land will be further encroached upon, the urban built-up areas show an expansion trend, and the areas of rural construction land and cultivated land decrease year by year.

Based on the three periods of land use status in the Bahe River Watershed from 1992 to 2022, a land use transfer matrix was constructed, and the data were subjected to visual analysis (Figure 4 and

Table 6. Land use transition matrices at each stage.

1992–2007	CL	FL	GL	WA	Con L	UL	Total	Transfer-Out Volume
CL	83.54	0.84	1.31	5.91	11.03	0.22	102.85	19.31
FL	2.57	0.17	0.01	0.30	0.88	0.00	3.92	3.76
GL	0.13	0.03	0.01	0.04	0.11	0.00	0.31	0.30
WA	0.81	0.15	0.01	7.60	1.91	0.05	10.54	2.93
Con L	7.04	0.88	0.10	2.04	17.77	0.00	27.84	10.07
UL	0.00	0.00	0.00	0.01	0.25	0.02	0.27	0.25
Total	94.09	2.07	1.45	15.90	31.94	0.29	145.73	36.62
Transfer-in volume	10.55	1.90	1.43	8.30	14.17	0.27	36.62
2007–2022	CL	FL	GL	WA	Con L	UL	Total	Transfer-out volume
CL	75.67	2.71	0.76	4.83	9.71	0.40	94.09	18.42
FL	0.98	0.39	0.00	0.14	0.53	0.02	2.07	1.68
GL	0.67	0.07	0.01	0.18	0.46	0.06	1.45	1.44
WA	2.35	1.69	0.25	10.17	1.29	0.15	15.90	5.74
Con L	7.70	1.86	0.33	3.07	18.41	0.56	31.94	13.53
UL	0.00	0.00	0.00	0.00	0.28	0.01	0.29	0.28
Total	87.38	6.72	1.35	18.39	30.69	1.21	145.73	41.07
Transfer-in volume	11.71	6.33	1.34	8.22	12.27	1.20	41.07
1992–2022	CL	FL	GL	WA	Con L	UL	Total	Transfer-out volume
CL	73.70	3.12	0.72	9.39	15.18	0.73	102.85	29.15
FL	2.57	0.41	0.05	0.21	0.67	0.01	3.92	3.52
GL	0.15	0.05	0.02	0.04	0.05	0.00	0.31	0.29
WA	1.59	1.15	0.07	5.51	2.08	0.12	10.54	5.02
Con L	9.35	1.98	0.49	3.21	12.47	0.33	27.84	15.37
UL	0.00	0.00	0.00	0.02	0.23	0.01	0.27	0.25
Total	87.38	6.72	1.35	18.39	30.69	1.21	145.73	53.61
Transfer-in volume	13.68	6.31	1.33	12.88	18.22	1.19	53.61

CL represents cultivated land; FL represents forest land; GL represents grassland; WA represents water areas; Con L represents construction land; and UL represents unused land.

). The data indicate that during the period from 1992 to 2022, the mutual conversion between cultivated land and construction land was particularly intense. The transfer-out amount of cultivated land was as high as 29.15 km², of which 15.18 km² was converted into construction land. In the coal mining subsidence areas, numerous villages were damaged and demolished, with the non-submerged portions being reclaimed as cultivated land after rehabilitation. From 1992 to 2022, a total of 15.37 km² of construction land was converted to other land types, of which 9.35 km² was transformed into cultivated land. This is driven by the reclamation and governance of land in coal mining subsidence areas [43,60]. Due to extensive water accumulation in areas with significant subsidence, the water body area in the study region rapidly increased. During this period, the water body area expanded by 12.88 km², with cultivated land and construction land contributing the most to this conversion—9.39 km² and 3.21 km², respectively. From 1992 to 2007, under the influence of rapid urbanization development, a large amount of forest land was converted into cultivated land and construction land. From 2007 to 2022, due to the governance of subsidence water accumulation areas carried out in the study area and the establishment of several urban parks, the area of forest land was restored. Under the regulation of ecological restoration policies such as the “Planning for the Protection and Governance of Mine Geological Environment in Huaibei City”, some water areas and cultivated land were converted into forest land.

