Simulating the Coupling of Rural Settlement Expansion and Population Growth in Deqing, Zhejiang Province, Based on MCCA Modeling

Abstract

Analyzing the relationship between rural settlements and rural population change under different policy scenarios is key in the sustainable development of China’s urban and rural areas. We proposed a framework that comprised the mixed land use structure simulation (MCCA) model and the human–land coupling development model to assess the spatiotemporal dynamic changes in rural settlements and its’ coupling relationship with the rural population in the economically developed region of Deqing, Zhejiang Province. The results showed that rural settlements and urban land increased by 14.36 and 29.07 km², respectively, over the last 20 years. The expansion of some rural settlements and urban land occurred at the cost of cropland occupation. Rural settlements showed an expansion trend from 2000 to 2020, increasing from 42.69 km² in 2000 to 57.05 km² in 2020. In 2035, under the natural development scenario, the cropland protection scenario, and the rural development scenario, rural settlements are projected to show an expansion trend and Wukang and Leidian are the key regions with rural settlement expansion. The distance to Hangzhou, nighttime light data, distance to rivers, and precipitation are important factors influencing the expansion of rural settlements. The coupling relationship between rural settlements and the rural population developed in a coordinated manner from 2000 to 2020. For 2035, under different scenarios, the coupling relationship between rural settlements and the rural population showed different trends. In the rural development scenario, the highest number of towns with coordinated development between rural settlements and the rural population is in Deqing, predominantly with Type I coupling. Overall, an important recommendation from this study is that the sustainable development of regional land use can be promoted by controlling the occupation of cropland for urban and rural construction, balancing rural settlement expansion and rural population growth, and formulating land use policies that are more suitable for rural development.

1. Introduction

The urbanization of the rural population is an important trend in the development of countries worldwide. China has experienced rapid urban expansion [1] and is one of the countries with the highest urbanization rate. According to the seventh national census, China’s urban population increased by 236.42 million, and the rural population decreased by 164.36 million from 2010 to 2020. In theoretical terms, the spatial area of rural settlements should shrink as the rural population decreases. However, the spatial area of rural settlements is increasing over time [2,3], mainly due to the need for space in terms of the living environments and economic investment activities of farmers [4]. The decrease in the rural population and the expansion of rural settlements have led to the disuse of rural settlements [5]. Rural settlements support production and living for the rural population and are an important part of rural land use. Due to the ineffective use of disused rural settlements, the production and living needs of the growing rural population can only be accommodated through the expansion of rural settlements. The expansion of rural settlements may be at the cost of cropland reduction [6,7,8,9]. This unsustainable degradation in the structure of rural settlements degrades the quality of cropland and exacerbates cropland resource shortages, which cause further food crises [10,11] in China. Coordinating the balance between urban construction, rural development, and food security is critical for China, which lacks sufficient reserve land resources. Therefore, it is important to assess the coupling relationship between rural settlement expansion and rural population growth.

Extensive research has been conducted on the coupling relationship of urban expansion [12,13] and urban population growth [14,15], urban–rural construction land transition and population flow [16], rural settlement expansion, and rural population growth [17]. In terms of research scales for analyzing the coupling relationship between rural settlements and rural populations, most of these studies have been undertaken at the macro level [3,18,19,20], including the national, provincial, and municipal (county) regional scales, while insufficient research has been conducted at the micro-scale including towns. The decoupling theory [21] and decoupling models [22,23] have been proposed as research tools for exploring the relationship between population and land change and have been widely used. There is currently no relevant theoretical or policy basis for the delineation of the study period. Current works are often based on a single time scale to explore the coupling relationship between rural settlements and rural populations in a certain time period, which means that there is a lack of clarity in determining the appropriate divisions for research time periods [24].

Research has been undertaken on the prediction of the coupling relationship between rural settlements and the rural population. For example, Zhang used the decoupling index and constructed a gray prediction model to predict the changing trend from 2015 to 2024 [22]. These studies have some advantages in simulating the quantitative change trends for future rural human–land coupling relationships, but the simulation results are not sufficiently visualized in space. Land use simulation models are widely used to simulate the spatial distribution of land use and cover change (LUCC) with the advantage of spatial expressions, such as the cellular automata (CA)–Markov model [25,26], the ANN–CA model [27,28], the CLUS-S model [29], the FLUS model [30], and the PLUS model [31]. Different models are applicable to different research scenarios. CA models are the basis for establishing other models [32,33,34], which are applicable to large-scale macroscopic simulations. CLUE-S is predominantly applied to large-scale macroscopic studies and is suitable for finding the hotspots of land use change [35]. The FLUS model is mainly applied in the simulation and prediction of urban planning [36,37]. Compared with other models, PLUS models can be used to combine future predictor variables and calculate the development potential for each land use type using the random forest (RF) algorithm, which can simulate changes in the land use distribution more accurately [38]. However, PLUS is still a traditional pure cell (CA) land use structure simulation model [39], which is suitable for simulating land use changes in large-scale study areas at the macro level. It has some limitations for simulating land use structure changes in mixed cells at the micro-level. Liang proposed a CA model based on the hybrid mixed-cell CA model (MCCA) [40], which can be used to simulate subtle changes in different land use composition structures within the hybrid cell and is suitable for simulation studies in rural settlements [41].

