Modelling and Forecast of Future Growth for Shandong’s Small Industrial Towns: A Scenario-Based Interactive Approach

Abstract

The industrial small-town development process in Shandong is influenced by the urban agglomeration strategy and the regional collaborative production, thereby resulting in a challenge of growth boundary planning. How to build a growth forecast decision support system to help small industrial towns maintain sustainable development with limited trial and error costs is an essential topic in the current research of small town-related fields. Empirical analysis reveals that the growth factors of small towns differ from the factors of cities due to the other-organization planning management system and self-organization construction activities that coexist in small towns. Besides, due to the size of small towns, the impact of policy changes in small towns is more significant than in cities. Furthermore, as part of the regional production chain, small industrial towns are most vulnerable to uncertain external disturbances. Therefore, it is necessary to formulate different development scenarios according to possible disturbances and output corresponding development forecasts. The research aims to build a decision-making support system for Shandong’s small-town planning based on an urban modeling approach using geographic information technology and scenario planning. Considering the mutually driving effects of the objective environment and subjective policies of Shandong’s industrial towns, as well as the corresponding dynamic mechanisms and comparing the theoretical basis and limitations of the different modeling approaches, this essay constructs a model system based on a mathematical model and a system dynamics model. It is also an interactive model accompanied by applicable rules and factors so that initial information and relevant development goals can be inputted into the model system to simulate the influence of different policies and identify the small industrial town growth scenarios.

1. Introduction

The rapid development of China’s urbanization has resulted in many problems and has become the main constraint restricting the further development of China’s urbanization. Due to the disequilibrium of the growth of urbanization, the development gap of China’s urbanization has increased. Resources and industries are highly concentrated in large cities, which has reduced the development space of small and medium-sized cities and towns, unbalanced urban and rural development, and created a lack of rural infrastructure and public services [1]. China’s government hopes to solve the issues with intra-regional and urban-rural development through macro policy regulation and control. The solution proposed by the government to address the imbalance of development is to implement the ‘urban agglomeration’ development strategy. Under the background of China’s urban agglomeration and urbanization development, the small-town unit has become the focus of the next phase of China’s urbanization. The small-town unit is not only an important node for promoting regional coordinated development but also provides employment and public services to rural areas. Therefore, smalltowns are the main way to reduce the difference between urban and rural development. The advantage of smalltowns in terms of spatial structure is that smalltowns can cooperate with central cities for industrial cooperation, absorb the surplus rural labor force, and help promote the rational distribution of industrial space layout and labor force distribution within the region. In terms of economic structure, the development of smalltowns can enable rural emigrants to have employment opportunities near their homes [2].

At present, the large-scale, long-term migration of rural migrants, excessive gaps between urban and rural development, and the urbanization of migrant workers has created problems that are difficult to solve by large cities alone [3]. Therefore, China urgently needs to give priority to the development of small-town units. While the urban agglomeration development strategy is an opportunity for the development of smalltowns, it also brings challenges to the development of smalltowns. Compared to polycentric development, China’s ‘urban agglomeration’ development strategy has stronger government guidance and is more like a visionary description [4]. Basically, the ‘urban agglomeration’ development strategy is like a regional industrial development guidance manual. China’s government wants to form a regional production chain by applying that strategy.