4.2. Land Use Prediction Under Multiple Scenario Simulations

The simulation results of land use in various scenarios in the Bahe River Watershed in 2037 show (Figure 5 and

Table 7. Land use change in the Bahe River Watershed under different scenarios.

Land Use Type	2022	2037			Land Use Change from 2022 to 2037
	Current Situation (km2)	Base (km2)	Maximum Subsidence (km2)	Planning (km2)	Base (km2)	Maximum Subsidence (km2)	Planning (km2)
Cultivated land	87.37	77.05	74.52	82.20	−10.31	−12.85	−5.17
Forest land	6.74	8.78	7.70	9.48	2.04	0.97	2.74
Grassland	1.36	1.41	0.92	1.77	0.05	−0.44	0.41
Water areas	18.39	20.84	26.29	12.47	2.45	7.90	−5.91
Construction land	30.72	36.13	35.19	39.66	5.41	4.47	8.94
Unused land	1.22	1.58	1.17	0.21	0.36	−0.05	−1.01

) that cultivated land remains the most dominant land use type in the watershed (accounting for more than 60%). The land use changes in each scenario mainly consist of the mutual transformation among cultivated land, construction land, and water areas.

Compared with the land use data in 2022, the changes in land use under each scenario are evident. In the baseline scenario, all land use types except cultivated land have shown varying degrees of increase. Notably, construction land, water areas, and forest land exhibit significant growth trends. Specifically, the area of cultivated land has decreased by 10.31 km², while construction land has expanded by 5.41 km², with most of this increase concentrated in the southeastern part of the study area. The water area has grown by 2.45 km², primarily due to the expansion of coal mining subsidence-induced water accumulation zones. Additionally, the forest land area has increased by 2.04 km², reflecting a certain level of attention to ecological restoration efforts. For other land types not mentioned above, changes in area are relatively minor. Overall, the trend of cultivated land being progressively encroached upon by construction land and water areas within the watershed persists. In the maximum subsidence scenario, the coal mining subsidence area and the predicted data of coal mining subsidence are derived from the “Huaibei City Territorial Space Ecological Restoration Plan (2021–2035)” (Table 2) [43]. Among them, the maximum subsidence depth is 11,000 mm, and the average subsidence depth is 5000 mm. These data are calculated by the coal mine surveying department based on the geological conditions of the coal seam and the mining process, using the maximum probability integral method. The land use change in this scenario is the most remarkable, with the largest increase in water area, reaching 7.9 km². The newly added water area is highly concentrated, mainly distributed in the coal mining subsidence area. The area of cultivated land has shrunk sharply, up to 12.85 km², and most of the reduced cultivated land has become water areas in the subsidence area. A large amount of cultivated land has been destroyed due to coal mining subsidence, causing a huge impact on local agricultural production. The areas of construction land and forest land have increased by 4.47 km² and 0.97 km², respectively, and their increments are both less than those in other scenarios. These changes in land use types fully indicate that under the development model dominated by mining activities and without restoration intervention, continuous mining activities continue to cause a large amount of surface subsidence and eventually lead to an increase in water area. This land use change caused by coal mining subsidence makes the landscape pattern in this area unstable in a certain period in the future. The planning scenario is influenced by a range of measures, including government-led ecological restoration initiatives and comprehensive management strategies for coal mining subsidence areas. Significant changes are observed in the areas of cultivated land, forest land, and water bodies. Specifically, the decline in cultivated land area has been moderated, with a total reduction of 5.17 km². The forest land area has increased by 2.74 km², contributing to the protection of ecological land to some extent. In contrast with other scenarios, the planning scenario is unique in that it exhibits a contraction in water areas, with a decrease of 5.91 km². This reduction is primarily due to the reclamation of most subsidence-induced water accumulation zones as cultivated land. The construction land area has expanded markedly by 8.94 km². Influenced by the planning policies, this expansion predominantly occurs on the periphery of the densely developed construction land zone in the southeastern part of the watershed, leading to intensified encroachment on surrounding lands.

4.3. Hydrological Simulation Under Multiple Scenarios

4.3.1. Spatial Distribution Characteristics of Water Accumulation Range Under Multiple Scenarios

After coal mining subsidence, notable alterations take place in land use. This phenomenon induces substantial changes in the runoff generation and concentration processes within the basin, thereby augmenting the rainstorm and flood risks in the area. Consequently, analyzing and forecasting the variations in surface runoff processes holds crucial significance for the early warning of rainstorms and flood risks as well as land use planning.