A pattern of synergistic increase and decrease in the rural population and rural settlements has not yet emerged in China [2,42]. The contradiction between rural settlements and the rural population in the plains and economically developed areas has become more prominent [43]. To cope with the difficulties of disused homesteads and inefficient rural land use caused by conventional urbanization, in 2015, China proposed a policy of rural homestead system reform [44,45]. Deqing is a county in Zhejiang Province and is one of the pilot areas of China’s rural homestead system reform. The implementation of this policy will have a considerable impact on the spatial layout of rural settlements in this region. In order to solve the land–food–population conflict in rural areas, in this paper, we use the logical chain of “policy guidance–factor-driven rural settlement change–simulation prediction” to find the optimal path for rural development. In this paper, we aim to analyze the spatial and temporal evolution of land use in Deqing from 2000 to 2020, which can be used to infer the human–land relationship, geographical differentiation, and the spatial structural characteristics of the process of reforming the residential base system in the county. A new hybrid land use structural–temporal dynamic change modeling method, the hybrid MCCA, is proposed to simulate the spatial pattern of land use for Deqing in 2035. A grid-scale LUCC model is constructed to show the evolutionary characteristics of the spatial and temporal patterns of rural settlements under different development scenarios by tracing the past (2000–2015), expressing the current situation (2015–2020), and predicting the future (2035). Another coupled model is established at the town scale to analyze historical rural human–land correlations and depict the future rural human–land correlations under different policy scenarios. This study has considerable implications for the coordinated development of regional people–land, arable land conservation, and urban–rural development.

2. Materials and Methods

2.1. Study Area

Deqing is a county in Zhejiang Province that is located between latitude 30°261′–30°421′ N and longitude 119°451′–120°211′ E. It is in the western part of the Hang Zhou-Jia Xing-Hu Plain in the Yangtze River Delta and is surrounded by large- and medium-sized cities such as Hangzhou and Shanghai, encompassing a total area of approximately 936 km². Its topography slopes from west to east [46]. The climate of Deqing is warm and humid and forms part of the subtropical humid monsoon zone, with four distinct seasons, an annual average temperature of approximately 13–17 °C, and an annual average precipitation of 1379 mm. The county has 11 towns [47]: Moganshan, Yuyue, Xinan, Xinshi, Luoshe, Leidian, Qianyuan, Wukang, Sanhe, Zhongguang, and Fatou (Figure 1). The total land area of the county is 936 km². In 2020, the area for rural settlements is 57.05 km², accounting for 6.55% of the total. In addition, according to the Huzhou Statistical Yearbook, the total population is 443,200 in Deqing, with 267,800 people registered in rural areas. The land size of the rural settlements per capita is 213.03 m², far exceeding the upper limit of 146 m² per capita stipulated in the General Land Use Plan, Deqing (2006–2020). Deqing County has experienced two rounds of pilot reforms for the rural homestead system, one started in March 2015, and the other started in September 2020 [48]. Since 2015, Deqing has been innovating and exploring rural land use policies for promoting the clustering and development of rural settlements. Meanwhile, combining its geographical location advantages for attracting population inflow. This has eased the contradiction between the supply and demand of people and land.

2.2. Data Preparation

Four types of data were used in this study: land use cover data, driving factor data, restricted area data, and rural population statistical data (

Table 1. Research datasets used in this study.

Category	Data	Year 1	Original Resolution	Data Resource
Land use cover data	Land use cover data	2000, 2005, 2010, 2015, and 2020	30 m	https://www.resdc.cn/data.aspx (accessed on 2 March 2022)
Driving factors	POP	2020	100 m	https://www.worldpop.org/datacatalog/ (accessed on 5 April 2022)
	Temperature	2015	1000 m	https://www.resdc.cn/data.aspx (accessed on 2 March 2022)
	Precipitation
	Distance to center of rural settlements	2015	/	https://www.webmap.cn (accessed on 5 April 2022)
	Distance to roads	/	/	https://www.openstreetmap.org/ (accessed on 10 February 2022)
	Distance to railways
	Distance to river	/	/	https://www.resdc.cn/data.aspx (accessed on 2 March 2022)
	Distance to Hangzhou	/	/	https://map.baidu.com/ (accessed on 10 April 2022)
	Distance to center of county
	Distance to living facilities
	Distance to schools
	Distance to government and social groups
	Distance to factories
	Distance to medical service site
	Elevation	2015	30 m	http://www.gscloud.cn/ (accessed on 2 April 2022)
	Slope
	Aspect
	Night light data	2020	500 m	https://doi.org/10.7910/DVN/YGIVCD (accessed on 10 April 2022)
Constraint data	The boundary of restricted area	2015	/	https://www.deqing.gov.cn (accessed on 10 April 2022)
Rural population statistical data	Rural population	2000, 2005, 2015, and 2020	/	Huzhou Statistical Yearbook, and Census data

¹ Driving factors collected from different periods were approved, and we made the periods for the driving factors as recent as possible.

). The land use data were obtained from the Chinese Academy of Sciences Resource Environment Data Center (http://www.resdc.cn/ accessed on 6 August 2022). Chinese land use remote sensing monitoring data with a spatial resolution of 30 × 30 m were used to reclassify land use types into seven categories: cropland, woodland, grassland, waterbody, urban land, rural settlements, and other land using ArcGIS 10.2. There were 18 driver datasets, including the DEM data, meteorological data, point-of-interest data, night light data, and road data. Restricted area boundaries as the constraint data were collected from the Deqing government departments. The rural population statistical data are from the Huzhou Statistical Yearbook and the Census Data for 2000, 2005, 2010, 2015, and 2020. All the remote sensing datasets were unified with the coordinate system CS_Krasovsky_1940 and were resampled to a resolution of 250 × 250 m.