Urbanization and development in modern China have progressed through three stages: the slow progress stage before 1979, the rapid development stage from 1979 to 2009, and the quality improvement stage from 2009 to the present. In China’s past development models, cities have accumulated most of the population and resources; as a result, the imbalance between urban and rural development affected overall urbanization seriously [5]. Therefore, China’s government proposed the ‘urban agglomeration development strategy at the national level and planned a national and regional urban network system. At the same time, for the development of urban agglomeration, China has issued the ‘Characteristic Small Town Development Strategy’, which was designed for small towns [6]. ‘Small towns’ in China refer to ‘Xiang’ or ‘Zhen’ with a population of 20,000 to 100,000. In the 6 years since 2016, China has actively adjusted the development strategy of small towns, small towns have become increasingly prominent in regional coordinated development, and character development has become a critical research interest [7,8]. In the context of China’s new urbanization policy, Small towns have an important strategic role in the construction of urban agglomeration systems. As an intermediate link in the urbanization chain, small towns gained opportunities for accelerated development. Industries in large cities move to small towns [9], and the migration destination of the rural population also shifts from large cities to small towns [10]. In this process, studies on the growth forecast of small towns have gradually increased in the academic community, and many scholars have used research methods to analyze the factors of small-town development. At the beginning of the twenty-first century, scholars explored the influence of six factors, including natural, economic, institutional, scientific and technological information, humanities, and environment, on the expansion of small towns [11]. Based on China’s land management system, some scholars have analyzed the intrinsic motives of spatial expansion of towns and cities and concluded that resources and environmental conditions are the prerequisites for town density. They also concluded that transportation patterns, land development patterns, residential development patterns, and road traffic taxation systems are the key factors affecting the intensity of town development [12]. Based on previous research, researchers analyzed the strong correlation between socioeconomic factors, including urbanization rate, secondary industry employees, asset investment and construction land [13]. In addition to sociological research, traffic factors are also considered to be important factors affecting the expansion of small towns, scholars studied the expansion process of the developed area of Minghu town in Guangdong Province, and analysis suggested that the improvement of transportation conditions is one of the driving forces of town expansion based on the transportation network [14]. Finally, the terrain is also an important influence factor. Some scholars analyzed the suitability of the spatial expansion of construction land in Karst mountainous towns, which showed the significant influence of topography and geomorphology on town growth [15].

However, current research on the growth of small towns still faces three limitations. (1) Most quantitative studies on China’s small towns exclude the objective impact of policy changes and regional collaboration from the analysis of the small town’s development. Consequentially, the research findings lack adaptability [8,12,16,17,18]. (2) Current research is mostly based on a government-led planning system, which is limited to the complex rule-based modeling of town growth simulation [19]. This simulation lacks feedback and expression from the public. As a result, there is a lack of efficient interaction between researchers, planners, and service users. In addition, this type of research usually requires a large amount of data, and the cost of research and design is high [20]. (3) As part of the regional production chain, the economic and social development of small industrial towns will be significantly affected by changes in the supply and demand relationship of the market, such as the innovation of production technology, adjustment of production capacity in central cities, changes in energy and labor costs. Therefore, compared with small towns with agriculture, real estate, or tourism as the primary source of income, small industrial towns encounter external disturbances more frequently. In addition, it is more difficult for small industrial towns to adjust the structure of industries and land use. Therefore, small industrial towns are most vulnerable to uncertain external disturbances. Formulating different development scenarios according to possible disturbances and output corresponding development forecasts is necessary. Therefore, this essay attempts to introduce the modeling approach based on geographic information technology into the planning and management process of small industrial towns to support the decision-making process. This paper first discusses the classification and characteristics of typical growth prediction models. Based on the mutually driving effects of the objective environment and subjective policies of Shandong’s industrial towns and the corresponding dynamic mechanisms, the following section presents a growth model applicable to Shandong’s small towns. The last part reveals the growth factors under different policy backgrounds. The scenario planning theory is applied to enable the model to change its output according to the changes in policies and objectives. The research flowchart is presented(Figure 1).

2. Town Growth Modeling Approach

2.1. History of the Development of Town Growth Models

The modeling approach is an abstraction and generalization of the urban system, a quantitative description of the urban-rural conversion process and the spatial pattern of the built environment. The modeling approach can predict town growth, identify land boundaries and help designers make scientific land use decisions. Research on the modeling approach began in the 1950s. However, due to the large data requirements, redundant structure, and high R&D costs, the modeling research method was criticized by scholars represented by Douglas Lee and fell into obscurity from the 1970s to 1980s [21]. The modeling approach research regained attention in the 1990s [22] after the improvement of data environment and computer analysis capabilities, especially the development of geographic information technology [23]. Due to the acceleration of urbanization, urban malaise, suburban sprawl, and other problems have emerged. New analytical methods and tools represented by the modeling approach became a greater necessity in urban planning at that time [24]. At the same time, the modeling approach is no longer regarded as a tool to predict the boundaries of towns but as a way of theoretical research. Bottom-up town growth simulations became possible in the 21st century, and town modeling research experienced a period of rapid development after improvements in computing power [25]. At this stage, the modeling research methods begin to show a trend from static model to dynamic model, from subsystem model to integrated model, and from rule-based model to interactive model [26]. The interaction model refers to a model system that assists the communication of multiple subjects in the design process by setting up an information technology center. The interaction model system can improve design efficiency and quality by optimizing the information transmission process in a design program, and it solves the ‘black box’ problem of the rule-based model [27].