The MIKE 21 model was employed to simulate the water accumulation range and maximum water depth under the maximum subsidence, baseline, and planning scenarios for a 100-year design rainfall event in the study area (Figure 6). The Bahe River Watershed exhibits a relatively flat topography with minimal elevation variations, resulting in relatively uniform simulated water patches. Nevertheless, significant spatial heterogeneity persists in the water accumulation range across the different scenarios. Spatially, water accumulation under each scenario is predominantly distributed along the river system and its surrounding areas, forming a band-like pattern. This outcome arises because the elevation data for each scenario were pre-adjusted to account for the river network prior to simulation, causing substantial surface runoff to converge into the river following rainfall. Under extreme rainfall conditions, river overflow occurs, leading to increased water accumulation within the river system and its adjacent areas. Additionally, higher water accumulation is observed in areas with dense construction land and its surroundings across all scenarios, whereas cultivated land experiences relatively lower water accumulation over large areas. This phenomenon can be attributed to the prevalence of impervious surfaces in densely constructed areas, which hinder infiltration. In contrast, cultivated land has highly permeable soil, enabling rapid rainwater infiltration and reducing surface runoff. These findings indicate that different land use types significantly influence the spatial distribution of water accumulation.

From the spatial distribution characteristics of water accumulation areas in each scenario, it is evident that the water accumulation range in the maximum subsidence scenario is significantly larger than that in the other two scenarios. In the maximum subsidence scenario, the extent and depth of water accumulation were calculated based on the current DEM data after being corrected with the subsidence prediction data for 2037 (Table 2) [43], using the MIKE 21 model. In this scenario, a substantial water accumulation area forms in the central coal mining subsidence zone of the watershed. This phenomenon occurs because mining-induced surface subsidence creates low-lying terrain in this region. During rainfall events, a significant volume of rainwater flows into the subsidence area. The relatively enclosed nature of the subsidence zone hinders timely drainage, leading to extensive water accumulation in this area. In contrast, the baseline scenario exhibits relatively mild subsidence, resulting in less severe water accumulation.

The water accumulation in the planning scenario is the least, with a small amount of water mainly converging around the river system and some water patches. This is because in the planning context, the policies of land reclamation and ecological restoration in the coal mining subsidence areas are comprehensively considered. The areas of cultivated land, forest land, and grassland increase, and the surface permeability is high, facilitating the infiltration and drainage of rainwater. The coal mining subsidence area in the middle of the watershed is planned as the Fengqi Lake coal mining subsidence wetland. The originally relatively closed subsidence water areas are connected to the surrounding rivers, lakes, and ditches, which is conducive to the rapid drainage of water accumulation.

4.3.2. Distribution Characteristics of Water Accumulation Intensity Under Multiple Scenarios

This study employs the MIKE 21 model to simulate the distribution of maximum water depth within the watershed under three scenarios: maximum subsidence, baseline, and planning. Based on guidelines from documents such as the “Code for Design of Urban Flood Control Project” (GB/T 50805–2012) and prior research, and considering the simulated extreme rainfall event with a 100-year recurrence interval, the intensity levels of water accumulation are defined as shown in

Table 8. Classification of water accumulation intensity.

Water Accumulation Intensity Levels	Depth of Accumulated Water (m)
Extremely low	<0.3
Low	0.3–0.5
Medium	0.5–1.0
High	>1

. Using GIS spatial analysis to quantify water accumulation intensity, the proportional area data for each intensity level of water accumulation zones were calculated, as presented in

Table 9. Proportional area of different water accumulation intensity under multiple scenarios.

Scenario Types	Water Accumulation Intensity	Proportion (%)
Base scenario	Extremely low	48.51
	Low	7.03
	Medium	13.11
	High	31.35
Maximum subsidence scenario	Extremely low	54.24
	Low	6.59
	Medium	12.27
	High	26.90
Planning scenarios	Extremely low	68.49
	Low	4.24
	Medium	8.17
	High	19.10

.