2.3. Research Framework

The research framework of this paper comprises two parts: the MCCA model for simulating the LUCC data and a coupling model for estimating the relationship between rural settlements and population (Figure 2). We used the MCCA model to simulate the distribution of rural settlements in Deqing in 2020 and 2035 based on elevation, GDP, precipitation, temperature, and POI data under different scenarios. The coupling model was used to assess the decoupling of the spatial distribution of rural settlements and the rural population in different years and under different scenarios.

2.4. Method

2.4.1. MCCA Model

(1)

Mining of quantitative transition rules

Combined with the situation of the study area and previous research results [49,50], 18 natural and socioeconomic factors were selected as the driving factors for the data (Figure 3). Table 1 describes the driving factors, including natural, climatic, socioeconomic, and environmental factors. We used RF regression (RFR) to examine the relationships between the changes in land use components and their driving factors. We also used the out-of-bag root-mean-square error (OOB RMSE) to evaluate the fitting precision of the RFR. The RFR has the advantage of being able to measure the contribution of driving factors to the variation of each structural change in land use [51]. As shown in

Table 2. Parameter settings of the MCCA model.

Parameters				Cropland	Woodland	Grassland	Waterbody	Urban Land	Rural Settlements	Other
RFR	Number of regression trees		100
	Sampling rate		0.1
	Accuracy index	OOB-RMSE		0.014	0.009	0.002	0.008	0.044	0.060	0.043
Land use type simulation	Parameter	Neighborhood	3 × 3
		Step size	1	1	1	1	1	1	1	1

, the sampling rate was set as 10%, and 100 regression trees were used to construct each RFR model. The OOB RMSE for all the land use components was less than 0.05, indicating that the RFR was well-trained and capable of capturing the relationships between the LUCC and the associated driving factors.

(2)

Land use demands and land use transfer matrix rules

For the model calibration and prediction, the historical (2000–2020) and future land use demands (2035) were prepared. We used the method for linear regression [52], which can project future land use demands for a specific period while outputting the historical land use amounts.

Table 3. Area ratio of different land use types in Deqing in 2000 and 2020 and the area demand ratio for 2035 under different scenarios (%).

Year	2000	2020	2035
			NDS	CPS	RDS
Cropland	50.30	43.26	37.71	39.71	37.8
Woodland	37.55	36.28	35.37	36.37	34.37
Grassland	0.63	0.65	0.66	1.66	0.76
Waterbody	4.93	6.62	8.04	8.04	8.04
Urban land	1.55	4.64	7.34	6.34	9.34
Rural settlements	4.92	6.56	7.44	6.44	7.35
Other	0.12	1.99	3.44	1.44	2.34

shows the land use demands. We used the land use transfer matrix [53] to mine the land use type conversion relationship, as shown in Equation (1).

$$T_{ij}=\left[\begin{array}{ccc}{T}_{11}& \dots & T_{1n}\\ ⋮& ⋮& ⋮\\ T_{n1}& \dots & T_{nn}\end{array}\right]$$

(1)

where n is the number of land use types; i and j (i, j = 1, 2,…, n) are the types at the beginning and the end of the study period, respectively; and T_ij is the area for land use type i converted to type j.

We normalized the coefficients of the land use transfer matrix for 2000–2020. We then obtained the land use transfer matrix rules for simulating the land use distribution in 2020 and for the natural growth scenario in 2035.

$$\mathrm{T}=\left[\begin{array}{ccc}1& \cdots & 1\\ ⋮& ⋮& ⋮\\ 1& \cdots & 0\end{array}\right]$$

(2)

where 1 means that the two land types can convert to each other, and 0 means that the two land types cannot convert to each other.

(3)

Accuracy validation of simulation

To evaluate the accuracy of the MCCA model, we used OA, Kappa, and relative entropy (RE) to evaluate the overall accuracy, change accuracy, and structural accuracy, respectively. OA represents the number of samples simulated correctly as a percentage of all the actual samples and the equations for Kappa can be expressed as:

$$Kappa=\frac{O{A}_0-O{A}_e}{1-O{A}_e}$$

(3)

where OA₀ is the overall accuracy, and OA_e is the probability of the consistency of the predictions due to the probability of the consistency of the predicted results due to chance.

The equations for RE can be expressed as:

$$R{E}_i={{\displaystyle \sum }}_{k=1}^k{R}_i\left(k\right)\log \left[\frac{R_i\left(k\right)}{S_i\left(k\right)}\right]$$

(4)

$$R{E}_{mean}={{\displaystyle \sum }}_{k=1}^{M}R{E}_i\left(k\right)/M$$

(5)

where RE_i is the relative entropy of the actual and simulated land use structure of the hybrid unit i, with a higher value indicating a greater difference between the actual and the simulated land use structure; R_i and S_i are the actual and simulated land use structures, respectively; k is the land use structure, k = 1, 2,…, K; M is the total number of mixed cells; and RE_mean is a measure of similarity in the land use structure at the regional level.

