Which of the following scenarios is best solved using a regional planning model

Urban sprawl, which arises from the rapid growth of the economy and population, has become a major challenge for sustainable urban development worldwide (Yao et al., 2016, Yao et al., 2017, Hashem and Balakrishnan, 2015, Liu et al., 2014). For assisting urban planning, methods or models are required to guide and constrain urban area growth (Long, Han, Lai, & Mao, 2013). Urban growth boundaries (UGBs) have been a common tool used by planners to control urban development in open spaces, protect superior rural areas that make significant contributions to the urban environment from development, and promote efficiency in urban management, especially where there is residential development in established and planned suburban areas (Gennaio, Hersperger, & Burgi, 2009). Moreover, this planning tool is also important for increasing the density of urban services and reducing urban infrastructure costs (Tayyebi, Perry, & Tayyebi, 2014). A recent study was carried out by Long, Han, Tu, and Shu (2015) regarding planner designed UGBs, and they reported that UGBs were effective in containing human mobility and activity. In addition, the control function of UGBs increases over time during urban development, and the effect of UGBs is clearly stronger in exurban areas than in central urban (Long, Gu, & Han, 2012). Therefore, the UGBs will play an increasingly important role in the future of new urban land management.

UGBs are most often established in high growth areas such as metropolitan areas. The earliest UGB can be traced back to the green belt in London in the 1930s (Nelson & Moore, 1993). In recent decades, UGBs were first adopted widely in the United States (Hepinstallcymerman et al., 2011, Jun, 2004) and then, they were gradually brought into other countries such as China (Han, Lai, Dang, Tan, & Wu, 2009), India (Venkataraman, 2013), Canada (Gordon & Vipond, 2005), Albania (Carter, 1992), Australia (Coiacetto, 2007), Switzerland (Gennaio et al., 2009), etc. To date, UGBs are being used by an increasing number of local governments in various countries around the world to direct urban growth (Ma, Li, & Cai, 2017). As UGBs attract an increasing amount of attention, there are also a growing number of requirements to develop efficient and feasible techniques to define those boundaries for different applications. However, many UGBs are delineated by conventional methods that are only based on the personal experience of planners, which may lead to the lack of an adequate scientific basis and quantitative support (Long et al., 2013). Land use suitability evaluation models used to evaluate UGBs based on a series of spatial factors, e.g., topography and traffic conditions (Cerreta & Toro, 2012) have also been widely used in previous studies (Bhatta, 2009). Although easy to implement, these methods ignore the urban landscape characteristics, which will have negative effects for establishing elaborate urban boundaries (Cao et al., 2012, Ma et al., 2017). Moreover, many geographic factors that drive urban change operate across different spatial and temporal scales in a very complex way (Tayyebi et al., 2011, Tayyebi et al., 2011). Suitability evaluation models commonly fail to reflect these relationships and interactions, which may result in UGBs failing to realistically accommodate future urban expansion (Tayyebi et al., 2011, Tayyebi et al., 2011).

To overcome the disadvantages of the abovementioned studies, many researchers have established UGBs by adopting the cellular automaton (CA) model. The CA model differs from previous models (manual method and suitability evaluation model) in its ability to represent spatial interactions implemented in the immediate neighborhood or the hierarchical structure of the neighborhood (Li et al., 2017a). CA can simulate the dynamics of urban growth at the landscape level (Verburg & Overmars, 2009). Through iterations and updates, CA can efficiently incorporate the interactions between urban growth and its corresponding geographic driving factors (Clarke & Gaydos, 1998). Thus, by using the CA model, UGBs can be generated from the simulation results in these studies. For example, Tayyebi et al., 2011, Tayyebi et al., 2011 proposed two rule-based CA models for the Tehran metropolitan area, which can directly predict the size and shape of urban boundaries. Long et al. (2012) delimited UGBs for the Beijing region using a constrained CA and compared the results to those established in the city master plan. The results show that CA is a helpful planning tool for the establishment of UGBs. Some of the researchers have tried to combine CA with intelligent algorithms such as logistic regression (Hu & Lo, 2007), artificial neural networks (Tayyebi et al., 2011, Tayyebi et al., 2011), particle swarm optimization (Feng, Liu, Tong, Liu, & Deng, 2011), and ant colony optimization algorithms (Ma et al., 2017). In addition, a recent study also proposed CA models based on partial least squares (PLS-CA) regression or generalized pattern search (GPS-CA) which can better explain the dependent variables and reduce simulation uncertainties. These CA models also have great potential to improve the CA-based UGB delineating method (Feng, 2017, Feng et al., 2016). In summary, the use of these intelligence algorithms allows CA models to simulate the local interaction between land use patterns and various driving factors (Li and Yeh, 2002, Lin et al., 2011, Liu et al., 2010).

