A Land Use Planning Literature Review: Literature Path, Planning Contexts, Optimization Methods, and Bibliometric Methods

Abstract

Land use planning studies are accumulating in unprecedented quantities, and have created a wide space for an extensive summary, the synthesis of fundamental developments, a sharpening of the focus of future study issues, and the dissemination of concise information among the academic community and the policy making environment. Nonetheless, the publication rate and content coverage of land use planning literature reviews have been relatively low. This study aims to contribute to the field by first assessing the effectiveness of the bibliometric method as a literature study method. It then summarizes the developments in the land use planning literature in general path building process, the planning context, and the development of methods. The study findings highlight that the bibliometric method tends to reward information carriage papers rather than original contributors. More than 85% of the time, published articles are cited for general information rather than their core research contribution, suggesting the incomprehensiveness of the bibliometric method in studying real knowledge development. In terms of the land use planning context, significant gaps are observed, particularly in urban land use, including the detachment of the general goal-oriented objective formulations from basic land use allocation theories and conceptual models. A serious concern also arises that the formulations of certain urban land use optimization objectives appear to contradict the original concept of a sustainable city. To address these gaps, this paper suggests several key research frontiers, including linking the basic land use allocation and utility theories to land use optimization, and a need to shift current urban land use planning/optimization approaches from spatial optimization, which changes land uses to meet flow resources, towards optimizing the flow of resources to fit the resilient nature of existing spatial configurations and physical establishments in the built environment. Additionally, evaluating the existing built environment for optimality should be prioritized rather than solely focusing on new developments. In terms of method development, the study suggests that advancing current loosely coupled methods into more integrated systems may be a promising frontier. In general, the paper strongly suggests the possible reiteration of the land use planning study landscape to make sustainable development discourse more concrete and to make the method development stage involve more integrated systems. Immediate research interventions may be the standardizing of land use planning studies through the development of protocols and standard benchmark problems.

1. Introduction

Land is the spatial carrier of all sorts of human life. It shapes a community’s socio-political and economic establishments through the interplay of use and value under a given tenure system [1,2,3], wherein the rights and responsibilities of the individual, groups/communities and the duty of the government are defined within a context of an overall national development framework and regional/urban development aspirations, shaped within ideologically framed national development policies. At any scale, the spatial configuration of land use is a physical manifestation of the distribution of the structure of benefits and costs to a society [4,5]. Enforcement of fair distribution of such benefits and costs among communities and among the groups and individuals within a community is one fundamental reason that land use must be planned. This is especially true where the market fails, as is so often the case, to fairly balance benefits and costs [6]. On the other hand, changing the relation of use–value drives land use land cover change (LULC), where such alteration of ecosystem services (ESs) causes changes to the spatial relation of human activities (human–spatial interactions) [7,8]. This is the second fundamental reason that calls for the effective planning of land use. Land use planning is instrumental in alleviating the potential for incompatible changing of regional/rural land into land for urban use that detrimentally affects the productivity of the primary food supply and ecological services and alleviates land scarcity within the built environment through different use policies [9]. In general, as a human economic development carrier resource, as a natural endowment and ESs provider [10], and as an institutional entity that shapes the socio-political behavioral relations of humans by tenure conditions, both rural and urban land require effective planning for their sustained productive use. In other words, the degree to which human actions have an effect on changes in the environment has remained a key subject of study.

To this end, societies in successive generations have utilized land use planning, shaped in the context of corresponding dominant ideologies/thinking regarding development. Since the pioneering land use planning model proposed by von Thunen [11], the field of land use planning, especially in urban areas, has evolved through generations of models, including structured mathematical models in the 1950s and 1960s, where bid-rent and optimal firm location theories played a significant role in conceptualizing the spatial allocation of activities [12]. Spatial simulation models dominated the 1970s and 1980s, and since the 1990s, the focus has shifted towards sustainable development in the land use planning literature, where the hegemony of market forces culminated even in countries which are fundamental market advocators, such as the UK [9]. From a technical perspective, sustainable development-framed planning literature now integrates demand–supply quantity structure and spatial simulation, aiming to achieve a balance between economic, societal, and ecological outcomes [13,14].

Within sustainable development thinking, optimizing land use allocation based on the trade-off between different objectives goes beyond economic bidding. It considers the relative productivity of various land use structures for various capabilities (such as access to transport, recreation, shopping, etc.), or alternative function values (ecological reserve, built-up areas, mobility) [13,14]. With the shift in theoretical ground, the land use modeling task has become more challenging, with an increasing number of objectives, specific policy restrictions/requirements, and the need for stakeholder engagement at different stages of the planning process. As a result, traditional structured mathematical models have become less attractive, and self-learning machine learning algorithms (MLA) and dynamic metaheuristic algorithms (MHA) have gained prominence in the land use planning literature. These not only abruptly minimize the costs of processing data, but technological advancements in spatial data acquisition have further increased their capacity to determine patterns of land use and their prediction capacity regarding future use of both urban and regional/rural land.

Within the paradigm of sustainable development, the concept of ESs effectively bridges the gap between science and policy in regional/rural land use planning [8,15]. In urban land use planning, the sustainable built environment discourse, characterized by compact built-up areas, harmonious functionality, mixed-use development, and the relationship between physical structures and the natural ecosystem, serves as the major mainstream form [16,17]. Conceptualizing land use optimization as a contemporary approach to the classical highest-and-best-use value allocation holds promise for the achievement of sustainability [18]. Understanding the drivers of land use change and analyzing spatial measurements and the laws of spatiotemporal changes in land use are essential for modeling existing phenomena and simulating future spatial patterns of land use [6].

MHA and ML technologies have not only overhauled the technical capability of examining land use plan scenarios. They also have contributed to the shift in spatial development thinking by enabling the capacity to handle temporal perspectives and uncertainties. However, their effect over the fundamental spatial development theories and conceptual models looks unattended, especially in the built environment. The contemporary literature on land use planning tends hegemonically drifting away from spatial domain towards the domain of the technical capability of artificial intelligence in land use planning. This observation can easily be justified by the fact that land use optimization objectives are often formulated as being general management-oriented, especially in urban land use planning research. Only a limited number of works address the basic land use allocation conceptual modes spanning economic geography, utility theories, and spatial morphology. Nonetheless, the whereabouts of such theory-grounded conceptual models has not been explored in the optimization-based land use planning literature, including in review studies.

The better way to address a knowledge gap is conducting a literature study. As in any other disciplines, literature studies play a vital role in land use planning to trace the development trends of concepts/contexts and methods, develop hypotheses regarding new directions, identify current hotspots, and suggest potential research directions. However, review studies in optimization-based land use planning are relatively limited in number compared to the continuously accumulating volume of studies [18]. During data retrieval in this study, for example, a broader title “land use optimization*” search parameter in the Web of Science core collection yielded only six articles (1.13% of 530) written in English. In addition, since any review work addresses only a few, if not a single issue, such as a single or certain group of optimization methods, a few objectives of land use optimization, etc., more reviews on land use planning are required. More importantly, reviews that assess the harmonization/synergy of theories/conceptual models/ are worth more given much of the available literature is heavily focused on the development of methods and their applications.

Against this backdrop, and recognizing that the existing reviews have limited themselves to specific planning contexts or methods, this paper aims to achieve the following objectives.

(i)

Assessing the credibility of the bibliometric method as a main literature study method of investigating knowledge development;

(ii)

Constructing the temporal trajectory of the land use planning literature to allow researchers to contextualize their problem within a specified time frame or perspective, or explore the influence of a specific time’s thinking on the application of methods in depth;

(iii)

Compiling and synthesizing state-of-the-art of land use optimization methods, characterizing their defining nature and identifying research frontiers;

(iv)

Providing a concise summary of existing optimization-based land use planning concepts and exploring the whereabouts of fundamental classic land use allocation theories/concepts and utility models within the popularly governing optimization-based land use planning research, emphasizing the built environment.

Against the above background, section two describes the data and methodology employed in the study. Section 3 presents the results of the bibliometric analysis and it evaluates the suitability of the method for the exploration of real contribution of publications. Section 4 summarizes previous review works and provides a foundation for Section 5. Section 5 delves into the exploration of the development of land use planning knowledge and methods. Finally, Section 6 concludes the paper by summarizing the key findings and implications for reiterating the land use planning context and further method development.

2. Materials and Methods

The literature study is a comprehensive and thorough research method that serves a research community with a solid foundation in a specific discipline [19]. The purpose of conducting a literature study is not just to compile available knowledge so as to facilitate readers with a comprehensive summary of a specific topic of interest. The integration and synthesis of findings of different studies rather makes the literature study a more powerful method of addressing an issue of interest than any single study. It could be due to this power that journals editors have increasingly been attracted to review manuscripts over regular articles, as evidenced by the high citation and downloading rates of review papers compared to regular articles [20]. A comprehensive review is also a valuable source of up-to-date information for practitioners.

Literature study methods vary depending on the objectives and the types of data. Based on the aim of the study, data (type, source, and acquisition technique), and the desired outputs, ref. [19] categorized the literature study approaches into systematic, semi-systematic, and integrative. A systematic review that focuses on a specific research question aims to synthesize and provide a comprehensive analysis of existing evidence on a particular research theme [21]. To select relevant published articles for inclusion in the review, a systematic strategy is applied. The outcomes of the analysis often lead to policy implications. On the other hand, the aim of an integrative review is to synthesize and critically investigate existing knowledge on a specific issue. Data sources are not limited to published articles, and their collection does not strictly adhere to systematic selection strategies. The outcomes of such qualitative analyses typically include identification of themes and theoretical models/frameworks. In between these two approaches, there is the semi-systematic review, which focuses on specific research areas and traces the development of a research theme over time. In this method, the source selection strategy may or may not be systematic. The outcomes of a semi-systematic review often include theme identification, historical overviews, and the identification of research frontiers [22].

This paper falls under the category of a semi-systematic review. Its purpose is to provide a synthesis of the literature on optimization-based land use planning over the past two decades and to identify potential research directions to facilitate further sharpening of land use planning conceptual underpinnings and methods based on cautiously sampled articles. It employs both quantitative (statistical)-based bibliometric and traditional reading (text analysis) methods.

The study resources were sampled from the Web of Science core collections using title as a search parameter and alternative titles and keywords as search values. Five alternative title values (land use optimization*, land use optimization, land use, land use planning, and urban land use planning) and three alternative keyword values (optimization, metaheuristics, and genetic algorithm) were tried iteratively with each title value. To maximize diversity in the retrieval, the keywords retrieved along with each search title value were mapped in a word cloud for visual analysis. Based on a visual inspection of the keyword clouds, we identified that the keyword value “optimization” returned for the title value “land use planning” was denser than any other alternative. Accordingly, the sample retrieved using the search title value “<TI = land use optimization*>“ was selected for analysis. As of 1 August 2023, a total of 670 publications met the search criteria, spanning the period from 1993 to 2023. Among these, 339 were articles published in 201 journals, affiliated with 789 institutions from 122 different countries (Figure A1). After, the articles were filtered following the PRISMA screening process [23]. Approximately half of the publications (331) were excluded for the reason they were not articles. The remaining 339 articles met the inclusion criteria, including language (English), availability of full text, and availability of keywords. Furthermore, no duplicates were found.