2.2. Types of Classical Town Growth Models

The growth modeling approach simulates the city’s natural development and is a helpful tool for shaping the city’s vitality. Scholars and urban design practitioners are researching the theories of the growth modeling approach. However, while related theories are increasing, the growth modeling approach is still a tool in the experimental stage. According to previous research [10,11], there are four main types of classical town growth models: spatial interaction models, mathematical models, cellular automata models, and agent-based models (

Table 1. Comparison of different urban models for the application of small town growth (source: Author).

Model Name	Methodology	Organizational Model	Application Limitations
Spatial autocorrelation model	Gravitational theory	Other-Organization	Low accuracy and lack of planning theory support
Mathematical model	Economics/statistical theory	Self-Organization	Representing full benefits through economic gains, ignoring trial and error costs
Cellular automaton	Bionics method	Self-Organization	Rules are too complex for a government-led system
Agent-based models	Theories related to Artificial Intelligence	Self-Organization	Rules are too complex and costly to construct for a government-led system

).

Spatial interaction model: Spatial interaction models were first developed by Lowry and are based on a gravity model of residential (population) distribution at fixed workplaces [28]. It later emerged as the Time Oriented Metropolitan Model [29], the Integrated Transportation Land Use Portfolio [30], and the Integrated Land Use model [31].

Mathematical/Statistical model: Mathematical modeling uses a series of equations or formulae to represent a town system’s static or equilibrium steady state. Mathematical models can quantify population growth and the allocation of land resources in towns [32]. Land bidding theory and discrete choice theory [33,34,35] can further develop these models. Based on the corresponding economic theory, scholars have proposed a series of model systems [36] to simulate changes in urban land use under housing market rules.

Cellular automaton model: The objective of the cellular automaton model is to analyze the behavior of complex town systems by observing the simple behavior of local individuals in the system. The cellular automaton is considered a marker of the transition from the conceptual to the experimental phase of town growth models. It can evolve into scalable and sustainable building development strategies [37].

Agent-based model: In the agent-based model, different “agents” are positioned in the model according to the object of study. The natural characteristics of the simulated object are used to give the “agents” cognitive abilities, initial states, judgment processes, and behavior patterns. Afterward, a predetermined number of “agents” are positioned in the manually constructed “environment”. The process of “understanding the environment, behavioral judgment, and derivation” is repeated for each process over time [38,39].

According to the comparison and analysis of existing models, integrating the current urban spatial growth model from the perspective of urban spatial morphology may be an effective method to improve the application efficiency of the urban spatial growth model. Therefore, models from the different scales can be combined by converting indicators and constants. In this case, planners and architects can simulate the entire building process by coding the environment parameters.

3. Analysis of the Characteristics of Small Towns in Shandong and Selection of Applicable Models

3.1. Typical Characteristics of Small Towns in Shandong

There are primarily two typical characteristics of small towns in Shandong:

(1)

Characteristics of the construction model of the dual organization

The planning and development process of small towns in Shandong have been subject to the influence and control of upper-level planning in central cities and regions. There is also self-organized building behavior in the construction process of Shandong’s small towns. Meanwhile, the planning and development process of small towns in Shandong is also influenced by the mutually driving effects of the objective environment and subjective policies. The development of Shandong’s small towns is a decision-making process of multiple driving forces.

(2)

Specialized industrial structure features

The superordinate plans have always controlled the development and construction of small towns in Shandong. In the context of urban agglomeration development, the industrial development of small towns in Shandong is highly dependent on central cities. The internal industrial structure is clear and unitary. This pattern of industrial structure and economic organization would be advantageous for discussing economically relevant variables in the spatial model of towns. Therefore, small industrial towns are vulnerable to uncertain external disturbances.