(4)

Different land use scenario settings

To evaluate the regional spatial layout for rural settlements affected by different land use scenarios, we designed the natural development scenario (NDS), the cropland protection scenario (CPS), and the rural development scenario (RDS). Under the NDS, we continued to use the transition matrix rules for 2000–2020 and predicted the land use demand for 2035 by using linear regression. The open water category (a subcategory of waterbody) was not allowed to convert to the other land use types [41].

Referring to the requirements of the Deqing Land Master Plan (2006–2020), the occupation of cropland for construction is strictly controlled, the balance of cropland occupation and replenishment is implemented while ensuring that the quantity of cropland is not reduced, and the quality is improved. Under the CPS, the conversion of cropland to other land is limited, and there is an appropriate increase in the area of cropland demand for 2035.

In 2020, Deqing implemented a five-year policy of rural homestead system reform. During this period, Deqing introduced a series of rural homestead management regulations, such as the linkage between urban and rural construction land and the withdrawal of unused homestead bases. This affects changes in the land use structure. Under the RDS, we set rural settlements to be able to be converted into cropland and rural settlements into grassland. We appropriately reduced the area demand for rural settlements and slightly increased the area demand for cropland and grassland for 2035 (Table 3).

2.4.2. Rural Population Prediction

This study focuses on simulating the impact on the coupling relationship between rural settlements and the rural population under different LUCC scenarios, so we assumed that the rural population change is less affected by the policy. Based on the available data, we used the combined growth rate method [54] to predict the spatial distribution of the rural population at the town scale for 2035. We backtracked the historical average growth rate over 15 years (2005–2020) and used it as a composite growth rate to predict the rural population size in the study area for 2035. The equations for the combined growth rate can be expressed as:

$$P_n=P_0\times {\left(1+k\right)}^n$$

(6)

where P₀ represents the rural population in 2005, P_n is the rural population in 2035, k is the combined rural population growth rate, and n represents the number of years.

2.4.3. Coupling Development Model

To depict the rural settlement–population relationship in the process of rural development, the elasticity coefficient was used to partition the types of coupling development between rural settlement expansion and population growth [55,56]. The elasticity coefficient can be expressed as:

$$\mathsf{\lambda }=\frac{RPD}{RSL}$$

(7)

where λ is the elasticity coefficient between rural settlement expansion and population growth. RPD is the average annual change rate for the rural population, whereas RSL is the average annual change rate for the rural settlements.

According to the increase and decrease changes for RPD and RSL and the numerical comparison, the coupling relationship between the rural settlements and the rural population can be classified into six types. Taking Type I as an example, RPD and RSL are positive at the same time, and the RPD is greater. This indicates that the rural population and rural settlements increase at the same time and that the growth of the rural population is faster than that of the rural settlements. The six coupling types can be divided into two categories of coordinated and uncoordinated development. The detailed information is shown in

Table 4. Types of coupling relationship between the rural settlements and the rural population.

Type	RPD	RSL	λ	Coordination Relationship
I	>0	>0	>1	Coordinated development
II	>0	<0	—
III	<0	<0	<1
IV	<0	<0	>1	Uncoordinated development
V	<0	>0	—
VI	>0	>0	<1

and Figure 4.

3. Results

3.1. Land Use Transfer Matrix during 2000–2020

As shown in the land use transfer matrix in Deqing for 2000–2020 (

Table 5. Land use transfer matrix for 2000–2020 (km²).

Change to	1	2	3	4	5	6	7	Transfer out
1	396.04	3.59	0.19	18.54	26.05	18.08	9.24	75.69
2	5.00	336.88	0.43	0.15	2.10	0.48	8.16	16.32
3	0.07	0.39	5.42	0.00	0.00	0.00	0.01	0.47
4	2.56	0.16	0.00	42.89	0.17	0.22	0.05	3.16
5	0.19	0.09	0.00	0.06	14.18	0.02	0.00	0.36
6	2.87	0.11	0.01	0.25	1.07	41.55	0.13	4.44
7	0.03	0.05	0.01	0.00	0.04	0.00	0.99	0.13
Transfer in	10.72	4.39	0.64	19.00	29.43	18.80	17.59	-

1—Cropland, 2—Woodland, 3—Grassland, 4—Waterbody, 5—Urban land, 6—Rural settlements, and 7—Others.

), there were frequent area transformations between land use types during this period. There were pronounced transformations between urban land and cropland, rural settlements, and cropland. There was a substantial area of cropland transferred out, encompassing an area of 75.69 km². The area of cropland converted to urban land was the largest in 2000–2020, with an area of 26.05 km². The area of cropland converted to rural settlements was the second highest in 2000–2020, encompassing an area of 18.08 km². Rural settlements and urban land increased by 14.36 and 29.07 km², respectively, over the last 20 years, predominantly from cropland transfer. The cropland area converted to urban land and rural settlements are mainly distributed in the center of Deqing, such as in Wukang (Figure 5). Curbing the expansion of cropland consumed by urban land and rural settlements is critical for protecting the total cropland area. Grassland and waterbodies increased by 0.17 and 15.84 km², which is mainly due to the transfer of cropland. Woodland areas had no pronounced change during 2000–2020, and the conversion of cropland and grassland to woodland supplemented the loss of woodland. This indicates that there is a risk of cropland loss and a reduction in woodland and grassland, which will threaten food security and the ecological environment in Deqing. To cope with these challenges, Deqing should undertake sustainable planning for the urban and rural land layout, actively strengthen the protection of cropland, and promote planning and protection for woodland and grassland. Based on the analysis of the land use transfer matrix in Deqing during 2000–2020, the coefficients of the land use transfer matrix were modified, and the land use transfer matrix rules were calculated, which form the foundation for constructing the MCCA model.