Despite the fact that rural/regional land use planning and urban land use planning have traditionally been studied independently, this paper does not confine itself to only either one of them. Instead, it seeks to facilitate knowledge transfer between the two. Moreover, the fundamentals of land use planning science are the same. More importantly, both rural/regional land use planning and urban land use planning have been governed by the current theoretical premise, sustainable development, and most analysis methods have been in use in both alike.

The literature investigation was conducted in two stages. First, a general exploration of knowledge cooperation within the land use planning community and its development trends was conducted using the bibliometric method. Given our focus is on tracking the longitudinal development of knowledge and cross-sectional cooperation, cluster analysis and temporal trajectories of knowledge cooperation were conducted based on citation statistics (we did not analyze territorial and institutional cooperation because they were not our primary focus). The bibliometric study was conducted on the 339 articles using CiteSpace 6.2.R4. Subsequently, the real cooperation between authors (i.e., citation relationships among articles) was examined. The main aim was to investigate whether articles were cited for their core research findings or simply as being sources of general literature information. We supposed that if most articles were cited for their respective core research contribution, the bibliometric method would be competent to trace knowledge development; otherwise, the traditional text reading method could be employed. Examining citation relationships within clusters revealed that articles have been referred to more for the general information they provided than for their core research contexts. This result was consistent with the findings from sampled articles that cited citation burst articles.

The outcome of the bibliometric analysis suggests that the actual development of a particular research theme/area of knowledge can be explored through traditional text reading methods. The exploration was conducted based on 25 articles, which are members of the 339 articles retrieved and were utilized to generate the bibliometric statistics. These 25 articles were selected based on four criteria: being review-type articles (all six were included), highest citation ranked (top five), recently released articles (first five), relevance to the optimal land use planning context in the Web of Science (top ranked nine articles). To select representatives of each criterion, the entire 339 articles list retrieved was iteratively sorted in descending order. To avoid redundancy, articles selected for one criterion were excluded from the list in the next step of selecting representatives of the other criteria. Figure 1 below depicts the data acquisition and entire research route.

3. Bibliometric Indications

The bibliometric analysis focused on citation and keyword patterns. The clusters were generated by setting the citation threshold value to 2 (Figure 2 and Figure 3). Four clusters each had at least 23 members, two other clusters each had four members, and the remaining 34 each had less than four members. The four largest clusters, with member counts ranging between 23 and 58, are rural land use, cluster #0 (n = 58); practical efficient regional land use planning, cluster #1 (n = 47); change in land use’s effect on small regional models (CLUE-S), cluster #2 (n = 44); and land use patterns, cluster #3 (n = 23). In these clusters, the articles citation count ranges between 17 and 49. In cluster #0, the most-cited articles are ref. [24,25,26,27,28], with repective citation counts of 49, 37, 34, 34, and 22. Articles [29,30,31] are members of cluster #1, cited 42, 18, and 17 times in their sequence. Article [32] in cluster #2 and article [33] in cluster #3 each got 21 citations (

Table A1. Citation counts by major clusters.

Cluster ID	Citation Counts	References	DOI
0	49	Stewart TJ, 2004, COMPUT OPER RES, V31, P2293 [24]	10.1016/S0305-0548(03)00188-6
1	42	Cao K, 2012, COMPUT ENVIRON URBAN, V36, P257 [29]	10.1016/j.compenvurbsys.2011.08.001
0	37	Cao K, 2011, INT J GEOGR INF SCI, V25, P1949 [25]	10.1080/13658816.2011.570269
0	34	Aerts JCJH, 2003, GEOGR ANAL, V35, P148 [28]	10.1353/geo.2003.0001
0	34	Ligmann-Zielinska A, 2008, INT J GEOGR INF SCI, V22, P601 [27]	10.1080/13658810701587495
0	22	Aerts JCJH, 2002, INT J GEOGR INF SCI, V16, P571 [28]	10.1080/13658810210138751
3	21	Costanza R, 1997, NATURE, V387, P253 [32]	10.1038/387253a0
2	21	Liu XP, 2017, LANDSCAPE URBAN PLAN, V168, P94 [31]	10.1016/j.landurbplan.2017.09.019
1	18	Deb K, 2002, IEEE T EVOLUT COMPUT, V6, P182 [30]	10.1109/4235.996017
1	17	Liu XP, 2013, ECOL MODEL, V257, P11 [31]	10.1016/j.ecolmodel.2013.02.027

).

In cluster #0, the most cited article are [24], which focused on a special-purpose geographic information systems (GIS)-based genetic algorithm (GA) for solving additive and spatial objectives; ref. [25], which dealt with boundary-based GA operators; and ref. [27], whose key contribution is the density-based constraint design. The most-citing article in this cluster [34] is a comparison of four metaheuristics and their hybrid application in optimizing urban land use plans. In cluster #1, the top-cited articles are [29,30,31]. The work of ref. [29] was similar to that of ref. [25]. The work of ref. [30] is introducing the NSGA-II, and ref. [31] integrated system dynamics and a hybrid form of particle swarm optimization (PSO) to solve land use allocation problems. The most-citing article in this cluster ref. [35] deals with land use intensity-restricted multi-objective spatial optimization and cited ref. [24] for its core contribution (of special operators), while it referred the other articles for a general literature information. Each of these four papers has little connection to the context of the CLUS-E method; they also have no connection in common. In cluster #2, the work of ref. [36] employed a hybridized multi-agent system and PSO to simulate a multi-objective optimal land use plan. Ref. [33] is a simulation of long-term multiple land use cover change (LUCC) implemented in a system dynamics–cellular automata coupling method. The key work of ref. [37] is focused on improving the application of simulated annealing (SA) with interactive stakeholder involvement in the land use planning process; none of the cited articles in the cluster are about SA. Ref. [38] dealt with a low-carbon multi-objective land use allocation simulation implemented in NSGA-II. Cluster #3 showed a similar trend.

Overall, the key finding from the bibliometric analysis is that the citation network does not necessarily reflect the core contributions of the cited sources.

In Figure 2, the plot indicates six clusters, of which four are strong. Co-citations are very strong between clusters #0 and #1, and clusters #1 and #2.

In Figure 3, the clustering map represents the strength of each cluster from 1990s to the present. Rural land use experienced the strongest bursts, which happened multiple times. With the exception of the burst times, the studies kept an almost steady momentum. This could be due to increasing advocacy for regional land use planning within the concept of sustainability. In part, it can also be explained by China’s overwhelming engagement in land use planning research for more than a decade. The practicality of the burst in around 2010 could be associated with a period of revolution in terms of method efficiency. Recently, the use of CLUE-S has risen significantly, paralleling the growing exercise of separating spatial layout activities from planning tasks. On the other hand, land use pattern as a title shows declining trend.

It is also a common experience that review authors hold the opinion that keywords can map the development trajectory of a certain study issue, such as ref. [35], who used the first-appearing keyword as an indicator. By following this method, this paper mapped the temporal trajectory of all of the keywords of the retrieved articles in two different ways, i.e., (i) choosing selection criteria “By Degree” and setting the parameter threshold value = 10 (Figure 4); and (ii) choosing selection criteria “By Frequency” and setting the parameter threshold value = 1 (Figure 5). In both cases, the density of the keyword map appears very sparse compared to the size of the dataset. The reason for this is that most authors prefer to index their works by discipline or study thematic scope-level keywords. Such high-level keywords are less likely to indicate the core contribution of the study within a specific discipline or research theme. Such discipline or research into thematic-level keywords could make the literature redundant, and obscures the exploration of the progress in our knowledge.

At this juncture, we turn to our first objective of assessing whether the bibliometric method effectively indicates the real contribution of published articles. To answer this question, first, we explore the connection between time and citations. The average publication years of the articles in clusters #0, #1, #2, and #3 were 2004, 2009, 2017, and 2005, respectively. These average publication years explain the respective average ages of the articles within a cluster. The standard deviation of publication years across the clusters is 5.12, the corresponding value for the membership size among the clusters is 12.67, and the corresponding value for the number of citations among the clusters is 62.50 (

Table A2. Summary of the largest six clusters.

Cluster ID	Size	Silhouette	Label (LSI)	Label (LLR)	Label (MI)	Avg. Year
0	58	0.74	land use	rural land use (43.41, 10−4)	land use pattern (1.65)	2004
1	47	0.777	case study	practical efficient regional land use planning (25.91, 10−4)	using accessibility map (1.85)	2009
2	44	0.849	case study	clue-s model (56.76, 10−4)	potential area identification (1.39)	2017
3	23	0.963	land use pattern	land use pattern (75.62, 10−4)	land use pattern evolution (0.41)	2005
10	4	1	a hierarchical optimization approach to watershed land use planning	watershed land use planning (16.92, 10−4)	case study (0.08)	1993
11	4	0.995	two-stage land use optimization for a food–energy–water nexus system: a case study in Texas, Edwards region	energy-water nexus system (12.42, 0.001)	case study (0.07)	2018

). These figures suggest that citation count lacks congruence with the age of the cited sources and with the cluster membership size. This finding suggests a tentative assertion that there is no strong evidence to support the claim that citation strength/count indicates the degree of relevance of the cited articles. Of course, the lack of relationships among the highly cited articles (and articles citing them) within each cluster, as discussed in previous paragraphs, strengthens this temporary supposition.

We further explore an additional 41 articles sampled from among the citation burst citing articles. Considering 6 (of a total 12) of the citation burst articles, 41 cited articles were selected randomly. For each of the 6 citation burst articles, 2 to 10 citing articles were considered (Figure A2). Among the 10 sampled articles that cited ref. [36], 3 referred to the source for their core research context, while 7 articles referred it for a general background literature information. Only 1 among 10 articles that cited ref. [28] and only 1 among 3 articles that cited ref. [39] referred to the respective source for their respective core research content. In total, 85.4% (35 of a total 41) of the cited articles referred to their respective sources for general literature information. Adding the articles discussed above in relation to bibliometric analysis (12) and to these 41 articles, 88.7% (47 of the total 53) of the citing articles demonstrate that most sources are referred to for general background literature information rather than for respective key research content.

The same is true with temporal analysis. The mapping shows the progress of the general literature information along time.

Unless each source is cited for what it specifically contributes, citation frequency (and other citation statistic information) does not necessarily indicate the relevance of the cited source article. In line with this finding, ref. [40] reports significant content variation among the top 10 papers with the highest local citation scores. This finding indicates a potentially severe interpretation flaw regarding consistency of trends or content. Now, it is safe to suggest caution regarding the potential risk of relying solely on the citation-based bibliometric method if the objective of the review is making a critical synthesis or exploration of actual new developments or emerging issues. However, this article never argues against the use of the bibliometric method in tracing knowledge development if titles and keywords reflect the key contribution of the article. In line with this, the current paper indicates that the top cluster, “rural land use”, was a prominent research theme during 2000–2010; “land use pattern” was the hotspot of 1995 to 2000 and 2005 to 2010, and tended to weaken thereafter; during the period of 2010 to 2015, “practical efficient regional land use planning” was at its peak; since 2015, application of CLUE-S has been growing, but its citation burst is modest (Figure 3).