3.2. Prerequisites for Constructing a Model of Small-Town Growth in Shandong

(1)

Organizational model

A town growth model applicable to small towns in Shandong must be designed to account for both other-organization planning control and self-organization building behavior in the organizational system and can simulate the disturbance of policies or environment that shape small towns at the macroscopic scale and can simulate the spontaneous building practices of township residents at the microscopic scale. Correlating the data-driven mathematical model with the goal-driven system dynamic model for predicting small towns’ future growth trends can effectively simulate small towns’ dual organizational growth process. The former module represents the passive adaptation of small towns to the established objective conditions and can simulate the bottom-up growth process of small towns. The last module expresses the goals of town planning participants and can simulate the government-led growth process of small towns.

(2)

Limited budget

In the development stage of improving the quality of urbanization in Shandong, small towns, as the main focus of current urbanization efforts, will have an unprecedented development progression. Nevertheless, it is essential to note that the trial-and-error cost for small-town planning is limited. Applying the modeling approach can reduce labor intensity, shorten the planning period, and improve working efficiency. The large-scale town modeling research methods represented by big data models only partially apply to small towns. A suitable town model for small towns must consider the research and construction costs of the model, which requires the model to be universal in terms of methodology and some of its internal parameter variables.

(3)

Planning practice

Shandong’s smalltowns have been selected as the research object for the model system because the small-town industry in Shandong is relatively simple, which makes small town become a good practice object for model research. Under Shandong ‘s urban agglomeration system, the economic development of small towns is highly dependent on central cities, and there are usually only one or two main industries in small towns. At present, it is difficult for urban models to balance more complex urban economic behaviors and urban land decision-making behaviors. The use of Shandong’s small towns as experimental objects is conducive to the combination of model research and planning practice and will find problems in the promotion and application of urban models. Secondly, the existing government-led model used for small towns in Shandong is not conducive to the sustainable and healthy development of these towns, whereas the use of an urban model system to construct a city planning platform based on multi-party negotiations is conducive to improvements at the planning level in small-towns. Douglas Lee argues that the town model computational process must be transparent, easy to understand and apply [40]. According to previous research and practice, using a limited number of variables to simulate the full range of town phenomena is unrealistic. The reasons for the spatial interaction model, the mathematical model, or the CA model not being widely applicable are mainly due to the following points: First, the principle of the model is too complicated, and it is pretty different from the overall process of urban design or urban planning, so it cannot be understood by designers or decision makers. Second, the model is not adaptable and cannot be adjusted to new objects of analysis by modifying parameters or implanting a new status based on the original model. Therefore, in many cases, applying the urban growth model cannot reduce labor costs in urban planning or urban design. Third, the existing urban model system cannot reflect the public will. Even the CA model requires the designer to translate the public will. This process is incomprehensible for the public, who cannot have direct involvement, so much information may be lost or misinterpreted. Finally, as an urban dynamic model with a time dimension, although the urban form of different stages can be generated by iterative derivation, the original conversion principle has not changed, and the dynamic concept of real-time data monitoring, real-time feedback, and real-time correction is not fully realized. Using human-computer interaction to simplify the model structure can help the model system be more applicable to town planning [41].

Combining the three perspectives, the growth model that applies to small towns in Shandong must be a dual organizational structure model that costs less and is easier to understand and visualize. By comparing the theoretical basis and limitations of the different modeling approaches, this essay constructs a model system based on a mathematical model and a system dynamics model. It is an interactive model with suitable rules and factors, so initial information and relevant development goals can be inputted into the model system. Automatically generated spatial data are outputted to a visual interface for decision-making.

4. Constructing a Growth Model Applicable to Small Towns in Shandong

4.1. Model Framework Construction

To accommodate the dual organization and construction model, the overall structure of the model for small towns in Shandong should consist of two parts: (1) both statistical and suitability growth forecasting for small towns and (2) reflecting the quantitative analysis of objective patterns and expression of subjective guidance in the planning process. The model calculation is also divided into two parts: the research and planning stages.