The data indicate marked variations in the proportions of water accumulation zones across different intensities within the watershed under various scenarios. Specifically, the high-intensity water accumulation zone in the maximum subsidence scenario covers the largest area, accounting for 31.35% of the total watershed. The baseline scenario ranks second with a proportion of 26.90%, while the planning scenario exhibits the smallest high-intensity water accumulation area, comprising only 19.10% of the total watershed. For low intensity and extremely low intensity zones, the planning scenario demonstrates a notably larger area compared to the maximum subsidence and baseline scenarios, reaching a proportion of 72.73%. Overall, the maximum subsidence scenario exhibits the highest water accumulation intensity, whereas the planning scenario shows the lowest intensity. When combined with the water accumulation distribution map, the high-intensity water accumulation zones in all scenarios are predominantly located within the coal mining subsidence area in the central part of the watershed. Coal mining-induced subsidence has created localized depressions in this region, hindering effective drainage and resulting in significantly greater water accumulation depths compared to surrounding areas, thereby posing a relatively higher risk. Additionally, varying degrees of water accumulation are observed in construction land-intensive areas across all scenarios, particularly in the built-up area in the southeastern part of the watershed. Among these, the maximum subsidence scenario encompasses the largest impact area, while the planning scenario performs best. This outcome can be attributed to the extensive surface hardening that impedes rainwater infiltration. Moreover, in the maximum subsidence scenario, the upstream coal mining subsidence area severely disrupts the hydrological cycle of the watershed. In contrast, the planning scenario is less affected by coal mining subsidence due to ecological restoration measures, which enhance vegetation coverage and water system connectivity, leading to improved water accumulation conditions.

4.3.3. Changes in Hydrological Characteristics Across Different Scenarios

Based on the above analysis, the spatial distribution of the water accumulation range, maximum water depth, and water accumulation intensity in the watershed is closely related to land use types, elevation, mining activities, and planning policies. To further explore the hydrological response characteristics and inherent mechanisms of the watershed, we will conduct an analysis from the aspect of the numerical evolution of the outflow discharge at the watershed outlet. In this study, we utilized MIKE 21 (2024) software to export the runoff data of the watershed outlet for 72 h after simulating a 24-h designed rainfall and calculate its runoff characteristic values (Figure 7).

The data indicate that the maximum subsidence scenario exhibits the lowest mean runoff, whereas the planning and baseline scenarios show relatively higher mean runoff values. Regarding peak runoff, the maximum subsidence scenario reaches the highest value of 192.3 m³/s, while the planning and baseline scenarios exhibit lower peak values. From the perspective of runoff distribution patterns, the flood peak in the maximum subsidence scenario occurs latest at the 18th hour, exhibiting the most significant lag. Compared to the baseline and planning scenarios, the maximum subsidence scenario exhibits the most severe land subsidence, creating larger low-lying areas with enhanced rainwater storage capacity and reduced water flow. Consequently, it has the smallest mean runoff value. Additionally, as rainfall progresses, upstream inflow accumulates within the subsidence zones, leading to a delayed flood peak timing. Once the storage capacity is reached, rainwater overflows from the subsidence areas, causing a large volume of water to pass through the outlet in a short period due to the rapid convergence process. As a result, the maximum subsidence scenario demonstrates the highest flood peak value. The subsidence conditions in the baseline and planning scenarios are relatively moderate. Compared with the maximum subsidence scenario, the subsidence areas in these scenarios can distribute part of the rainwater while not retaining a large amount of rainwater to the extent that extreme peaks occur. Therefore, they have a certain reduction effect on the peak flow.

It can be seen from the cumulative runoff data that the initial loss period of the maximum subsidence scenario occurs from approximately 0 to 16 h and then enters the strong runoff generation period, gradually entering the recession period at the 26th hour. The initial loss period of the baseline and planning scenarios occurs from approximately 0 to 12 h, entering the strong runoff generation period 4 h earlier than the maximum subsidence scenario and gradually entering the recession period at the 22nd hour. From the perspective of the total cumulative runoff, the total cumulative runoff of the maximum subsidence scenario is the smallest, the baseline scenario is in the middle, and the total cumulative runoff of the planning scenario is the largest. This is because the subsidence area in the maximum subsidence scenario is large, which has a storage effect on the upstream inflow. Runoff is generated only after this area is filled with water, thus resulting in the smallest cumulative runoff at the watershed outlet. Under the influence of land reclamation and ecological restoration strategies, the planning scenario can discharge the upstream inflow in a timely manner and therefore has the largest cumulative runoff.