In conclusion, while the bibliometric method is often considered robust [41], the findings in this paper suggest that most articles are not cited for their core purpose or contribution. Therefore, reliance on the mechanical summary statistics in bibliometric analysis may be insufficient to investigate past trends and identify possible frontiers in land use optimization. Therefore, sections four and five of this paper present the application of an explorative reading.

4. Previous Review Works

4.1. Brief Summary of Content Coverage

Only six among the retrieved 339 articles were literature studies, of which four were related to the use of metaheuristic methods in agricultural land optimization. Refs. [7,40,42,43] discussed the subject from different perspectives. Refs. [18,43] are about urban land use optimization.

These articles provide valuable insights into various optimization methods and into land use planning contexts. They make an immense contribution, exposing readers to comprehensive knowledge within the coverage of the articles. Given all six review articles have been considered, a simple comment regarding the coverage of what has been written about land use planning within the Web of Science over the past two decades is that 100% of the search parameter are met; this signifies that what has been reviewed within the scope of our search is already known. In this section, a concise summary of each article is provided, their composite value is summarized, and frontiers within the scope of these six articles are identified.

The authors of Ref. [42] are concerned that agricultural intensification is creating the challenges of pollution and biodiversity loss, which thus limits its capacity to meet food demand. According to the authors, land use optimization is the upcoming frontier solution to the issue of food demand. They emphasize the need for high-credibility optimization methods to achieve sustainable land use plans that balance the needs of the ecosystem and the economy. Accordingly, their study numerates the metaheuristic methods that have been applied in agricultural land use planning case studies, explores factors that determine the success of the methods, and explores the alternative mechanisms for the involvement of stakeholders in the planning process. They reviewed 50 articles (38 were case studies), following the PRISMA method [43]. Reportedly, simulated annealing (SA), taboo search (TS), evolutionary algorithms (EA), differential evolution (DEq), and swarm intelligence (SI) have been in use. Researchers often select methods based on previous success stories of similar problems. Posteriori stakeholders’ involvement, where trade-off solutions are filtered based on multicriteria analysis, has long been common practice. Yet, a great majority of other studies suggest that the stage at which stakeholders participate in the planning process is subject to the nature of the problem; ref. [43] discusses more on conditions of stakeholder participation.

Countering a given land use planning problem, the key contribution of the paper is that it provides a general guideline for the selection of an optimization method. The suggestions of the paper are based on the characteristics of a given problem and the method applied in the studies reviewed. In cases wherein a greater number of constraints may limit the search space navigation, SA and TS tend to perform well. SA is preferred for its parallelization capacity, while TS is an effective local search for improved solutions in the immediate neighborhoods of the current solution by penalizing for revisiting already visited neighborhoods. On the other hand, evolutionary algorithms are recommended for handling multiple conflicting objectives. One cause of the scalability challenges of heuristics is that algorithms are problem domain-dependent; if not, algorithm design should be tailored down to the level of the problem’s specific context. Hybridizing two or more methods is a widely applied solution to deal with the scalability challenge arising from the problem-specific nature of metaheuristics [40]. The authors of Ref. [42] demonstrated that combining global optimizers and local search methods could effectively solve multidimensional combinatorial land use optimization problems. The hybridization complements the relative exploration strength of one and the relative exploitation strength of the other. Despite coupling being well-known alternative, parallelization and the use of heuristic-type operators are other alternatives available to deal with challenge of scalability.

Ref. [43] extended the discussion of method selection by adding decision variable type as a criterion. They argue that exact methods are suited for discrete-type variables, and heuristics are compatible with continuous variables. Mixed-integer programming (MIP) is better suited for handling mixed variables. The approach of Ref. [43] to the optimization method selection is more scientific. Instead of relying solely on tabulating a matrix of problems and methods applied to solve them and then establishing a generic judgment regarding which method(s) could better solve a given nature of a problem, they approached the problem by examining the strengths and drawbacks of various methods (TS, SA, SI, genetic algorithm/GA, artificial immune system/AIS, linear programming/LP, and fuzzy programming/FP); objective formulation type (Pareto-based versus the variants-of-aggregation method; weighted sum/WS versus goal programming/GP); the constraint-handling mechanism (penalty function versus defining feasibility criteria); and the conditions of stakeholder involvement (priori, interactive, posteriori). Consequently, they suggest a well-structured hierarchical evaluation process to select an appropriate method. The first step is identifying the nature of the problem. If the problem does not require extensive trade-off analysis, priori methods (WS, GP/RP, TS, and SA) are preferred, and iterative stakeholder involvement is convenient. If the problem requires extensive trade-off analysis, the decision as to whether objectives have to be scalarized or the problem has to be converted to ε-constraints form is made. If possible, Pareto-based methods are recommended, with stakeholder involvement typically being posteriori.

The scope of ref. [7] is broader, addressing both the method selection issue and issues of spatial measurements and drivers of land use change. From a spatial development theoretical standpoint, the authors highlight the importance of considering the mutual feedback between humans’ socioeconomic activities and the eco-environment in the context of land use change. This mutual feedback conceptualization recognizes that socioeconomic activities could cause land use change, which in turn affect the eco-environment. The reverse is also true. Changes in the eco-environment could influence the spatial patterns of humans’ socioeconomic activities on land. This mutual feedback conceptualization helps to identify and characterize factors contributing to spatial land use change. Increased attention paid to the effect of ecological limitations on human activity on land means that the driving factors of the analytic paradigm of land use spatial change are of equal weight in terms of economic growth and restraint. At the other end of the spectrum, mutual feedback also involves optimizing land use for a given activity, considering land use change as a factor.

In terms of methods, the authors argue that the CLUE-S and Markov models are effective tools for integrating socioeconomic and natural driving factors. These coupled methods are capable of handling conflicting land use relationships. Planning models such as LP, system dynamics (SyD), and multi-objective programming (MOP) facilitate analysis by quantifying economic and social driving factors and determining the equilibrium between demand and supply. Since different problems require different spatial resolutions and different spatial resolutions return different outputs, the authors suggest appropriate level of spatial abstraction and use multiple data types/sources to redress possible gaps while analyzing the land use change. Incomplete individual datasets may further trigger the use of multiple data sources/types.

The study conducted by [7] focuses specifically on the application of GA in multi-objective land use optimization. Using CiteSpace 5.8.R3 software, the authors map the knowledge collaboration among countries and institutions based on 1154 articles retrieved from multiple databases. They also traced the temporal development of GA knowledge by mapping the frequency of the first-appearing keywords. According to their findings, the period from 1995 to 2004 marked the mainstreaming of GA methods for land use optimization. From 2004 to 2008, much research focused on optimizing the GA itself. Hybrid application of GA dominated the years 2009 to 2016. Since 2017, the literature has progressed towards deepening the integration of GA with big data. Consequently, the authors suggest three future research directions including further integration of GA with emerging AI capabilities, incorporating a temporal dimension into land use optimization to cope with the dynamic nature of land use change, and integrating GA with broader knowledge and practices in the field of land use planning.

Both Ref. [18] and Ref. [43] assessed the literature on urban land use planning. The focus of ref. [18] is the representation and formulation of objectives of the sustainable built environment, including contiguity, compactness, and compatibility. The authors examined 55 articles following the PRISMA protocol. Reportedly, the compactness objective appeared in 16.67% of the studies, the contiguity objective in 13.67%, and the suitability objective in 11.9%. Aggregating the objectives according to the three pillars of sustainable development, economic objectives appeared in 46.67% of the studies and ecological/environmental objectives in 43.33%, signifying a comparable emphasis on economic development and ESs in urban literature. The representation of social objectives was only 10%. The majority (42.86%) of studies applied Pareto-front-based objective construction; 36.73% applied weighted sum, and the remaining 20.41% followed goal programming. With regard to methodological approaches, 80% of the studies surveyed applied GA in its various customized forms, followed by PSO (12.73%) and ant colony optimization (ACO) at 9.1%. About 80% of the cases depended on raster data models.

Given the generic nature of the bibliometric method, the paper, however, falls short of evaluating concepts/theories that influence the identification and formulation of objectives and the performance of the various optimization methods encountered in the articles studied.

The review by ref. [44] is more specific. It examines the establishment of compactness and contiguity objectives that characterize the structure/configuration of a sustainable built environment. Supposedly, compactness is achieved through consolidation of the same uses to form large clusters [27] enforced by various mathematical formulations including perimeter-to-area ratio, diagonal length of the minimum bounding rectangle, the weighted average ratio of area to the square of a perimeter, core–buffer cell assignments, minimizing the number of clusters, or maximizing the largest cluster. Alternatively, constraints such as setting a minimum threshold of a certain land use cell in a neighborhood may be imposed as a precondition to allow additional allocation of a specific use type to a neighborhood. The modeling of contiguity, on the other hand, is often based on graph theory operationalized using a path-based model, order-based model, or network-based model alternatives.

However, whether promoting larger clusters of the same uses is in line with the conventional wisdom of sustainable built environments is questionable, and is considered in next subsection.

Collectively, the review articles provide a wide-coverage summary of the land use planning literature both in the context of methods and in planning knowledge. Yet, there exist critical gaps left unaddressed. The first limitation is that the literature heavily dwells on methods, rather than on planning theories and conceptual discussions. Among the six review articles, only ref. [7] gave meaningful weight to planning concepts, which underscores that the effect of human socioeconomic activity on the ES is not the only aspect of the spatial planning domain. The mutual-effect feedback concept seems very important in deepening our understanding of sustainable development. Yet, whether the spatial change in activities caused by ESs change ends with sustainable solutions seems an emerging new direction for discussion and inquiry. Despite being heavily method-focused, the reviews also left many methodological concerns unaddressed. First, they lack standardized frameworks to assess and compare the performance of the methods they encountered while reviewing studies. Second, they did not notice that different performance indicators may rank the same set of methods differently. It is also doubtful whether computational time cost will prevail over planning quality. There is no theoretical ground to benchmark the trade-off between computational time and solution quality. Third, there is a tendency to assume tools/methods that evaluate data efficiently according to defined instructions are end-solutions in and of themselves (see ref. [42]). Instead, we argue that methods only facilitate the handling of huge data, and ease the analysis according to one’s definition of the problem. For a method to solve a problem effectively, conceptualization of the problem is the basic concern, which calls for detailed knowledge of land use.