3.2. Simulation and Prediction Results of the Land Use Structure

3.2.1. Accuracy Verification of the Simulation Results for 2020

The consistency between the simulated and actual land use patterns in 2020 was assessed based on four indices: OA, Kappa, mcFoM, and RE. As shown in

Table 6. Accuracy indices of the mixed-cell CA model for the experiment.

Indices	1	2	3	4	5	6	7
Use accuracy	0.9291	0.9762	0.9862	0.8689	0.8651	0.8361	0.5430
Producer’s accuracy	0.9331	0.9789	0.8995	0.8840	0.7951	0.8416	0.5956
OA	0.9265
Kappa	0.8921
mcFoM	0.4643
Mean RE	0.2799

, the values for OA, Kappa, mcFoM, and the mean RE were 0.9265, 0.8921, 0.4643, and 0.2799, respectively (Table 6). This indicates that the accuracy of the MCCA model met the simulation requirements. The use accuracy and producer’s accuracy for each land use type also met the application requirements.

3.2.2. Prediction of LUCC under Different Scenarios for 2035

Based on Section 3.2.1, the MCCA model was used to simulate the land use layout under different scenarios for 2035. The RGB images generated using different combinations of land use types were used to show the simulation results. The LUCC layout under different scenarios for 2035 showed different trends (Figure 6). In the NDS, the area of other types of land, urban land, rural settlements, and waterbodies all showed an increasing trend. Among them, other types of land had the highest increase of 72.86%, followed by urban land, waterbodies, and rural settlements at 58.19%, 21.45%, and 10.43%, respectively. In the CPS, the reduced cropland area was less than the NDS, and the area of woodland, grassland, water, and urban land showed an increasing trend. Meanwhile, the growth rate for urban land in the CPS (36.64%) was significantly less than that in the NDS (58.19%), and rural settlements showed a significant decrease (−1.46%). In the RDS, the area for grassland and woodland did not significantly change but the areas of urban land and rural settlements significantly increased by 21.83% and 7.78%, respectively. The results showed that the area of urban land increased under the three different scenarios (NDS > CPS > RDS). This indicates that economic development in the central region of Deqing is ensured. The urban land area and rural settlements in the NDS significantly increased, and the increased area of rural settlements was lower in the RDS. In the CPS, the rates of increase in the area for woodland and grassland were significantly higher than those in the NDS and RDS, and the decrease area of cropland was lower.

3.3. Spatiotemporal Characteristics of Rural Settlement Use Expansion and Rural Population Growth

3.3.1. Time-Series Changes in Rural Settlements and the Rural Population

Regions with larger rural populations generally showed a larger area for rural settlements. From 2000 to 2015, Deqing’s rural settlements continued to increase, but the rural population showed a decreasing trend (Figure 7b). From 2015 to 2020, the change in the rural settlement area showed a growth trend, and the area of rural settlements increased from 49.53 km² to 57.05 km². There was a significant increase in the rural population from 2015 to 2020 (26.50 × 10⁴ people in 2015 and 35.56 × 10⁴ people in 2020, respectively). Rural settlements showed a parallel growth trend with the rural population during 2015–2020 (Figure 7). During 2020–2035, the rural population showed an increasing trend, and the rural population increased from 35.56 × 10⁴ people to 43.54 × 10⁴ people. In the NDS and RDS, the areas for rural settlements increased from 57.05 to 63.01 km² and from 57.05 to 61.47 km², respectively. In 2035, compared with the NDS, the rate of the expansion of rural settlements is estimated to decrease in the RDS (Figure 7a). Rural settlements and the rural population showed a parallel growth trend during 2020–2035 (NDS and RDS). The changes in rural settlements and the rural population showed a different trend in the CPS. The area for rural settlements decreased from 57.05 to 55.57 km² (Figure 7a), while the change in the rural population showed a growth trend.

3.3.2. Spatial Distribution of Rural Settlement Use Expansion

To explore settlement use expansion in Deqing, we calculated the growth rates for rural settlements during 2000–2035 (NDS, CPS, and RDS) in Deqing (

Table 7. Rural settlement change during 2000–2035 (NDS, CPS, and RDS) (km², %).