5. Discussion

5.1. Intrinsic Causes of Hydrological Changes Under Multiple Scenarios

The hydrological response characteristics of each scenario vary significantly. The influence of coal mining subsidence and ecological restoration measures on the hydrology of the watershed is dual-sided. The study area is positioned in the middle of the Huaibei Plain and is classified as a coal mining subsidence area with high groundwater levels. Due to the high-intensity coal mining in history, this area has suffered from the impacts of coal mining subsidence [61]. Clarifying the causes of water accumulation in the subsidence area is the key to exploring the variations in its hydrological responses in multiple scenarios. Existing studies have indicated that the geology, hydrology, meteorology, and human activities in the Huaibei Plain are the crucial factors causing the problem of water accumulation in the subsidence area [62]. From a geological perspective, the study area is characterized by abundant coal resources, with coal seams that are both widespread and relatively shallow. The regional geological structure is complex, featuring well-developed fault zones and poor rock stability. The overburden of the coal seam primarily consists of loose sediments with low compressive strength, making it highly susceptible to collapse above mined-out areas [63,64]. Additionally, the groundwater level in the Huaibei Plain is relatively high (averaging less than 5 m), and the proximity of the phreatic water layer to the coal seam increases the likelihood of groundwater inflow during mining operations. Combined with the extensive surface water network and numerous rivers and lakes in the Huaibei Plain, which enhance the interaction between surface water and groundwater, coal mining-induced subsidence disrupts river connectivity, leading to the formation of waterlogged areas [65]. From a meteorological and climatic perspective, the Huaibei Plain exhibits a temperate semi-humid monsoon climate, characterized by concentrated and intense rainfall (primarily occurring from June to September and accounting for over 70% of the annual precipitation) [66]. This high-intensity rainfall significantly enhances surface runoff, creating favorable natural conditions for water accumulation in subsidence areas. Additionally, prolonged and intensive underground mining activities, coupled with the accelerated urbanization process in the study area, have disrupted the regional hydrological equilibrium, exacerbating the severity of water accumulation issues in the subsidence zones.

Under the maximum subsidence scenario, in this study, we performed land use prediction and hydrological simulation based on surface subsidence rates and predicted subsidence extents. According to the watershed outlet discharge data and water accumulation range data, the average and cumulative runoff under the maximum subsidence scenario are the lowest, while the peak runoff is the highest and occurs at the latest time. Additionally, the water accumulation range is the widest. This phenomenon arises because, under the maximum subsidence scenario, the high closure degree of depressions within the watershed forms a natural “Reservoir,” resulting in a larger catchment area compared to other scenarios. Simultaneously, surface subsidence damages the geological structure, leading to a high degree of fragmentation in water patches and obstructed drainage pathways [67]. Studies by scholars such as Zhang et al. [6] and Ying et al. [68] have also highlighted that subsidence areas possess a certain water storage capacity. During precipitation events, upstream inflows are initially stored within the subsidence zones, reducing surface runoff. Runoff only forms when the storage reaches its upper limit and accumulated water begins to overflow [69]. Consequently, the water accumulation area is extensive, the average runoff is relatively low, and the peak occurrence time is delayed. However, in this study, the peak runoff data for the maximum subsidence scenario are found to be the largest compared to other scenarios, whereas the peak runoff data for the baseline scenario are the smallest. This suggests that varying degrees of subsidence may exert distinct influences on the hydrological characteristics of the watershed. Studies conducted by scholars such as Ma et al. [70] and Zhang et al. [6] indicate that subsidence areas possess the capacity to mitigate flood peaks, a finding that aligns with the flood peak characteristics observed in the baseline scenario of this study. However, the results of the maximum subsidence scenario in this study deviate from those reported in prior studies. This discrepancy may stem from the relatively small scale of the watershed examined in this study. In the maximum subsidence scenario, the proportion of the subsidence area is substantial, leading to pronounced terrain undulations. Subsidence depressions intercept and reconfigure the primary flow paths, resulting in a pronounced concentrated flow effect [5]. Additionally, the water system network within the watershed exhibits a high degree of fragmentation, and the proximity of the subsidence area to the outlet exacerbates this phenomenon. When the water storage capacity reaches its upper limit, water exits the depressions in the form of slope flow, generating pulsed flood peaks [71]. As a result, the peak runoff of the maximum subsidence scenario reached 192.3 m³/s, exceeding values observed in other scenarios. In the planning scenario, the simultaneous enhancement of land reclamation rates and ecological restoration levels leads to a reduction in the waterlogged area within coal mining subsidence areas. Concurrently, this scenario exhibited the smallest cultivated land reduction (5.17 km²) compared to 2022 levels. The construction land and the forest coverage registered at 39.66 km² and 9.48 km² respectively—higher than alternative scenarios. Relative to the baseline scenario, the planning scenario demonstrated greater mean runoff, cumulative runoff, and peak runoff at 3.798 m³/s, 5918.628 km³, and 185.052 m³/s, respectively. This is primarily attributed to the expansion of construction land in certain areas, which enhances runoff generation [72,73,74]. Additionally, the coal mining subsidence area has been effectively managed, reducing the subsidence extent and consequently diminishing its water storage capacity. The water patches within the watershed remain relatively intact, and the connectivity of the water network is strong, facilitating efficient drainage [75]. Although the planning scenario helps maintain the hydrological characteristics of the watershed to a certain extent, it also suggests that the confluence process occurs more rapidly.