4.2. Value and Frontiers

The six reviews discussed above explored various land use planning concepts and methods. These studies systematically analyzed the strengths and weaknesses of different global and local land use optimization methods spanning evolutionary algorithms (EAs), SA, LP, and integer programming (IP). By consolidating and summarizing the findings, these studies provide valuable insights and guidelines for selecting an appropriate method. Regarding method selection, two distinct approaches have been identified. The first approach is establishing an association of the problem domain and the applied method type based on previously published studies. This approach is, however, overly simplistic, and the success rate of obtaining appropriate methods is lower. The second approach involves a scientific analysis of the nature the problem domain and the construct behavior of the optimization methods. Additionally, decision making time constraints and solution quality criteria, such as diversity and reliability, are important considerations for the method selection [45,46,47,48]. Another factor that influences method selection is problem size. Exact methods are less suitable for larger problem sizes, e.g., exceeding 200 [24]. This is primarily due to the complex interrelationships between geographic and ecology, as well as the characteristics of the geographic units (e.g., shape, distance, adjacency, etc.). The type of variable involved is also a determining factor in method selection [18]. However, it is important to note that an objective and comprehensive analysis considering all these parameters is still lacking for method selection. A thorough and objective analysis would greatly enhance our understanding of the available methods, and allow for a more concrete selection strategy.

Since the contextualization of a certain land use planning problem may belong to a certain conceptual/theoretical/framework and the mathematical formulation of a problem is often distinct, the method selection strategy suggested by ref. [42,43] still remains one including generic guidelines. This fact is supported by the fact that different studies reporting different findings and presenting different interpretations offer the same method(s) applied to similar problems. A study by ref. [49] assessed the performance of a nondominated sorting genetic algorithm II (NSGA-II), PSO, and a multi-objective evolutionary algorithm (EA/D) based on solution dispersion, the diversity of the solution space, and the number of dominant solutions in Pareto-front parameters. Reportedly, PSO exhibited the best diversity of solutions, while the EA/D outperformed the other two in terms of computational time. In the study by ref. [50] that compared LP, SA, mult-objective land use allocation (MOLA), and multidimensional choice for land use planning (MDCHOICE), reportedly, MOLA performed better in landscape metrics, LP returned the best demand requirement convergence level, and MDCHOICE had the best ability to harmonize land uses (i.e., diversity of the solution space). It is also very interesting to encounter a situation wherein a certain method may be biased to a problem type. The same study reported that SA was the weakest for the metrics mentioned above in all the test problems, but excelled in solving the forestry land use allocation problem. Ref. [51] assessed the performance of GA, Cuckoo Search, and PSO in agricultural land allocation problems, aiming to diversify crop plantation plans. The authors reported GA’s superior capabilities, as evidenced by 103% simulated crop yield growth and 97% simulated profit growth, while simulated water consumption still reduced by 5%. In a study that compared Markov–cellular automata and Markov–cellular automata–CLUE-S hybrid models, both were applied to a land use problem with the aim of maximizing ESs values; in [52], the authors reported that the non–stationary Markov state transition and cellular automata (CA) probability hybrid returned a high level of layout precision.

In general, various studies indicate that the performance of different methods depends on the chosen comparison metrics. In a parallel manuscript that explores the nature of GA in land use optimization, produced by the authors of this paper, findings from a semi-systematic literature study indicate that the effectiveness of GA (if GA represents the major characteristics of evolutionary algorithms) is problem-specific. Findings from the aforementioned reviews and another review by the authors of this paper alike indicate that relying solely on general literature information and historical trends of the usage of a method is less likely to guarantee an appropriate method. Since analyses of artificial intelligence-based optimization algorithms are limited [52] and a conclusive environment to evaluate an optimization method is not available, a detailed and objective parameter-based method selection strategy may be required. It may be necessary to concretize the generic judgments suggested by these studies into comprehensively parameterized and objective indicators, based on the classification of available methods.

Despite the fact that the review articles are devoted to optimization methods, they also contribute to the advancement of concepts in spatial development planning. Ref. [52] suggests that the depth and breadth of sustainable urban land use optimization are barely explored. One of the limitations highlighted is the low (only 10%) representation rate of social development-related objectives. It is worth noting that studies that encompass objectives that represent the three dimensions of sustainable development are rare; only two (of total 55) studies included social development dimension-related objectives (in a review by [18], for instance). In studies that consider the social dimension, the representation is often vague, such as a form of social security service values or accessibility to main roads [36] and spatial compactness [53], wherein the representation creates confusion with geographic compactness objectives that address the ecological environment. Another significant limitation worth mentioning is the lack of standardization or consensus concerning the use of proxy variables to calculate the benefits of the objective parameters of each sustainable development pillar. These gaps provide opportunities for further research, especially that which establishes consensual objectives that represent social development and proxy measurements to gauge the performance of each pillar of sustainable development. This can be made more concrete by developing protocols that standardize the measurement of ES values and social development.

5. Review of the Regular Articles

As with any regular research article, a literature study has to be bound by a defined scope. Each review study addresses a specific aspect of a given research theme in a specific discipline. In this regard, it is natural for the review articles assessed above to leave much unaddressed regarding planning concepts/theories and methods of optimizing land use allocation. This section explores the development of the general literature following a chronological sequence, the whereabouts of basic land allocation theories/conceptual models and utility models within the popular optimization-based land use planning literature, and the state-of-the-art of optimal land use planning methods. Potential frontiers for research and discussion are provided, following a summary of the existing state of the literature.

5.1. Literature Path Building

Based on the logical assumption that any research project builds upon existing knowledge [22], this subsection aims to evaluate the development of the general land use planning literature over the last 20 years. The task has been accomplished by analyzing the topical issues studies identified in their review/introduction sections, in chronological order. In other words, the task is that of establishing a matrix of time and the emergence of new ideas in the land use planning context (or the evolution of methods).

Table 1. The trajectory of the development of optimal land use planning general literature 2002–2023.

Ref. Time	Identified Gap	Suggested Direction/Implications
2002	Optimization methods are problem-dependent. No generalized behavior was established	Improve the efficiency of algorithms through comparison of multiple scenarios
2008	Mismatch between optimization methods and planning perspective. i.e., assumption of the determined time	Intertemporal approach
	Global optimization-implemented objectives used at the general management level(this problem still persists in urban land use)	Detailing objectives to a level at which quantifying value resources is practically possible
2015	Coupling was not mature enough	Broadening the application and undertaking more research on hybridization approach
	Limitations of local scale optimizers (game theory) applied independently	Hybridizing local optimizers with global optimizers to take relative advantage of exploration and exploitation
2017	Determinate assumption of constraints	Modeling uncertainty
2018	Any trade-off considered acceptable/a feasible alternative	Minimizing the magnitude of trade-offs among objectives is a quality advantage
2020	Land use change driving factors not considered	Building the probability of land use change factors into simulations
2021	Global optimizers lack layout capability, while local optimizers lack structure capability	Coupling top-down and bottom-up methods attained a normative stage of application
2023	Spatial layout determined by local optimizers is affected by the historical trends of the land use change process	Spatial suitability analysis/horizontal process
	The heterogeneous nature of spatial units providing ESs is affected by the historical transition record matrix of the logical transition rule	An emerging issue open for inquiry

provides a condensed summary of the historical trends in progression.

Land use optimization can be applied at different spatial scales including river basins [15] and cities [36], and at global [54], continental [55], or national scale [56]. Rural/regional land use optimization primarily focuses on managing ESs or attaining a balance between ESs and economic benefits. On a regional level, the aim may be to strike a balance between construction/development activities and ecological preservation [9]. Yet, more effectual land use planning may be developed at local resolution scale, for example, in a city or a region of a constellation of cities, a watershed, or in a manageable administrative unit such as a county or district.

In a study conducted by [57], various simulation models used in ecosystem management were assessed. These models included stochastic variation, variants of LP, 0-1 IP, nonlinear programming (NLP), Bayesian belief network (BBN), hierarchical dynamic programming (HDyP), SA, and spatial dynamic programming (SDyP). The study highlighted that optimization techniques in environmental modeling are highly problem-dependent, because ecosystem model conceptualization and spatial complexity vary from one problem to another problem significantly. As a result, the solution the authors recommended is improving the computational efficiency of a solver rather than iterating experiments across multiple implementation algorithms. The authors’ implicit conclusion is that the methods are structurally the same; however, they are not.

The authors of ref. [58] investigated the use of econometrics, mathematical dynamic programming, and GA for land use optimization in dynamic environments. They highlighted that most optimization models are detached from the perspective of land use planning. These models generally assume a static state at a specific point in time, whereas land use planning extends over a future time period. Ref. [59] also discussed the application of various optimization techniques including conventional programming constraint, chance-constrained programming, LP, multi-objective LP (MOLP), genetic programming, and the combination of LP with artificial neural networks (ANN) and GIS. Reportedly, the methods were employed to evaluate problems involving different sets of objectives, constraint assumptions, contexts, and spatial scales, and the studies were applied in various countries. The authors observe that optimization techniques have primarily been used to support general management systems in land use allocation. Consequently, the authors argued that the practice of translating available optimization models into practical problems is rather limited. There is knowledge gap regarding how to link practical issues with the optimal utilization of all land resources under conflicting demands, specifically at watershed level.

When it comes to modeling land use change, ref. [45] discussed several methods, including multi-agent systems (MAS), generalized LP, unbalanced support vector machines, and aggregated multivariate regression. They also examined Markov chains, multivariate statistics, SyD, CLUE-S, and CA. They argued that coupling is not exhaustively utilized, despite the potential hybridization of two or more methods being very wide. Consequently, they proposed the combined use of ACO, Markov chains, and CA in their own study. Within their discussion of a wide range of land use optimization methods, ref. [14] emphasized Game theory (GT) and GA. GT allows for the simulation of the behavior of various stakeholders, assisting in decision-making in a conflict-affected environment, but it lacks the ability to execute global optimization [14]. This limitation can be addressed by coupling it with other global optimization techniques [45], which are already in wide use.

It is worth noting that the aforementioned studies highlight the application of different land use optimization and land use change simulation methods, including variants of swarm intelligence, evolutionary algorithms, structured mathematical models, agent-based models, system dynamics, various land use change transition models, other auxiliaries, and different combinations. Combining these methods may offer more comprehensive and effective approaches to land use planning and decision making in the complex environment of land use allocation.

The incorporation of uncertainty has been acknowledged in various disciplines, such as water allocation, waste disposal site allocation [60], agricultural plantation [61], and other land use applications [62,63]. However, much of the existing literature tends to assume deterministic conditions and overlooks dynamism [64]. Neglecting uncertainty in land use planning may compromise the scientific credibility of studies and lead to plan implementation failures. To address this issue, the authors of ref. [64] emphasize the importance of systematically analyzing the sources of uncertainty, which is a growing experience among Chinese scholars, which emerged from their understanding of the failures of past planning methods that assumed exact constraints. Another widely discussed strategy is the transparent involvement of stakeholders’ engagement in the planning process. Where uncertainty is acknowledged, failures arise either from the lack of systematic methods of analyzing the sources of uncertainties, or from the absence of transparent stakeholders’ engagement during decision making regarding objectives [14].