Region		2000–2015	2015–2020	2020–2035 NDS	2020–2035 CPS	2020–2035 RDS
County	Deqing	6.85 (16.04%)	7.53 (15.20%)	5.95 (10.43%)	−1.46 (−2.56%)	4.44 (7.78%)
11 Towns	Wukang	1.65 (52.13%)	3.63 (75.46%)	3.97 (47.13%)	2.98 (35.29%)	2.42 (28.75%)
	Xinshi	0.21 (2.53%)	1.35 (15.76%)	1.03 (10.47%)	1.21 (12.21%)	0.67 (6.75%)
	Leidian	3.02 (56.54%)	1.47 (17.61%)	1.52 (15.49%)	0.06 (0.57%)	1.78 (18.12%)
	Qianyuan	0.06 (1.49%)	0.87 (20.37%)	0.40 (7.72%)	−0.32 (−5.98%)	0.02 (0.31%)
	Luoshe	−0.10 (−3.92%)	−0.002 (−0.08%)	−0.20 (−8.28%)	−0.37 (−16.00%)	−0.97 (−40.97%)
	Zhongguan	0.58 (6.95%)	0.16 (1.78%)	−0.15 (−1.70%)	−2.14 (−23.73%)	−0.91 (−10.04%)
	Yuyue	0.57 (15.93%)	0.02 (0.46%)	0.25 (6.13%)	−0.96 (−23.27%)	0.78 (18.85%)
	Xinan	0.46 (15.93%)	−0.02 (−0.40%)	−0.67 (−12.22%)	−1.31 (−23.88%)	0.36 (6.52%)
	Moganshan	0.03 (12.39%)	−0.001 (−3.49%)	0.01 (5.03%)	−0.05 (−23.68%)	0.05 (21.13%)
	Sanhe	0.31 (18.35%)	0.05 (2.33%)	−0.04 (−2.12%)	−0.44 (−21.62%)	0.09 (4.26%)
	Fatou	0.06 (14.29%)	0.004 (0.81%)	−0.17 (−35.89%)	−0.12 (−25.20%)	0.15 (31.63%)

). The hotspot analysis tool (Getis-Ord Gi*) in ArcGIS was used to calculate the expansion hotspot index (Zi*) at the grid cell (250 × 250 m) to obtain the spatial distribution of hot–cold spots for rural settlements in Deqing (Figure 8). The region with a higher local expansion hotspot index represents the hotspot areas in rural settlements with use expansion (high-value agglomeration areas), while the region with a lower local expansion cold-spot index represents the cold-spot areas for rural settlement use expansion (low-value agglomeration areas). During 2000–2020, Deqing’s rural settlements showed an expansion use trend. In the earlier sub-period (2000–2015), the total rural settlement increased by 6.85 km², and the growth rate was 16.04%. Among them, 11 towns had increasing trends, and the growth in rural settlements predominantly occurred in Wukang (1.65 km², 52.13%) and Leidian (3.02 km², 56.54%). During the later period (2015–2020), the area for rural settlements continued to increase, the growth area increased to 7.53 km², and the growth rate for rural settlements area decreased to 15.20%. Nine towns showed increasing trends, and the growth in rural settlements predominantly occurred in Wukang (3.63 km², 75.46%), Leidian (1.47 km², 17.61%), and Xinshi (1.35 km², 15.76%). The hotspot clustering for rural settlements was pronounced during 2000–2015 and 2015–2020 and the high-value clustering areas were concentrated in Wukang and Leidian (Figure 8a–c).

Under the different scenarios for 2035, we focused on regulating the changes in the transfer matrix rules for the three types of land: cropland, urban land, and rural settlements. For 2035 (NDS), we continued to use the land use transfer rules from 2000 to 2020, and the expansion of rural settlements was found to be dominated by encroachment into cropland. In 2035 (NDS), the increased area of rural settlements is estimated to be mainly distributed in the towns of Wukang, Leidian, and Xinshi (Figure 8d), in line with the findings for 2000–2020. For 2035 (CPS), the amount of cropland converted to rural settlements and other land use types was projected to be restricted, achieving successful protection for cropland. The decreasing areas for rural settlements would be distributed in Wukang, Leidian, and Xinshi, and the rural settlements’ cold-spot clustering region would encompass the towns in the eastern part of Deqing (Zhongguan, Xinshi, Leidian, and Yuyue) (Figure 8e). For 2035 (RDS), cropland that is converted to rural settlements and urban land was projected to be permitted, and rural settlements being converted to cropland and grassland would be approved. The areas for rural settlements in Wukang, Leidian, and Yuyue are estimated to significantly increase (Figure 8f). Compared with the NDS and CPS, the RDS shows a slower expansion use trend for rural settlements, and the spatial patterns for rural settlements would be more stable and more intensive in Deqing under this scenario.

3.3.3. Spatial Distribution of Rural Population Growth

From 2000 to 2015, there was a steady decrease in the rural population in Deqing, with a decrease of −5.50 × 10⁴ people and a decrease rate of −17.19% (

Table 8. Rural population change in Deqing during 2000–2035.

Year	2000	2015	2020	2035
Population growth (×104 people)	—	−5.50	9.14	7.90
Growth rate (%)	—	−17.19%	34.51%	22.16%
Growth type	—	Decrease in steady	Increase in activity	Increase in activity