By comparing the hydrological response characteristics across different scenarios, it was found that the impact of varying degrees of subsidence on the hydrological features of the watershed exhibits dual characteristics. When the degree of subsidence is moderate, the subsidence area stores water resources and regulates peak floods. Conversely, when the degree of subsidence is high and the proportion of subsidence is large, the watershed’s flood peak tends to increase sharply. The spatial distribution and scale of the subsidence area are relatively sensitive; the closer the subsidence area is to the watershed outlet and the larger its proportional area, the more pronounced its influence on runoff data becomes. In planning and managing the subsidence area, adjustments to the land use structure should be considered to mitigate potential soil erosion caused by planned development [76].

5.2. The Impact of Multi-Scenario Hydrological Simulation on Policies

Based on the results of multi-scenario hydrological simulations, this study provides a scientific foundation for territorial spatial planning and ecological governance in coal mining subsidence areas. The “Overall Territorial Spatial Planning of Huaibei City (2021–2035)” (hereafter referred to as the “Planning”) emphasizes strict adherence to ecological protection red lines, prioritizing the conservation of critical ecological functional areas such as wetlands and water sources, while also focusing on addressing water accumulation issues in coal mining subsidence zones [52]. Furthermore, both the “Planning” and the “Urban Flood Control Planning of Huaibei City (2022–2035)” prioritize water resource management and disaster prevention, advocating for enhanced water resource protection, optimized allocation of water resources, and improved flood control and drainage systems [51]. The range and intensity of water accumulation simulated in each scenario of this study can serve as a basis for delineating ecological protection red lines, managing water resources, and mitigating flood-related disasters. In future planning, it is recommended that high-intensity water accumulation zones within subsidence areas be designated as part of ecological control zones. Flood control infrastructure, including drainage channels and pumping stations, should be installed in these high-intensity water accumulation zones [77]. Additionally, buffer zone designs such as highly permeable pavement and vegetation transition belts should be implemented both within and around the subsidence areas [78]. Based on site-specific conditions, terrain reshaping and water system reconfiguration should be conducted to enhance water network connectivity and optimize the drainage system [79]. Secondly, the “Planning” highlights the optimization of the functional structure of territorial space, particularly focusing on the layout of construction land and cultivated land. It also emphasizes regional coordinated development and industrial transformation. Research findings indicate that the expansion of construction land has led to an increase in flood peak flow and water accumulation intensity to some extent. Conversely, the rational utilization of coal mining subsidence areas can play a critical role in water resource storage and flood peak reduction [6]. Therefore, this study recommends enhancing the optimization of various land use structures and promoting the resourceful utilization of subsidence areas [4]. Land use under different scenarios significantly influences hydrological processes. It is essential to strengthen regional collaborative governance by coordinating land use planning between subsidence areas and surrounding regions. Promoting infrastructure connectivity between subsidence areas and other regions can enhance the overall flood control capacity of the area [80]. Planning should be conducted based on varying degrees of subsidence; for stabilized water-accumulated areas, they can be designed as subsidence wetland parks. For deep water accumulation zones, developing fishery-photovoltaic complementary industries can achieve synergistic benefits in both ecological restoration and economic development [81].