The planning research stage collects data and provides a foundation for the model calculation. The sections of this stage include: collating the upper-level spatial plan for the target town, obtaining the descriptions of the town development objectives in the planning text, clarifying the bottom line in the upper-level spatial plan, collecting historical development data, including land use growth data and population growth data, and obtaining the current land use layout of the built-up area of the town, which includes industries, road networks, infrastructure, and other features. A logistic regression analysis of the factors that may be influential must be conducted to identify the default land use pattern for towns. The logistic regression model belongs to the statistical/mathematical model and is widely used in empirical research. The logistic regression model applied in the small town growth forecast model is designed based on the multivariate logit model that is based on the random utility theory and discrete choice theory proposed by McFadden [42].

In the planning stage, planners or other participants classify small towns according to their predominant industries and apply the appropriate land use growth constant corresponding to each industry. At the same time, input the participant’s policy or guidance strategy through the interactive interface. To convert policy or goal disturbance into land use impact, a system dynamic model including suitability analysis is conducted for all the parcels within the developable land area of the town. Finally, the statistical and suitability growth projections are combined to obtain the base elasticity boundary for the town growth projection [43].

4.2. Small Town Growth Impact Factor Selection

Applying the elements of Ian McHarg’s classification study, Carr and Zwick proposed a method for selecting impactful factors based on seven categories [40], which include geographic, ecological, demographic, economic, policy, cultural, and infrastructural elements. These seven categories summarize the factors that may influence town growth. They can comprehensively describe the characteristics of each parcel in the analysis range, which provides a foundation for establishing a new selection system for primary factors. Based on the analysis of existing studies, the factors that may affect land use growth in small Shandong towns are summarized as follows.

• Geographic elements: slope and distance from bodies of water.
• Ecological elements: forests and wetlands.
• Demographic elements: population density.
• Economic factors: existing towns, distance from central business districts, distance from central industrial districts, distance from existing towns.
• Policy elements: conservation lands and urban agglomeration gravity.
• Cultural elements: historical and cultural conservation areas.
• Infrastructural elements: distance to roads, distance to major roads, distance to major nodes, road density, and distance to shorelines (river banks).

4.3. Classification of Small-Town Growth Patterns

Small towns in Shandong influenced by urban agglomeration development strategy are particularly influenced by the dominant industry in the urbanization process [44,45]. The industry provides support for the urbanization process. In the development process of small towns, industrial development manifests itself as capital agglomeration and job agglomeration, which attracts more people to central cities, especially more rural agricultural labor to towns. Development leads to an increase in urban population and expands the construction land area. Since industries differ greatly in terms of production patterns, land use scale, and infrastructure needs, and the influential factors of various leading industries that drive the urbanization of small towns also differ, the construction of a predictive model for the growth of small towns in Shandong requires small towns with different leading industries to be considered separately. Based on the urban agglomeration system, small towns in Shandong can be divided into three categories according to their functions: industrial towns that are part of the regional industrial chain, cultural and tourism towns that provide services to the central city, and agricultural towns that have weak links with other towns in the region. Small industrial towns can cooperate with central cities for industrial cooperation, absorb the surplus rural labor force, and help promote the rational distribution of industrial space layout and labor force distribution within the region. So, among the above three kinds of small towns, small industrial towns are most obviously affected by the regional collaborative production mode and the development strategy of urban agglomeration. They are also most vulnerable to the impact of uncertain external disturbances. This essay selects small industrial towns in Shandong as the research object and collects data from small towns in multiple regions to build the model. This essay, from the perspective of model research, selecting a single category of small towns as the research object is conducive to improving the system by focusing on the model structure. From the perspective of planning practice, this essay’s selection of small industrial towns as the research object is beneficial to improve the application probability of the model by specifying research objects.

4.4. Growth Interaction Model Framework

An interactive model framework for small-town growth was constructed to synthesize the above model structure and the study on the selection of growth influencing factors (Figure 2).

5. Statistical Forecast of Small-Town Growth

5.1. Status of the Case

There are currently 41,636 small towns (Xiang or Zhen) in China and 1824 small towns in Shandong [46]. The development gap between small towns in different provinces is significant, impacting the model’s generalization. Another issue is that industrial towns in Shandong Province are concentrated in plain areas, and the development of small towns is less affected by the terrain. In addition, many factors, such as history and culture, will affect the development of small towns. Therefore, Shandong Province is selected as the research object in this article.