). Among them, there was a decrease in the rural population of the two towns of Yuyue and Xinan with both decreasing in activity (Figure 9a). The rural population in seven towns, namely Moganshan, Luoshe, Leidian, Qianyuan, Wukang, Sanhe, and Zhongguan, all showed a steady decrease. Only the rural population in Fatou showed a stable increasing trend, and the rural population in Xinshi showed no significant change. During 2000–2015, Deqing accelerated the urbanization process, and a large number of farmers moved to the city to work, resulting in a decrease in the rural population. In both periods of 2015–2020 and 2020–2035, the rural population in Deqing actively increased, and the rural population increased by 9.14 × 10⁴ people and 7.90 × 10⁴ people, respectively, with an increase of 34.51% and 22.16%, respectively. From 2015 to 2020, six towns, namely Xinan, Leidian, Qianyuan, Wukang, Sanhe, and Zhongguang, showed an increase in their rural population. The rural population of four towns, namely Yuyue, Xinshi, Luoshe, and Fatou, steadily increased, and only the rural population in Moganshan decreased (Figure 9b). From 2020 to 2035, the rural population in nine towns increased. Among them, the four towns of Xinan, Wukang, Sanhe, and Zhongguang increased in activity. The rural population in five towns, namely Xinshi, Luoshe, Leidian, Qianyuan, and Fatou steadily increased (Figure 9c). From the analysis of the spatial characteristics between the rural settlements and the rural population, it was revealed that most of the areas with rural settlement use expansion (Figure 8a,b) concentrated in the region where the rural population decreased (Figure 9a) from 2000 to 2015. In both periods of 2015–2020 and 2020–2035, most of the areas with rural settlement expansion (Figure 8b–f) also showed increases in the rural population (Figure 9b,c), especially Wukang and Leidian, which are the economically developed regions of Deqing.

3.4. Spatiotemporal Characteristics of Coupling Development between Rural Settlements and the Rural Population

Table 9. Coupling relationship type in Deqing, 2000–2035.

Year	2000–2015	2015–2020	2020–2035
			NDS	CPS	RDS
Type (λ)	V (−1.26)	I (2.13)	I (2.02)	II (−7.69)	I (2.69)

shows the coupling relationship between the rural settlements and the rural population in Deqing. From 2000 to 2015, the coupling relationship showed that the rural population decreased, and the area for rural settlements increased, with a coupling relationship index of −1.26. Therefore, the type of coupling relationship between the rural settlements and the rural population was V in Deqing from 2000 to 2015, indicating uncoordinated development. From 2015 to 2020, the coupling relationship showed an increase in the rural population and the area for rural settlements, and the coupling relationship index was 2.13. The coupling relationship between the rural settlements and the rural population was I in Deqing from 2015 to 2020, indicating coordinated development. The coupling relationship between the rural settlements and the rural population tended to develop in Deqing in a coordinated manner from 2000 to 2020.

Under the different scenarios, from 2000 to 2035 (NDS), the coupling relationship showed that the rural population and area for rural settlements increased, with a coupling relationship index of 2.02 (I, coordinated development). From 2000 to 2035 (CPS), the coupling relationship showed that the rural population increased, and the area for rural settlements decreased, with a coupling relationship index of −7.69 (II, coordinated development). From 2000 to 2035 (RDS), the coupling relationship showed that the rural population increased, and the area for rural settlements increased, with a coupling relationship index of 2.69 (I, coordinated development). Under the different development scenarios for 2035, the trend for the coupling relationship between the rural settlements and the rural population is expected to develop in a coordinated manner in Deqing.

Figure 9 shows the spatial change in the coordination relationship between the rural population and the rural settlements at the town scale in Deqing in different periods and under different scenarios. From 2000 to 2015, 10 towns had a coupling relationship type that was an uncoordinated development (IV, V), accounting for 90.90% of land development in Deqing. Only the coupling relationship type for Fatou town was coordinated development (I) (Figure 10a). From 2015 to 2020, the type of coupling relationship in two towns was uncoordinated development (IV, V), accounting for 18.18% of land development in Deqing. Nine towns had coupling relationship types that were coordinated development (I, II), accounting for 81.82% of land development in Deqing (Figure 10b).

Under the different development scenarios for 2035, the coupling relationships in different towns showed different characteristics. From 2000 to 2035 (RDS), six towns in Deqing had coupling relationship types of coordinated development type (I, II), accounting for 54.55% of Deqing, while five showed uncoordinated development (V, VI), accounting for 45.45% (Figure 10c). From 2000 to 2035 (CPS), eight towns had coupling relationship types of coordinated development types (I, II, III), accounting for 72.73%, while three showed uncoordinated development (IV, VI), accounting for 27.27% (Figure 10d). From 2000 to 2035 (RDS), seven towns had coupling relationship types of coordinated development type (I, II), accounting for 63.64%, while four showed uncoordinated development (V, VI), accounting for 36.36%. Compared with the NDS and CPS, the RDS showed the largest number of towns with coordinated development between the rural settlements and the rural population in Deqing, with the coupling type predominantly being Type I. The key regions of Wukang and Leidian both showed coordinated development (I) (Figure 10e).

4. Discussion

4.1. Rural Settlement Use Expansion Pattern

The MCCA model can be used to calculate the land use mixture at the grid scale based on the simulated land use structure. We calculated the change in mixing rural settlements with other land use types and used it as a basis for discerning the settlement expansion patterns. The rural settlement spatial expansion patterns for the typical towns of Wukang and Leidian in Deqing from 2000 to 2035 (NDS, CPS, and RDS) are shown in Figure 11. From 2000 to 2020, Wukang and Leidian showed infill expansion patterns (Figure 11a) and from 2020 to 2035 (NDS, CPS, and RDS), marginal expansion patterns (Figure 11b–d). The other nine towns showed pronounced clustering of low-value cold spots in rural settlements (Figure 8) and did not have significant expansion characteristics. In the CPS, there were more cells with an increasing mixture than with a decreasing mixture in Xinshi (Figure 11c). This is because the cropland transfer to rural settlements was not approved, leading to a proportion of the cropland area being larger at the grid cells. Compared with the NDS and CPS, the spatial characteristics of the increased mixture cells showed more clustering in the RDS. This indicates that the trend in the expansion of rural settlements by other land use types being occupied is effectively curbed in the RDS, and the spatial clustering of rural settlements is clear (Figure 8f).