The planning and design for coal mining subsidence areas should comprehensively take into account multiple factors such as ecological restoration, geological safety, hydrological regulation, disaster prevention and control, and economic and livelihood issues, in order to strike a balance between ecological protection and resource utilization and achieve sustainable development encompassing resource extraction, ecological restoration, and economic transformation.

5.3. Advantages of Model Integration and Limitations of the Study

The PLUS-MIKE 21 integrated model can effectively simulate the hydrological response characteristics under multiple scenarios and identify the spatial key points. Traditional numerical simulations using models such as SWAT have difficulties in depicting the changes in water accumulation at the spatial level. However, MIKE 21, with its two-dimensional free surface flow characteristics and flexible grid system, can simulate the changes in water accumulation under different scenarios [82]. Meanwhile, the PLUS model can achieve high simulation accuracy and adaptation to the complex evolution of multiple land types [83]. Integrating the MIKE 21 model with the PLUS model to carry out multi-scenario hydrological simulations in coal mining subsidence areas plays a crucial role in grasping the dynamic response mechanism of coal mining subsidence areas, multi-scenario collaborative analysis, and decision optimization. Additionally, factors such as the resolution of remote sensing imagery, projected coal mining subsidence outcomes, and model calibration exert a certain influence on the precision of hydrological process simulation results. In the parameter settings for MIKE21 model calibration, the roughness for different land use types can be further subdivided to improve the overall accuracy of the model. The interaction between surface water and groundwater has not yet been explored in the literature. In future studies, it can be attempted to couple with the groundwater model to achieve dynamic simulation of the interaction process between surface water and groundwater.

6. Conclusions

This study is based on the PLUS-MIKE 21 model, establishing three land use scenarios and conducting a comparative analysis. It investigates the hydrological response changes resulting from varying land use patterns in the Ba River Watershed by the year 2037. The primary conclusions are summarized as follows:

(1) The land use prediction data for each scenario indicate that cultivated land consistently constitutes the dominant land type in the Ba River Watershed, accounting for over 60%. Compared to the land use distribution in 2022, the baseline scenario shows a reduction of 10.31 km² in cultivated land. In the maximum subsidence scenario, water areas expand most significantly, increasing by a total of 7.9 km². Under the planning scenario, through measures such as land reclamation and ecological restoration, the water area decreases by 5.91 km² while the construction land increases by 8.94 km². (2) In the maximum subsidence scenario, high-intensity water accumulation zones occupy the largest proportion, accounting for 31.35% of the total watershed, with water accumulation primarily concentrated in coal mining subsidence areas. Conversely, under the planning scenario, these zones account for the smallest proportion at 19.10% of the total watershed. Notably, water accumulation issues are most pronounced in coal mining subsidence areas and densely developed construction land. In contrast, cultivated land and forest land effectively mitigate runoff. (3) The PLUS-MIKE 21 integrated model effectively simulates the hydrological response characteristics induced by land use changes in the Ba River Watershed. In the maximum subsidence scenario, surface subsidence leads to a pronounced runoff storage effect, characterized by the smallest mean runoff volume, a notable delay in flood peaks, and the highest peak discharge (192.3 m³/s). Under the planning scenario, measures such as ecological restoration facilitate timely runoff discharge, resulting in the largest cumulative runoff volume and reduced flood peaks. Coal mining-induced subsidence significantly affects water accumulation and runoff distribution patterns within the Watershed. The hydrological response characteristics resulting from different degrees of subsidence are diverse: A relatively severe subsidence situation is prone to trigger dam-break flood peak risks under extreme rainfall. (4) In the planning scenario, the implementation of land reclamation and ecological restoration contributes to maintaining the total runoff volume of the watershed and mitigating flood peaks to a certain extent. However, the rapid confluence process increases the risk of soil erosion. Addressing the governance challenges in high-phreatic coal mining subsidence areas is complex and requires integrated planning that considers hydrological response characteristics, socio-economic factors, and ecological impacts to achieve sustainable co-development between subsidence area management and stable hydrological conditions.