Diao Town is in Zhangqiu District, Jinan City, Shandong Province, and has a population of 123,814 as of 2017. Jinan is one of the Shandong regional center cities (Figure 3). Zhangqiu, located in the east of Jinan, is an essential node between Jinan, Zibo, Binzhou, and other big cities and an important part of Jinan’s urban agglomeration. Diao Town is located northeast of Zhangqiu, which is conducive to the formation of contact and cooperation between Diao Town and surrounding cities and towns (Figure 4). Diao Town has received many factories in the process of industrial restructuring in Jinan. In recent years, the town’s industrial economy has developed rapidly, leading to the formation of two leading industries of chemical industry and machine manufacturing. Diao Town’s industry-dominant position is increasingly prominent. Six leading industries have been supported and developed, including the chemical industry, light industry, machinery manufacturing, furniture, cloth, printing, construction, and building materials. The company generates annual profits and taxes of more than 40 million yuan, accounting for 73% of the town’s tax revenue. Diao Town’s GDP ranks 411 among small towns nationwide and 31 in Shandong Province. As a typical small industrial town, Diao Town’s agriculture and service industry account for a very low proportion of GDP [47].

5.2. Logistic Regression Results

After accessing the satellite map of the target town from 1969 to 2019 and the evolution of urban land use, applying geographic information technology to achieve “form-number” transformation and logistic regression analysis of the transformation results, the results of the regression analysis are used to identify constants in statistical growth forecast. The specific analysis process is as follows: First, identify all environmental factors that may affect the town’s growth and vectorize all possible factors by extracting the attributes of all parcels in the analysis boundary. Then, a logistic regression analysis was conducted to reveal the main influencing factors of different development stages. Finally, according to the regression analysis results and the matching with the policies at that time, analyze the main factors affecting the growth of small towns in different policy backgrounds. The results of the regression analysis of impact factors for Diao Town 1969-1979 are in

Table 2. Results of regression analysis of impact factors for Diao Town 1969–1979 (Source: Author).

Variable Name	B	Std. Error	Wald	Df	Sig	Exp (B)
Distance to open water	0.000	0.000	2.377	1	0.123	1.000
Distance to industry	0.000	0.000	4.572	1	0.032	1.000
Distance to town	−0.006	0.000	242.981	1	0.000	0.994
Distance to highway	0.000	0.000	22.459	1	0.000	1.000
Distance to major roads	0.000	0.000	1.775	1	0.183	1.000
Road density	43.395	28.675	2.290	1	0.130	7.017 × 1018
Town agglomeration gravitation	0.000	0.000	0.010	1	0.920	1.000
Slop	0.133	0.030	19.095	1	0.000	1.142
Population density	0.001	0.000	31.482	1	0.000	1.001
Distance to node	0.000	0.000	13.241	1	0.000	1.000
Existing town	−4.172	0.190	480.107	1	0.000	0.015
Protected land	−7.418	0.444	279.676	1	0.000	0.001
CONSTANT	1.644	0.550	8.944	1	0.003	5.177

, and the results mainly reveal the influencing factors of independent development of small towns. The results of the regression analysis of impact factors in Diao Town 1979–2009 are in

Table 3. Results of regression analysis of impact factors in Diao Town 1979–2009 (Source: Author).

Variable Name	B	Std. Error	Wald	Df	Sig	Exp (B)
Distance to open water	−0.001	0.000	129.388	1	0.000	0.999
Distance to industry	0.002	0.000	356.500	1	0.000	1.002
Distance to highway	0.001	0.000	268.541	1	0.000	1.002
Distance to major roads	−0.001	0.000	192.713	1	0.000	0.999
Road density	15.959	24.513	0.424	1	0.515	8,533,084.803
Town agglomeration gravitation	−0.002	0.000	271.095	1	0.000	0.998
Slop	0.053	0.037	2.063	1	0.151	1.054
Population density	0.000	0.000	24.599	1	0.000	1.000
Distance to node	0.000	0.000	19.744	1	0.000	1.000
Protected land	−7.774	0.374	430.964	1	0.000	0.000
Distance to town	−0.003	0.000	145.033	1	0.000	0.997
Existing town	−21.383	783.849	0.001	1	0.978	0.000
CONSTANT	15.981	783.849	0.000	1	0.984	8,721,073.526

. The results mainly reveal the influence factors applying small towns’ development model, taking population transfer as the core task. The results of the regression analysis of impact factors in Diao Town 1979–2009 are in

Table 4. Results of regression analysis of impact factors for Diao Town 2009–2019 (Source: Author).