While ecological restoration programs like the Grain for Green Program in China have achieved significant success [15], previous studies on ecosystem-based optimal land use allocation studies have primarily focused on quantifying ES values, assessing trade-offs between different ES values and identifying the factors influencing them. However, the limited consideration given to minimizing trade-offs among services is a crucial planning concern. Establishing an explicit objective that aims to close the trade-off gap among the attainment target parameters will enhance the sustainable use of available land resources from the perspective of a rational balance of products (economic and ES). An explicit reduction in the magnitude of the trade-off gap, rather than looking for land use configuration(s) to meet a minimum requirement of certain land use purposes, may sound more rational in the context of land use planning.

Among common land use simulation models, the authors of ref. [65] discussed the use of CLUE-S, future land use simulation (FLUS), CA-Markov, and CLUMondo [66]. The authors observed that FLUS is not a well-developed method. Uncertainty has not been considered in studies that use FLUS.

The authors of ref. [8] examined the distinction between global optimizers and local simulation methods. Their study highlights that top-down methods, although suitable for handling policy constraints, may not yield precise spatial layouts. Bottom-up methods, in contrast, operate transition rules to assign suitable locations to activities; however, they generally lack accurate quantity determination capabilities. The significance of the global versus local optimization categorization of methods is that addressing the compelling rationale behind the use of hybrid methods is not a mere scalability problem. Moreover, the disaggregation of quantitative structure determination (demand and supply balance) and the determination of the spatial layout of activities into two cooperating sub-models in the land use planning process triggers a logical reason for the coupling of global and local optimizers. In line with this perspective, land use optimization practices often concentrate on providing structural-level quantitative solutions that require further adjustments locally [67,68]. Attention should be paid to the limitations of local spatial layout optimizers such as CA and CLUE-S, wherein the historical record of the land use transition matrix determines the final layout of the activities. Further optimizing the layout of activities for suitability, considering heterogeneity, can address some of the limitations. Heterogeneity can be attained by filtering the alternative layouts based on a multicriteria decision-making process coupled with stakeholder participation. In opposition to this line of analysis, ref. [68] argues that current approaches that incorporate land ecological suitability into land use planning, especially in urban areas in which the major focus is on predicting ecological space regardless of the level of socio-economic and ecological trade-offs, are inadequate. Traditional methods of land ecological suitability assessment that only address vertical processes are insufficient in addressing the horizontal competition processes.

Urban land use planning and regional land use planning differ in their mainstream conceptual foundations. The methods’ applications also tend to differ. Rural land use planning focuses on optimizing various ESs, or on the trade-off between socioeconomic attainment and ES value necessities. In contrast, urban land use planning is treated on the ESs consumption side of the equation. In regional/rural land use planning, local optimizers are commonly employed either alone or in conjunction with global optimization techniques. Urban land use planning primarily relies on global optimization methods, with limited emphasis on bottom-up approaches.

Unlike in regional/rural land use planning, in the spatial development context, planning methods are not adequately differentiated in urban land use planning; this often leads to problems, such as general management strategies wherein the land use demand and the spatial layout maps of activities are determined to be a unified problem. Nevertheless, the literature on urban land use planning has made some progress too.

Efficient land use and a layout of activities that facilitates a smooth traffic flow are critical concerns involved in urban land use planning, because inefficient resource use and traffic congestion are mainly the outcomes of the pattern of spatial growth rather than the magnitude of growth [69]. This analytical perspective has strengthened the spatial analytical view of the sustainable development of the built environment. Following this line of thought, a number of studies have explored sustainable forms of the built environment, spanning morphological structures [16] and the intensity of land use [27]. However, the concept of sustainability is not captured comprehensively, especially in the urban literature. Some studies argue that focusing on objectives or constraints related to compactness and compatibility is insufficient to address the idea of sustainable cities in a meaningful way. In response, ref. [27] proposed a density-based design constraint that promotes the consolidation of similar land uses, though the suggested solution echoes the problem itself in a different format.

Studies by [29,70] have primary focuses on methodological approaches, with a similar focus in their respective introductory reviews. The former explored an LP-GIS coupled method in the work of [71], SA-LP in the work of [26], a density-based design constraint in [27], and vector-GA and GP-GA methods in [24,72]. The authors criticized the inefficiency of grid-based GA for large combinatorial land use optimization problems, which has already been recognized in the literature. To enhance efficiency, studies have suggested the use of heuristic operators such as boundary-based crossover, boundary-based mutation, and constraint steering mutation.

The core study domain of urban morphology research is spatial separation, including the separation of households and workplaces [73,74], accessibility studies of various forms [75,76,77,78], and sprawl [79] analyses of the spatial separation of residents from workplaces and a wide array of consumption services. However, the integration of transportation into land use optimization studies has remained incomprehensive [71,72,73,74,75,76,77,78,79,80]. The relationship between land use and work commuting, as well as other travel purposes, is frequently overlooked. In many structured mathematical models, the primary focus is on the separation of workplaces and residences, and the interplay between transportation and other land uses is often neglected. An unfortunate reality is that travel behavior has been studied without adequate consideration of the effect of land use types on mobility behavior. As a step towards enhancing land use planning, ref. [80] proposed a model for residential location selection using a case study of a new development area. While there are a few studies that include transportation elements in land use optimization, they typically use Euclidean or rectilinear distance to roads [27,31], or a decreasing distance function based on road service capabilities [25,26]. However, Euclidean distance-based accessibility studies are often criticized for their inability to capture the role of road networks in determining the spatial configuration of activities and physical morphology, as highlighted in the transport-oriented development literature [79]. Among a few studies that are critical of the issue, the one by ref. [13] is exemplary. The authors developed a composite modal infrastructure-based accessibility modeling approach and applied it to maximize workers’ quality of life and the productivity of public facilities in a new development area. As mentioned elsewhere in this paper, the discourse around sustainable city forms has strongly influenced the urban land use planning research environment [81]. In this context, Ref. [70] discusses possible zoning approaches. Two options mentioned are tree and semi-lattice structures [82]. The semi-lattice structure is considered a natural method of city growth. However, ref. [70] indicates that its practicality is a hotly debated issue due to the challenges involved when manual calculation of the connections between different land uses is considered. Instead, they proposed a zoning determination method based on the groupings of different land use types. This alternative may simplify the zone delimitation process and enhance the efficiency of urban land use planning.

Overall, the sample articles demonstrate the clear establishment of a body of literature within which successive studies identify gaps that previous studies have left unaddressed in debates and studies over time. The literature has remained future-oriented. In the planning concepts/theories context, the determinate assumption persisted well into the second half the 2010s, and currently appears to be under the comprehensive control of the literature. On the other hand, the assumption of the equivocality of any trade-offs, modeling uncertainty, and drawing land use change driving factors into spatial development planning analytics, the effect of historical land use transition in determining the final spatial layout of activities, which ultimately affects the heterogeneity level of spatial units providing ES, etc., is an active hotbed of inquiry. A more important limitation is that the progress of urban land use planning lags behind its regional/rural counterparts.

From a methodological perspective, selection of a well-performing single method has been replaced by designing hybrid methods. The number of such methods has grown notable, and the depth of method coupling also has recorded promising advancements. Yet, limitations are expected. First, we observe that recent publications claim an already solved problem, as if freshly indicating an information gap across territories. Second, we observe confusion between construct differences among various optimization/simulation methods and scenarios (see, for instance, [57]). Third, the wide application of hybrid methods tends to represent a hegemonic drift, according to the claims of the general literature. In other words, studies simply claim that the hybrid methods they applied perform well, without benchmarking any single method. If any reference is available, the reference most often made is a general literature claim. At best, authors refer to conceptual discourses to justify the use of the hybrid method(s) they applied, instead of conducting their own experiments. Finally, the application of hybrid methods is expanding horizontally, but without comparison among them.

5.2. Land Use Planning Context and Method Development

This section explores the progress of planning knowledge and planning methods based on 19 regular articles sampled. Although distinguishing land use optimization studies into planning knowledge-oriented (theories/concepts associated with land use allocation and utility modeling) and method-oriented domains is difficult, we categorized the articles based on the relative depth of the domains and the study objectives that the authors stated explicitly in their abstracts, introductions, or conclusions. As reported in

Table 2. Categorization of the sampled 19 articles by the focus of their core content.

Ref.	Domain	Applied on	Core content	Objectives	Method
[8]	Planning concept	Large region	Effect of land use change on ES	Economic benefits Max. ESV	GMOP and PLUS
[10]	Method	Urban agglomeration	Embedding land use optimization in ecological suitability	ESVs land use suitability	MOLP; DyCLUE; MCR
[13]	Planning concept	City	Accessibility model in land use planning	Max. Accessibility; Max. Compactness; Max. Suitability	NSGA-II
[14]	Method	Urban and rural region	Coordination of land uses at local level	Max. Suitability of land for a certain use and Max. Compactness	GA; DyGT
[15]	Planning concept	Watershed	Effect of land use change on ES	Max. Agricultural production; Max. Sediment retention; Max. Carbon sequestration; Max. Water quality; Max. sustainability of water production	InVEST; Biophysical models
[27]	Planning concept	City	Compact form of the sustainable city concept in land use planning	Min. Open space development; Min. Redevelopment; Min. Distance of new development site; Max. compatibility	GIS-MOLA
[29]	Method	City	Efficiency of NSGA-II for implementation	Max. GDP; Max. Environmental benefit; Max. Ecological suitability; Max. Accessibility; Max. Compactness; Max. Compatibility; Min. Use conversion; Max. NIMBY	NSGA-II
[45]	Method	District of a city	Hybrid optimization method for modeling land use change	NA	Markov-CA; ACO-CA
[57]	Method	Watershed	ES-based optimization under different land use management scenarios	Min. fertilizer use, Min. nutrient outflow and Max. economic yield	Monte Carlo; GA
[58]	Planning concept	Management farming	Temporal dimension of land use planning	Max. income	GA
[59]	Planning concept	Watershed	ES-based optimization	Min. soil erosion and Max. Economic benefit	Simplex-LP
[64]	Planning concept	City	Uncertainty incorporation in land use planning	Max. GDP Ecological benefit (ESV)	GA
[65]	Planning concept	Large region	Land use change driving factors; Probability surface-based land use optimization	Priority of land use type i	CLUMondo, BBN
[66,67]	Method	Large (rural + cities)	Application of hybrid methods for land use optimization	ESVs	DyMOO; CLUE-S; MCR
[69]	Planning concept	City region	Method of integrating ecological benefits into land use planning	ESVs Land use suitability	MOOLP CLUE-S
[70]	Method	City	Zoning mechanism in land use planning	Max. Compactness; Max. Compactness; Max. Dependency; Max. suitability	PSO; GA; Local search
[80]	Planning concept	City	Residential choice model in land use planning	Max. Quality of life for workers and Max. Productivity of facilities	GA
[83]	Method	Urban region	Application of hybrid methods for land use optimization	Max. Farm production; Max. Water yield; Max. Habitat quality; Max. Sediment retention; Max. Recreational quality; Max. Aesthetic quality	SA-GA

, nine of the articles focused on spatial development planning context, and the remaining are more method development-oriented.