4.2. Impact of Various Driving Factors on Rural Settlement Use Expansion

The expansion of rural settlements is a result of a combination of population growth, natural factors, economic development, and climate change. Assessing the impact of different driving factors on rural settlement expansion helps us to better understand future changes in rural settlements. The importance ranking for the driving factors for the rural settlement expansion in 2020 is shown in Figure 12. We found that the distance to Hangzhou had the largest impact on rural settlement expansion and that the area for rural settlement expansion was mainly distributed in the area close to Hangzhou. During the development of Hangzhou’s counter-urbanization, there was a large number of urban population flows to Deqing [57]. Deqing is closely connected to Hangzhou city geographically and makes full use of the resources in Hangzhou’s reverse urbanization to promote its rural development [58]. The nighttime light data, the distance to rivers, and the average annual temperature were the main factors influencing rural settlement expansion (Figure 13). In recent years, Deqing has strengthened the construction of rural infrastructure. This includes the main road works with supporting works such as street light installation and roadside greening. The expansion of rural settlements is influenced more by the rural habitat environment. This is because the improvements in rural infrastructure have increased the happiness level of local people and attracted residents from outside the area, causing the demand for rural settlements to increase. Waterbodies promote rural settlement expansion, which is generally close to or surrounding water areas encroaching on cropland or other land use types. The pattern for rural settlement expansion is distinct from urban expansion [59], with rural settlements clustering closer to the watershed and with a smaller farming radius, which also confirms the transfer competition between rural settlements and cropland. Although there was no significant spatial distribution between the area of rural settlement expansion and the average annual precipitation, the average annual precipitation still contributed to the expansion of rural settlements.

4.3. Suggestions for Future Rural Settlement Development

The coupling relationship between rural settlements and the rural population in Deqing was dominantly uncoordinated from 2000 to 2015, which shows that with the continuous decline in the rural population, the area for rural settlements increased. However, following the implementation of the policy for rural homestead system reform, the coupling relationship between the rural settlements and the rural population tended to be coordinated from 2015 to 2020. During 2020–2035 (NDS, CPS, and RDS), compared with the NDS and CPS, in the RDS, rural settlements became more clustered and stable, and the coupling relationship between the rural settlements and the rural population became more coordinated. Therefore, the relevant government departments should actively promote a policy of rural homestead system reform to gradually reverse the dilemma of “people decreasing and land not decreasing” and make the relationship between rural settlements and the rural population more sustainable. There is a need for effective control of the occupation of cropland for urban and rural construction to balance the relationship between population growth, food supply, and rural development.

4.4. Future Enhancement of This Research

In this paper, since the variable selection of the MCCA model is more inclined to remote sensing data sources, the variables of farmers’ income, etc., were not considered enough. In addition, we were limited by the availability of the population data, and the rural population data were only available at the township level; thus, more attention should be paid to how to obtain more accurate and detailed rural population data and conduct coupling research. In further research, we will fully consider the use of the population density data for validation analysis and combine it with the regional population policy to obtain more rigorous research results. At the same time, we will continue to explore how to efficiently and quickly obtain historical rural settlement data [60] to obtain more refined rural settlement data, aiming to make the simulation results of rural land use structure more realistic and better simulate the development process of LUCC at a micro-scale.

5. Conclusions

In this research, based on the analysis of the land use transfer matrix in Deqing from 2000 to 2020 and the MCCA model, we simulated the spatial pattern for rural settlements in Deqing from 2020 to 2035 (NDS, CPS, and RDS). Combined with the coupling model, the coupling relationship between rural settlements and the rural population was simulated and analyzed for 2000–2015, 2015–2020, and 2020–2035 (NDS, CPS, and RDS). The results of the historical and future (under different scenarios) land use simulation indicated that the coupling relationship between rural settlements and the rural population varied. The change in land use transfer from 2000 to 2020 predominantly occurred between three land use types, namely cropland, urban land, and rural settlements. The cropland area being converted to urban land was the largest, and the cropland being converted to rural settlements was the second largest, representing an extensive loss of cropland. Rural settlements showed an expansion trend from 2000 to 2020 in Deqing. The area for rural settlements increased from 42.69 km² in 2000 to 57.05 km² in 2020. Further study found that the distance to Hangzhou, nighttime light data, distance to rivers, and precipitation contributed more to the expansion of rural settlements.

The rural settlements showed an expansion trend under the three scenarios for 2035. The key expansion region for rural settlements is Wukang and Leidian, which showed infill expansion changes in marginal expansion patterns between the periods of 2000–2020 and 2020–2035 (NDS, CPS, and RDS). In the RDS, the distribution of rural settlements was shown to be more stable and more intensive. The coupling relationship between the rural settlements and the rural population tended to develop in a coordinated manner from 2000 to 2020 in Deqing. Under the different development scenarios for 2035, the coupling relationship between the rural settlements and the rural population showed different trends. Compared with NDS and CPS, in the RDS, the largest number of towns (seven towns) would undergo coordinated development between the rural settlements and the rural population in Deqing, and the coupling type is predominantly Type I. These results not only provide new insights into the redistribution of rural land resources at the regional scale but also reveal supporting data to meet China’s rural revitalization goals.