Variable Name	B	Std. Error	Wald	Df	Sig	Exp (B)
Distance to open water	0.000	0.000	4.442	1	0.035	1.000
Distance to industry	0.001	0.000	61.261	1	0.000	1.001
Distance to highway	0.001	0.000	30.630	1	0.000	1.001
Distance to major roads	−0.001	0.000	98.618	1	0.000	0.999
Road density	−34.820	37.908	0.844	1	0.358	0.000
Town agglomeration gravitation	−0.002	0.000	64.408	1	0.000	0.998
Slop	−0.145	0.066	4.875	1	0.027	0.865
Population density	0.000	0.000	0.525	1	0.469	1.000
Distance to node	0.001	0.000	49.181	1	0.000	1.001
Protected land	−18.770	210.448	0.008	1	0.929	0.000
Distance to town	−0.011	0.001	167.241	1	0.000	0.989
Existing town	−33.987	523.461	0.004	1	0.948	0.000
CONSTANT	29.439	523.461	0.003	1	0.955	6,095,597,479,066.454

. The results mainly reveal the influencing factors of small towns as regional collaborative production nodes.

5.3. Analysis of Statistical Results

Continuously significant: distance from industrial areas, distance from built-up areas, distance from highways, and distance from traffic nodes.

For small industrial towns, the same elements that influence industrialization also influence urbanization. Since local industrialization provides processing services for the regional industrial chain, the influence of transportation elements such as highways and freight terminals on urbanization is always significant.

• Persistent insignificance: road density

The road density within the town has not significantly impacted the urbanization development of Diao Town in the past 40 years because the main modes of transportation for Diao Town residents are walking, public transport, and bicycles, and freight mainly depends on railways and highways. Since town residents do not commonly use private cars, there is no high demand for town roads.

• Progressively significant: distance from water bodies, urban agglomeration gravity, and distance from main roads

The distance from open water on urbanization is gradually significant, indicating that water sources’ impact on industrial activities is significant and residents prefer to live in areas along the water for a better living environment. The gradually significant factor of urban agglomeration gravitational influence means that the influence of urban agglomeration development strategy is prevalent. It also means that the production mode of small towns transitions from independent production to specialized production as a link of the regional industrial chain with the central town as the core. In addition, the influence of the main roads of towns on urbanization has significance due to the imperfection of the road system.

• Progressively insignificant: slope, built-up area

The slope is an essential factor affecting town expansion because the slope of the original terrain can significantly affect subsequent construction. The Shandong region is relatively flat, so the slope does not significantly limit the growth of small towns there. The built-up area factor is no longer significant, which indicates that the land use in towns is compact and that some villages on the peripheries of towns are beginning to disappear.

• Other: population density, forest (not significant 1979–2009)

The period of 1979–2009 experienced rapid growth of towns and cities, which was characterized by town sprawl and real estate development. Large amounts of land, including forest land, were being used for residential construction. After 2009, implementing the urban agglomeration system and strengthening land management led towns to return to an industry-focused development path.

In order of the most influential, the elements of small towns with industry dominating are the distance from industrial areas, distance from highways, distance from transportation nodes, distance from built-up areas, distance from bodies of water, urban agglomeration gravity, distance from main roads, population density, and woodland areas.

5.4. Town Growth Forecast

Figure 5 presents the results of the regression analysis. The results consist of how much Diao Town is projected to grow in conjunction with relevant policy norms. This result mainly reveals the future growth results of Diao Town applying the current policy background, assuming that there is no change in the development environment of small towns. According to Figure 5, the growth of Diao Town is mostly concentrated around Chaijia Village, Diao Xi, and Zhang Guanzhuang Village. The statistical forecast shows the growth result of small towns while the external environment remains stable. However, previous research has found that the growth of small industrial towns is vulnerable to uncertain external disturbances. Therefore, it is necessary to identify the development trend of small towns in different environments by applying scenario planning.