CA-based UGB models have made superior progress compared to previous UGB methods. However, the spatial patterns of urban expansion are significantly affected by regional planning on both regional and local scales (Lu et al., 2013, Tian and Shen, 2011). Most of the previous UGB models only focused on the local “bottom-up” effect of the CA model but ignored the “top-down” effect at the regional scale. The large-scale influences usually refer to the future demand for economic development and population increase that determine the future amount of urban land in a region (Verburg & Overmars, 2009). The local effects are indicative of interactions and feedback between land use patterns and multiple spatial driving forces, which include the road network, geographical locations, terrain conditions, etc. (van Asselen and Verburg, 2013, Verburg, 2006, Verburg et al., 2011). On both scales, these effects are influenced by the development policies of a region (Gao, Wei, Chen, & Chen, 2014), and the urban area dynamics are largely determined by forces that are exogenous to land allocation. Therefore, the influence of regional planning on both scales should be considered by coupling the “bottom-up” CA model with a ‘top-down’ approach (Verburg et al., 2009, Xiang and Clarke, 2016). However, there are no previous studies that attempted to build a UGB model by integrating both the macro urban demand and local dynamics.

Additionally, the ways in which different planning policies influence the spatial patterns of urban areas and future UGBs under various scenarios is of great importance for decision makers to assess the outcome and impact of different policies (Chen, Li, Liu, & Ai, 2014). For example, Long et al. (2012) incorporated urban planning to simulate a planning-strengthened scenario in Beijing to help illustrate the impact of urban planning on urban expansion. In addition, considering planning factors in the urban simulation can be employed by decision makers in the early stages of policy making; this operation provides an inexpensive and effective way for planners to obtain helpful information about the influences of different development policies or planning scenarios on urban development, which may prevent poor urban designs (Clarke, 2014). With this information, planners can better adjust the direction of urban development by modifying corresponding planning factors and planning policies, as well as delineating more appropriate UGBs. Most of the previous research only attempted to build UGBs under a single scenario, in specific time nodes or by a set of model parameters (Ma et al., 2017, Tayyebi et al., 2014, Inkoom et al., 2017). However, a very limited number of studies have tried to establish UGBs for large-scale and fast-developing areas under various planning scenarios. Another challenge in delineating UGBs is that some cities with amazing development speed show fractal characteristics in urban land forms, spatial form and landscape organization (Yuan, 2005). An example of this is the Pearl River Delta area in China. This results in UGBs in these areas potentially comprising numerous polygons and even showing a dispersed form. When delimiting the UGBs, polygons with low compactness and a small area should be eliminated, as they are not feasible for urban development. This indicates that the results of the simulation model cannot be directly used as final UGBs. Previous studies established UGBs based on CA simulation, which was mainly through manual modification (Long et al., 2013). Such modification is quite subjective and inconvenient to use. The effective establishment of UGBs from the CA model simulation results remains unresolved for practical problems.

In this paper, a novel UGB delineation framework is presented, in which UGB-FLUS is proposed to tackle these problems. This framework is implement by two steps: 1) urban growth simulation with a future land use simulation (FLUS) model and 2) delineating UGBs based on the simulated urban growth. The FLUS model is a CA-based method that is integrated with a top-down approach to solve UGBs problems. This FLUS model has been proven effective for projecting complex land use changes under various design scenarios (Liu, Liang, Li, & Xu, 2017). By using this FLUS model, the visions of planners can be embedded as the constraints or drivers for creating UGBs. In the second step, we proposed a component based on the theory of erosion and dilation to improve the effects of generating plausible UGBs from the simulation results, because traditional methods cannot effectively remove the small and dispersed urban patches. This method is used to merge and connect the cluster of urban blocks into one large area and simultaneously eliminate the small and isolated urban patches. The application of this proposed framework is carried out in the Pearl River Delta (PRD), which is one of the fastest growing regions in China.

Methods

The UGB-FLUS framework involves several techniques. First, a spatial simulation model based on the theory of cellular automaton (CA)—the FLUS model (Liu et al., 2017)—is used to project the spatial distribution changes in urban land use for the PRD region. This model has been successfully tested in the simulation of land use and land cover change in China (Liu et al., 2017), as well as on the global scale (Li et al., 2017b). In this study, we further modify the FLUS model so that it can be

Study area

The PRD region is in southern China and encompasses an area of 54,000 km2; it is widely recognized as one of the most developed regions in China (Fig. 5). Since the implementation of reform and opening in recent decades, the PRD region has successfully developed into an economic, cultural and traffic center for south China. At the same time, the PRD has experienced rapid urbanization with the fast growth of GDP and population. Thereby, this raises a series of land use problems such as the

Planning scenarios

One of the purposes of this study is to simulate urban growth under the designed scenarios that closely link to planning policies in terms of space and quantity. We developed six scenarios under different spatial planning policies; these scenarios and spatial policies are very typical and commonly used in regional planning in other areas (Al-Ahmadi, Heppenstall, Hogg, & See, 2009). All spatial planning policies are shown in Fig. 6.

According to the different influences of various spatial forces

Conclusions

In this study, a UGB-FLUS method is proposed to support the planning process in a complex, urbanized region such as the PRD, which is composed of a CA-based FLUS model and a morphological UGB delineation technique. A case study in the PRD area was developed to demonstrate how urban patterns under different scenarios can be generated using this method.

This model successfully creates different urban forms under different planning policies such as the “High-speed Railway Station-centered

Acknowledgments

This research was funded by the National Key R&D Program of China (No. 2017YFA0604404 and 2017YFA0604402), the National Natural Science Foundation of China (Grant No. 41671398), and the Key National Natural Science Foundation of China (Grant No. 41531176).