5.2.1. Land Use Planning Context

The rural and urban land use planning literature operationalizes the context of sustainable development in different ways. The former is mainly operationalized via the ES accounting concept. This concept brings the representation of real-world problems close to a level that can be tackled using relatively easily measurable parameters. On the other hand, urban land use planning problems have been formulated at the general management scale. Another key difference is the extent to which a given problem is established in reference to relevant theories/concepts. In both aspects, the literature indicates the availability of important knowledge in the regional/rural land use planning literature that can be translated to/adopted by the urban counterpart. This subsection is devoted to these points; first, we discuss the state of available knowledge, and then we identify potential areas that may initiate inquiries.

Current State

There is increasing understanding that various potential users of land require different time spans to satisfy their respective interests, leading to a fundamental development in land use planning; this is the transformation of the spatial dimensional planning problems into spatiotemporal dimensional problems. The authors of ref. [58] considered an optimal allocation of agricultural land that maximizes income from sheep and grouse mixed farming activities under two land management alternative scenarios (tenant farmer managers and landlord managers). In each land management scenario, the manager can choose five activities from a combination (1 to 5) that maximizes their income. In the capacity of government, the key optimization problem is reconciling the short-termism tenant farmers’ income that is supposed to be assessed on an annual basis, and that of the proprietors, who look for medium- to long-term investment returns. The reconciliation of the interests of the two groups/scenarios is a problem of temporal conflict over spatial development. Under the tenant farmers’ management, land use change is supposed to occur in every contract term, possibly every year. Under the landlord’s management, the land use change time would be determined by the cumulative return assessment for a period of t years. Since a one-time (temporal) solution produced in a static situation will not respond to the landlord management scenario, a five-year period and 10-year intertemporal alternatives were considered. The intertemporal model is a dynamic model that inputs dynamic variables (with grouse population, grouse price, and government subsidy all changing periodically).

The results demonstrated a strong correspondence between the length of management and the stability of land use change. Under the temporal scenario (tenant farmers’ management), land use changed nine times after subsidy reform, which is not ideal for proprietors. Within the five-year intertemporal option in the landlords’ management scenario, a three sheep/hectare farming system would become more profitable before the subsidy reform program ends. In the next two five-year terms, land use would change seven times. Under the 10-year intertemporal landlord management scenario, three sheep/hectare would be more profitable, and land use change would occur only once within 14 years. After the end of the subsidy program, grouse shooting would become a dominant activity accompanied by more stable land use. The findings indicate the frequency of land use changes that respond to decision makers’ and to land managers’ understanding of how land use planning changes over a defined period. Regardless of its application potential in policy decision making, the temporal component is underrepresented in the literature.

Uncertainty programming enhances resource allocation rationality. Flexibility accommodates various aspects of unpredictable elements spanning inputs, processes, and outputs. Consequently, it is another concept gaining a strong foothold in rural/regional land use planning research. The continuously changing land use structure involves spatiotemporal characteristics [65]. Socioeconomic, biophysical, and biological factors cause changes in land use. Such change parameters need to be captured periodically. Despite the fact that concern is growing, the static assumption still dominates sustainable urban land use planning. At this point, one can think of the temporal dimension as being associated with the accession and succession relationships of businesses and households within bid–rent land allocation theory. A corresponding practice can be referred to in timber harvest scheduling [84] which concerns spatial harvesting patterns over multiple periods of time [85]. Refs. [65,66] also incorporate other variants of temporal uncertainties. Although the practice of uncertainty modeling is common in land use optimization studies, ref. [64] argues that the predecessors to their work lacked a systematic approach to the determination of uncertainty types and sources. In their own study, they integrated uncertainty elements within both the input parameters and the objective functions. Input parameter values were simulated stochastically between the upper and lower demand targets determined based on historical data and experts’ evaluations under different statistical significance level scenarios. For the output (objective function values), the range of the value was constructed based on the 10,000 best solutions sampled from each run. The study by [8] analyzed the spatiotemporal evolution of ESs in association with LULC change. They optimized both ESs and economic benefits under three scenarios (business as usual, ecological priority, and balancing economy and ecology). They modeled the uncertainty of input parameters by applying gray multi-objective optimization programming (GMOP). Authors of ref. [15] also modeled uncertainty via gray value objectives and constraints. The periodic oscillation in [14] exposes readers to a different perspective of uncertainty. In the study, the geometric progression element in the dynamic GT created within the compensation addresses the dynamic interaction of parties involved in the land use planning process, where the current decision of one party determines the next decision of the other. The elements of uncertainty in the work by [15] are of two types. In their study of the effect of land use change over ES, the authors applied a range of input values and experts’ opinions based on a multiple-criteria analysis in order to filter final solutions. In their work, the objectives were water supply sustainability, water production sustainability, soil retention, carbon sequestration, and agricultural production, evaluated over 6859 land-unit scenarios (created by combining different slope grades, land use types, and water constraint levels). The study’s important value addition is its stakeholders’ posteriori involvement, via multi-criteria decisions, in selecting planning options. The decision variables were the attainment of optimized objectives and the number of sub-watersheds that meet the minimum water demand threshold required for the maintenance of socio-economic activities. Applying the InVEST and biophysical models, they found solutions for ES trade-offs in five scenarios; each scenario improved on all objectives. The importance of the study also lies in its demonstration of the importance of heterogeneity in land use policy in closing the magnitude of trade-offs among Ess, and promoting more rational sustainable development.

Unlike the regional land use planning problem, its urban counterpart has to respond to a multitude of conceptual models/theories governing land allocation behaviors that are more complex than the ESs in the former. Although urban land use optimization problems are often generic, some works suggest some degree of progress in conceptual modeling. Unlike predecessors, the authors of ref. [27] considered the overall compactness of the built environment by imposing objectives that minimize the distance of new development sites from existing developed areas, and minimize open-space development objectives. They also imposed density-based design constraints that require a neighborhood for a defined minimum threshold of a certain use type to obtain additional allocations for that use. Ref. [80] explicitly addressed a residential location choice conceptual model. The residential choice model is among the core concepts that explain the spatial separation of residents from workplaces and other functions, based on utility theory. Spatial separation degrades peoples’ quality of life (i.e., the accessibility of work, medical care, shopping, and leisure facilities). Within their land use model, the workers live in old districts for facilities and commute to the new development areas for employment. As a response, the authors integrated workers’ residential areas in land use models, with which they aimed to maximize the quality of life of the workers and the productivity of facilities in the new development area. Their findings demonstrated that the land in a new development area can be utilized more effectively if the land use allocation is decided upon based on a housing location model. Ref. [13] draws on another important theory, the access to transportation model, into urban land use planning, along with other common objectives (compactness, compatibility, and suitability). The transport accessibility map of their case study was a composite of driving accessibility, cycling accessibility, and walking accessibility maps. The accessibility objective function can measure the degree of matching between the land use type and transport characteristics. Compared to a control problem without an accessibility objective, the problem with an accessibility objective implemented in NSGA-II returned a spatial layout that was more compatible with the existing development potential.

In a nutshell, the land use planning context can be traced in terms of basic theories/conceptual models that govern land use allocation and dynamic elements in modeling reality. The articles studied in this paper and the general literature are proof that, in relative terms, the latter is gaining more depth and breadth. A shift has occurred from a temporal to intertemporal perspective, and from deterministic assumptions to uncertainty accommodation in many forms (including input parameters and output scope, and the participation and interaction of stakeholders) to capturing the degree of uncertainty. However, progress on the fundamental conceptual side of land use allocation theories (and spatial morphology in urban land use) is very limited. Instead, the study of land use planning has drifted away from their spatial domain. One possible explanation for the weak developments in the conceptual realm of planning knowledge could be the fact that a significant portion of the optimization community is populated by professionals and practitioners from computer/software engineering and related disciplines, whose primary interest could be the efficiency of algorithm implementation. If this conjecture holds, it indicates that the degree of cooperation between land use planning experts and those working on the algorithm/model development side is limited, defined by code development and code use, with only a limited possibility of joint conceptualization of the spatial development problem. Another conceptual construct exists. Much of the recent literature advocates for a precise distinct zoning of uses. The purpose of suitability zoning is to demonstrate precise separation of functions from each other based on the minimum cumulative resistance (MCR) model. The implication is, however, depriving some locations of certain ESs, and other locations of other ESs. This calls for mechanisms that maintain a desired degree of heterogeneity and offer different ESs with a fairly accessibility resolution.

Frontiers

Many of the conceptual models related to land use allocation problems in the built environment have remained immature, or their scope of application is narrow; they may be completely unaddressed within optimization-based land use planning research. Furthermore, the specifications of certain objectives contradict basic land allocation and urban spatial morphological concepts.

In one way or in another, transportation is an integral part of land use planning. In traditional structured mathematical models, the accessibility of a form of travel is modeled by an origin–destination time cost dataset [77]. In cities with advanced economies, the availability of a large amounts of origin–destination census data detailed at the census or at postal-address-block resolution makes time cost modeling possible. Where such data are not available, usually in developing cities, often, survey data are collected at a defined spatial unit, say a district, and utilized in the modeling. In some studies, accessibility is modeled by a linear/rectilinear geographic distance. In rare instances, such as in the work by ref. [13], the use of a composite geographic accessibility index map may be sound, for the reason that the accessibility map considers different means of access (driving, cycling, and walking). The origin–destination model requires the real commuting time/distance and impedance, whereas the input in the case of [13] is the mobility service of given transport systems aggregated into a composite index. Furthermore, in case neither the actual origin–destination time/distance is readily available nor the mobility infrastructure well developed, the use of spatial proximity data for functional uses may be an alternative, in which accessibility maps can be created for each travel purpose, and the map index value can be utilized as input. The accessibility computed for each purpose of travel aggregated into a composite index map may serve as an alternative mechanism. Furthermore, neighborhood-based accessibility is an advantage given the problems of inconsistent traffic zone sizes. Finally, neighborhood-based accessibility is in congruence with the mixed-use development version of the sustainable built environment [16,86].

While transport infrastructure remains a key entity for spatial organization, the spatial structure may be predicted well if comprehensive model factors are considered. The logistic regression method and the minimum cumulative resistance (MCR) model that are common in rural/regional land use planning research may be customized for application to the built environment. Logistic regression can be utilized to construct a probability map of land use types based on change-driving factors. The degree of resistance (MCR) from a defined source of planning elements such as a high-density residential land, a protected special function, or industrial establishments, etc. can be applied to further fine-tune the layout of uses to meet a given compatibility objective.