6. System Dynamic Model Forecast

The system dynamic forecast model of town growth boundary planning is a control system using public participation constructed through the townland suitability analysis and the hierarchical analysis method [48], which reflects an applicable expression of subjective growth guidance in the planning process [49]. Compared to the stable objective growth pattern for small towns, subjective growth guidance has more uncertainty because it requires immediate adjustment according to changes in the external environment. The uncertainty of the planning environment is random and unpredictable. It cannot be addressed with a subjective scenario setting or controlled by “experience, high-quality data, and long monitoring time”. Therefore, the system dynamic forecast model introduces the concept of scenario planning to achieve reactive response through internal interaction and to assess and address uncertain disturbances.

The scenario system is based on predicting external disturbances that occur during the development of small towns and the corresponding adjustments to small-town plans under each scenario [50]. The scenario system’s goal is to enhance small towns’ ability to resist external disturbances and ensure small towns’ stable and sustainable development. The scenario system consists of three parts: scenario setting and identification, scenario selection, and scenario detection.

Model research stage: scenarios are pre-defined based on the impact of external uncertain perturbations on small towns. Scenario pre-determination and identification occur in the research phase. Each scenario includes a weight system for that scenario and a corresponding scenario baseline. The scenario system predetermines four broad categories of “social, economic, environmental, and policy” based on the impact of external uncertainty perturbations [51].

Planning stage: multiple planning strategies are developed based on scenarios. The corresponding planning strategy is customized for each scenario and accounts for each scenario’s characteristics. The planning response for the scenarios is primarily reflected through changes in weights. When the corresponding scenario is in effect, the target ranking of that scenario affects the suitability analysis in proportion to the suitability growth projections for small towns.

Planning implementation and management stage: small town development is monitored based on scenario selection baselines and evaluation of outputs. The scenario selection baseline is a statistical monitoring value in a small town that identifies the impact of uncertain external disturbances on the small town. The baseline is the foundation for the scenario to go into effect. Scenario detection means that when a specific scenario takes effect, the detection parameters corresponding to that scenario take effect simultaneously. The detection parameters are used to evaluate whether the small town can ensure stable and sustainable development with the scenario in effect.

By introducing the logic regression growth results of Diao Town and the weight system based on subjective guidance and correction, the future growth boundary of Diao Town can be obtained, as shown in Figure 6.

7. Conclusions

Compared with the previous research on small-town growth, the theoretical model proposed in this paper combines data-oriented market behavior prediction with goal-oriented policy guidance and presents a prediction method suitable for small towns in Shandong. This paper also proposes an analysis method based on the characteristics of small towns in Shandong and found that partition data according to the time nodes of policy changes can better explore the information latent in the data and enable the impact analysis of policy changes on the construction process of small towns. The regression analysis results indicate that, under the influence of different policies, the growth factors of industrial towns are different due to the size of small towns, and the impact of policy changes in small towns is more direct and significant than in cities. However, small towns are always dependent on central cities regardless of the policy context and dominant industry. This essay constructs a growth model framework applicable to small towns in Shandong by incorporating the new data environment supported by the quantitative research methods of the built environment. Applying the data-digging method [49] and the interaction models to simulate the organizational mechanism of small towns can help the model’s application in planning practice. Meanwhile, the theory of scenario planning is introduced to enable the model to change its output immediately according to changes in marker variables and can fit the organizational characteristics of a small town in Shandong.

The number of samples selected for the study is limited, and there needs to be a more cross-sectional comparison among small towns in the same category. The study relies on open data and lacks research on the joint action mechanism of economic factors, human factors, and town spatial forms. A future study that combines planning practice, introduced growth rate, industrial energy efficiency, and other relevant economic considerations would be more comprehensive and result in more accurate predictions of town growth. In addition, future research will incorporate other influential factors such as terrain, culture, and history and generalize this model for the whole of China.