Travel behavior is another key element that needs addressing in the optimal allocation of land. Within a given spatial structure, the outcome of travel behavior reflects feedback to a modal assignment. Currently, travel behavior is detached, and not only from optimization-based land use planning research; the perspective of travel behavior in research on land use types is also lacking. In order to model the built environment in a better way, the work of ref. [80] can be furthered to include broader aspects of human behavioral models within the built environment, so that travel behavior modeling may contribute more valuable perspectives to optimal land use planning research. More concretely, the work by [80] can be extended to integrate different land use models into a comprehensive and unified model that minimizes the negative consequences of spatial separation to a meaningful degree. Otherwise, the spatial layout of one model may contradict the results of another model, and ultimately the built environment will remain in a severely sub-optimal state.

The scientific potential of land use planning that does not appropriately consider fundamental land use allocation conceptual models, for example, the gravitational forces in the economic geography model rather than simply, the per capita GDP, is less credible. Distribution of economic benefits, for example, land value, workplaces, and investment, etc., and therefore associated social benefits (access to employment and investment in local, physical, and social infrastructure) can be captured well if concepts such as bid–rent theory and economic geography are considered in decisions regarding the spatial layout of land use activities. The macro-morphological structure of a city is crucial to modal assignment policies [3,81], and is another key concept that needs to be reiterated in land use studies. To this end, the results of a recent case study (Mek’ele city, Ethiopia) demonstrate a significant variation in accessibility, attributed to the area’s morphological structure, evaluated in a centralized–multicenter–decentralized continuum spatial structure scenario [87].

To a relative degree, regional land use planning accommodates more frequent land use changes than its urban counterpart. This is because regional/rural land is the major source of demand for land uses and the ESs can be exchanged among the different uses relatively easily. Consequently, optimizing land use allocations for the attainment of better ecosystem products is normative. The relatively fixed supply of urban land and long-lasting structures (buildings, physical infrastructures, and even societal fabrics) make land use change in the built environment relatively static. This is why much of the existing land use planning research is a simulation of expansion areas. However, it can be argued that the simulation of future expansion does not respond to the overall land use, and by extension, to the overall functionality of the city. The simulation is indispensable for handling the expansion part. A complement to the already developed part of the city, which usually covers a major share of its total spatial extent, is also required. The possible alternatives seem to be optimizing the distribution of flow resources, for example, by restructuring population density patterns and transport modal assignments to the existing major morphological and functional structures and capabilities, which is a reversal of the conventional method of land use optimization that restructures the spatial configuration of activities to respond to flow resources. From a policy perspective, it is crucial to underscore that flow resources’ allocation to existing land use could be a more reasonable planning approach than changing land use to respond to flow resources.

Research that evaluates the performance effectiveness/efficiency of the built environment is rare. Spatial performance evaluation studies of cities may allow policymakers to revisit their past trends and derive important lessons about the effects of certain land use planning approaches. Furthermore, they may either allow them to devise remedies to the effects of possible flaws of spatial configurations within already developed areas via redevelopment programs or through considering compensatory programs in expansion areas. More importantly, how planning policies affect the land use structure has not been fully addressed by analyzing alternative scenarios. Scenario analysis simply allows for comparison of relativity. Demonstrating new methods returning land use maps that evaluate better values over the pseudo-objectives does not necessarily indicate that the city would perform at its optimal state.

We also encounter a situation wherein urban land use optimization contradicts the basic principles of a sustainable built environment. As a concept, compactness has been mainstreamed in two forms in sustainable city discourse. In the original conceptual realm, it is based on minimizing sprawl and encouraging infill development for the efficient use of land resources [4,16,17]. In this mainstreaming, attainments increase accessibility to services, enhancing social mixing and energy use efficiency, and reducing pollution rate, etc. Within this original mainstreaming of the concept, many studies [87,88,89,90,91,92,93] have demonstrated the positive contribution of the decentralization of services and workplaces to the local level. The basic concept of the compact city can be either of high density, a mix of uses, or high intensity of use [16,86,94]. The normative association of compactness and sustainability is mainstreamed by the size of the entire city as a function of the productivity of land [95,96]. The mathematical representation of the compactness objective function in land use optimization, however, overlaps with the second mainstreaming, which claims consolidation of the same use to certain clusters [26,27,28]. The ultimate effect of this mainstreaming is aggravation of the spatial differentiation. Spatial convergence minimizes emission from transport, promotes walking and cycling or use the of public transport, shortens travel distance, converges travel time in cities with high traffic congestion, conserves ecological land, ensures equitable access to social infrastructure and services, etc. In principle, the objective of maximizing larger clusters/minimizing the number of clusters contradicts the basic premise of compact development. In theory, clustering of the same uses should promote spatial separation among the purposes of travel. It is also worth underscoring that no empirical study has demonstrated the relative advantages of clustering over mixed-use methods in terms of resource use efficiency and social equity. Sometimes, compactness/contiguity, number of clusters, and cluster size parameters are employed in land use optimization studies with the aim of restricting conversion of certain uses or preserving certain locations. In such cases, molding them as constraints rather than objectives may sound more rational. Otherwise, imposition of the parameters, as has prevailed, is a way of setting certain spatial configurations outside the algorithmic environment. In the work by [14], for instance, the additive parameter (land use suitability) declined for agriculture (by 2.5%) because of the preferential treatment of compactness by the applied operators. In that study, the sensitivity analysis challenged the modeling of shape-oriented objectives in combination with additive objectives. Compactness showed higher sensitivity to the distribution of the weights than its suitability counterpart did. In general, if clustering is admitted for a specific policy rationale, identification of the organizing planning elements must be clear, rather than simply minimizing the number of clusters or maximizing the size of clusters for certain uses.

Another dimension of the conceptual level discussion within the land use optimization can be whether certain shapes (contiguity and compactness) should be the outcomes of satisfying the defined aspatial objectives, or whether spatial objectives have to be determined within the preferentially imposed shape.

Detailing objectives to the level that practicable measurement can be achieved with established units of analysis (such as the monetary value of use conversion cost, the value of industrial products or services, the fair locational distribution of land value, and quantifiable social development indicators and social fabric integration in urban design, etc.) is another frontier in urban land use planning research. In addition, the relative degrees of influence of the factors of land use change have not been brought for discussion and inquiry. Finally, there are no standard values that define a certain land use in relation to objective parameters that can be addressed by transferring knowledge from the ES value, taking into account rural/regional land use studies with appropriate customizations.

5.2.2. Optimization Methods

Current Status

As indicated in Section 5.1, the current state-of-the-art approach for land use optimization involves the utilization of hybrid algorithms. Hybrid forms could be the coupling of two global optimization tools, or two local simulation tools, or a combination of a global structure planning approach and one or more local simulation tools. According to Table 2, 15 (78.95%) of the 19 regular study articles applied two or three different methods to solve the optimal land use allocation question. Two factors have been supposed to drive the adoption of multiple methods. Firstly, there are scalability limitations in terms of the efficiency and effectiveness of the global optimizers. By incorporating multiple methods, these limitations can be overcome. Secondly, the disaggregated spatial allocation of activities from planning activity necessitates the independent treatment of land use planning demand for different activities, and the spatial mapping of the planned activities. This is further justified by the specific requirements of policymakers at the local spatial scale.

The scalability limitation of metaheuristic global optimizers, particularly in the context of land use optimization, is a well-known challenge. With an emphasis placed on the application of GA (because it is the dominant method of implementation), it is needless to cite sources confirming the common approach to deal with the scalability challenge is to design purpose-fit operators. The studies by ref. [25,28] are examples wherein modified genetic operators act on neighborhood windows and their boundary information demonstrated the effectiveness of maintaining diversity in the population, and another (constraint-handling) mutation effectively filtered out infeasible solutions within the evolution process.

Outside the metaheuristic algorithm’s domain, the authors of ref. [59] applied simplex-LP. Despite criticisms of LP regarding its requirement of secularization of objectives [61], the simplex-LP method solved the land allocation problem with success in a situation wherein water and energy dynamics, environmental sensitivity, and economic goals were at stake. The authors formulated the problem as a multiple-objective implementation of simplex-LP in ADBASE, without the need for aggregating the objectives. The performance of the model was demonstrated by the 18.62% growth in economic benefits and the 7.78% reduction in soil erosion relative to the base-case scenario. The associated magnitude of land use change (a 93% increase in orchard land and a 50% decline in dry farmland) highlights both the effectiveness of the simplex-LP method in solving the problem and the limitations of traditional land use planning methods that do not consider balanced trade-offs among attainment objective parameters.

The authors of ref. [67] acted outside the algorithmic environment. They introduced a partially known solution strategy to deal with scalability associated with multidimensional objectives. Initially, their problem had three objectives (minimum fertilizer use, minimum nutrient outflow, and maximum agricultural output). To improve the GA implementation performance, they reduced the objective space to two by first optimizing the minimum fertilizer use objective. Then, they used the optimal fertilizer use map as an input for optimizing the trade-off between nutrient outflow and agricultural yield objectives. The fertilizer use map underwent a performance check using the Monte Carlo method before its use. The authors reported better convergence for the simulation that utilized the optimal fertilizer use map compared to a control problem that was evaluated from a scratch.

Despite multiple options are available to deal with scalability problems, the use of hybrid methods remains a priority, and there are various forms and techniques of hybridization. Ref. [70] applied a semi-parallel hybrid of PSO and GA with the addition of a local search. In this model, the entire initialized population was evaluated using PSO. Individuals with lower fitness were then returned and evaluated using the GA. Finally, the solutions returned from both methods were pooled together for further local search operations. In ref. [83], a servitude companion hybrid type of SA-GA was applied to examine the impact of LULC change on the distribution of ESs values. To improve the solution quality, the crossover and mutation functionalities of GA were incorporated into each temperature cycle of the SA. After random solutions are initialized in SA, and fitness is evaluated, solutions selected for the next generational evolution undergo crossover and mutation. The cycle terminates when the SA termination criterion is met. Ref. [31] also applied the same hybrid method. The advantage of the servitude companion type of hybrid method is not only the exploitation of the respective strengths of the involved methods but also its potential of avoiding the need for handling multiple methods and possible challenges of dataset integrity loss associated with the input-to-output management that otherwise would occur using simple couplings. Additionally, servitude companions facilitate the use of the hybrid method by end-user researchers who may lack coding skills/knowledge.

In the study described in ref. [45], the authors applied ACO, Markov chain, and CA for modeling spatiotemporal land use change. Markov and CA were used to manage the land use conversion, while the spatial distribution of land use was controlled through the integration of ACO and CA capabilities. Local transition rules for land use change were developed based on concepts from ACO and CA. The simulation effectively predicted the area of the two larger land use categories, but errors occurred for land uses with smaller areas. While the authors make claims about the effectiveness of the model, and the spatial matching (total quality) was 73.99%; the spatial accuracy for construction land was as low as 58.49%, and the maximum area prediction error was 18.2% for the construction land. However, this is not necessarily a drawback of the coupling method, but may be attributed to specific limitations in formulating the problem itself.

In order to address the land use planning problem comprehensively in their study case, ref. [67] employed multi-objective dynamic planning (MODyP), CLUE-S, and MCR in a sequential cooperation coupling format. In this approach, the first step is to determine the area requirement for each land use based on the maximization of ESs values, wherein MODyP was utilized. Then, a logistic regression equation was applied to determine the factors that influence land use distribution (which in their specific case were land use demands, regression results, elasticities of transformation, and the transfer matrix). These factors and their relative degrees of influence serve as inputs for simulating future land use changes and optimizing the spatial layout of the land use activities using the CLUE-S model. Finally, the minimal cumulative resistance model is applied to adjust the land structure for suitability zoning.

In a 2023 study by ref. [68], the hybridization method was further developed to address the context of planning knowledge. The authors criticized existing studies for their methodological limitation of integrating ESs into land use planning from three aspects:

(i)

They questioned whether spatial evaluation and stakeholders’ surveys generate appropriate alternative schemes that shape the effectiveness of planning decisions;

(ii)

They highlighted the integration of ESs into land use without optimization technology (See [15]);

(iii)

They pointed out the efficiency limitations of manually generated planning schemes, emphasizing that representing government policy restrictions in a model is often challenging.

Additionally, they mentioned the difficulty in quantifying ESs and land use configuration, as the relationship between certain ESs and spatial layout is often non-linear.

To tackle these challenges, they proposed a two-step spatially explicit optimization approach that they believed would better incorporate ES into the land use planning problem. In the first step, they optimized the land use structure (quantity) by applying MOLP to maximize ESs. In the second step, they determined the spatial layout of uses based on suitability objectives that considered biophysical and geographical factors. The layout determination task involved horizontal and vertical comparisons of suitability maps and stepwise allocation. After the land use structure was determined, land use allocation was decided by prioritizing spatial units with the highest suitability for each use type. If multiple units with the highest suitability were available, the average value of their respective neighborhood determines the use. The suitability of the land units is then assessed for alternative uses, and the information is fed into the implementation algorithm at each iteration cycle until all cells have been allocated.

In a related study by ref. [9], LP, MCR, and dynamic-CLUE (DyCLUE) were applied to integrate land use ecological suitability into land use planning at an urban agglomeration scale. The MOLP combined the objectives into a single objective for quantity determination. Logistic regression was utilized to compute the relative influence of land use change factors (traffic and location factors, natural environmental factors, and social and economic factors), the output of which was the input for the MCR that land units should overcome during horizontal movement, considering ecological and construction sources. The Dy-CLUE model explored the optimal spatial layout of land use activities, considering the outputs of MOLP and MCR.

CLUMondo is another variant, but relatively rarely utilized, land use simulation method based on identification of land use change factors and analysis of their relative degree of influence. The model estimates the priority of a use type for each grid cell by employing the parameters fed to it. The land use type with the highest priority becomes the new allocation. The application by ref. [65] is an example. They applied the method under different land use policy scenarios (historical trends, government planning, and windbreak and sand fixation).

GT has been widely used as a classic optimization tool in various disciplines, including land use planning. Its application to the land use planning coupled with GA was first introduced by [14] in 2015. In that study, GT played a role in simulating coordination between local-scale competitions after the overall land use structure was determined. The process began with the execution of GA for each land use type (such as farm, forest, garden, development) to generate an efficient land use map. The efficient map was then compared to the existing land use map to identify competition zones. These competition zones represented areas where, during development, there was a competition between discrete leapfrog construction and agricultural or ecological lands. The role of the dynamic-GT was simulating the negotiation process between the government and agricultural/ecological land use interest holders in the competition zones dealing with compensation pay rates. The government iteratively improves the compensation offer geometrically, and the farmers hold the option either to accept or to reject the offer based on their assessed or perceived income within a certain period. If the deal fails, the government will continue to improve the pay at a decreasing rate. This process continues until a certain cycle (in their case, an order of n = 10), during which either the farmers accept the rate or the government ceases further improvement—a situation in which the negotiation terminates. Upon successful termination, the government will acquire the entire competition zone. Another study by ref. [97] in 2019 also applied a similar approach of combining GA and GT to solve an urban land use allocation problem. In cases wherein the competition was among different agricultural uses [14], planning knowledge can be utilized to determine the use of the competition zone based on the sum of area of each use, selecting the use with the largest area in the competition zone.

Another common hybridization strategy to enhance the scalability of metaheuristic algorithms is the integration of a multi-agent system and heuristics such as GA. This approach has been applied in various studies [98,99].

In summary, land use planning and spatial simulation tools have achieved significant advancements in conceptualizing the technical rigor and use of multiple (usually two to three) methods to solve a single spatial development problem. Advancements have been driven by scalability problems with the implementation tools themselves in handling complex spatial problems, and by increasing the depth of intervention in the spatial planning context. Both situations have contributed to the emergence and advancement of hybrid methods. Within such great achievements, two important limitations can be identified, one being the use of gray and stochastic values to molding uncertainty. In theory, these assumptions justify the importance of a fairly wider search space for solutions. However, in practice, they are less likely to realize a fair trade-off balance among objectives. Objective function values corresponding to either the lower or the upper boundary of the input range is not unlikely. Instead, the authors of this paper demonstrated a fair trade-off among objectives by converting the input range to a scalar objective function form, wherein the objective function immunizes the deviation between the input boundaries and the allocation values. Since both boundary values are considered in the evaluation loop, the allocation tends lay around the mid-point of the input range, meeting the ultimate purpose of minimizing the magnitude of the trade-off between the objective parameters. We also argue that molding uncertainty in the objective space (See [64]) adds less, if any, value to the planning, because output is a function of input.

Given the methodology contemporarily dominating the state of the art is hybridization, the second limitation is the lack of a clear characterization of the available hybrid modalities. In order to facilitate understanding of hybridization approaches and to shape the environment for further study, this paper tries to fill this gap. Based on the degree of integration between/among the methods in their hybrid use, we label them in four ways:

(i)

Servitude companion (SC)—certain functionalities of one global optimizer are mapped into another global optimizer as integral parts.

(ii)

Semi-parallel cooperation (SPC)—solutions that are unable to pass one global optimizer are evaluated by another global optimizer, and solutions of both streams are pooled together (and may undergo further local search operation).

(iii)

Sequential coupling (SqC)—the output of one (usually the global optimizer) is used as input to the other(s) (usually local optimizers). This is also observed between two local layout simulators.

(iv)

Bonded integration of transition rules (BITR)—land use transition rules are drawn based on principles drawn from multiple methods. Any one of the methods that contribute conceptual principles for devising the rule can serve as a hosting agent, while simulating land use transition.

It is also worth mentioning that other options for dealing with scalability problems in metaheuristic methods, such as linking practical ground knowledge to algorithms (such as knowledge-informed GA), are actively in application. Multi-objective simplex-LP and partially known solutions, though rarely applied, are also available strategies that may require readmission for further inquiries.

Frontiers

Within the limit of the sample articles explored in this review, several frontiers can be explored to further enhance current state-of-the-art hybrid methods.

One key frontier is advancing the current relatively weak SPC and SqC to create more robust and seamlessly integrated algorithms. Designs that integrate the beneficial parts of one method into another method are important to avoid the problems of input-output flow management between/among methods in coupled use. In other words, the integration of multiple methods in hybrid methods may be stronger where two or more methods act simultaneously, rather than one method waiting for outputs of the other. The way ref. [70] applied the PSO-GA coupling is equivalent to testing PSO and GA independently and then selecting the better-performing one. The SA-GA servitude companion in ref. [83] and the local transition rule ref. [45] developed from ACO and CA are examples of integrated methods. Attention is required in proportion to the value of these advanced (though infrequent both in diversity and in application rate) methods.

Another area that remains unexplored is the impact of algorithm parameter settings on the application of heuristics for land use optimization. The norm in practice is that many studies define their own global parameter sets based on general information from previous studies, rather than conducting their own experiments to determine their own set of parameters. With numerous combinations of parameters and implementation algorithms/tools, the assessment of the effectiveness of an algorithm/tool often remains subjective, lacking an objective evaluation benchmark. To address this, a standardized spatial test problem could be developed, covering a comprehensive range of parameters wherein a standardized mathematical expression is devised for each objective parameter. Standard test problems are common in many applications such as the travel salesman [99,100,101], job shop scheduling [47], investment represented by the Knapsack problem [102], and in almost every engineering problem [103].

In addition, studies that examine performance across the various hybrid methods in terms of solution quality, computational cost, fitness, etc. are another potential research area that could unleash the full potential of hybrid methods. The key motivation behind this area is discovering whether hybrid methods inherit the performance features of their parent methods, and if so, to what degree. Inquiries in this area are important guides for end-users to determine well-performing hybrids. These investigations may be more sound in cases in which conventional coupling methods are assumed for use.

In land use optimization, the fitness attainment level demonstrates the magnitude of the improvement in the modeled objectives of the resultant spatial structure in terms of mechanical indicators. Usually, spatial quality evaluations are lacking. There are situations wherein neither the knowledge-informed/modified global optimizing algorithms nor resistance to use change objectives/constraints are responsive to spatial quality. The highest fitness may be achieved at the cost of spatial quality. This situation calls for the incorporation of objectives or operators that promote the quality of spatial configuration. In urban land use optimization studies, the only available mechanism so far is the use of a density-based design [27] that preconditions allocation of a certain use by the number of existing and newly proposed cells of that use type. Nonetheless, the method leaves us with the task of determining the clustering location of the algorithm’s stochastic decision. In other words, the method is sensitive to the number of target use cells in a neighborhood, without defining geographic location; it is often ineffective.

Domain harmony is another issue that needs intervention in urban land use optimization. While it is clear that additive parameters and shape-oriented parameters (compactness, contiguity, number of clusters, and size of clusters) are different in their mathematical structure, evaluating both objectives together is the norm without considering the issue of harmonization for mathematical representations. Whether the incompatible mathematical structures affect the overall state of the land use map is unknown. If so, whether such a mixed-domain objectives problem can be modeled as two sub models (where the output of either the sub model can be utilized as an input to the other sub model) needs attention.

Finally, we note the spatial scale. In the contemporary optimization-based land use planning literature, distinctive geographic units are the information carriers of a study area (region, watershed, city region, city, district, etc.). Determining appropriate regular size cells is the key to balancing the trade-off scale among objectives. Unstandardized geographic units, on the contrary, may affect the quality of objective trade-offs, regardless of the constraints that might have been met, as in ref. [59]. The information carrier units of [59] were sub-watersheds of same use, but were not of uniform size. This resulted in unbalanced trade-offs among objectives; irrigation land and rangeland attained only lower bounds; orchard land increased by 93% above its lower bound; and dry farming increased 50% above its existing state. The outcome of such an allocation imbalance was a high sensitivity of ESs to change in even a single land unit. Reportedly, reduction in the benefit showed the highest sensitivity to the reduction in orchard and irrigated farming areas, whereas benefit increment was only sensitive to an increase in the orchard area. Similarly, reduction in the rangeland area showed high sensitivity to increased erosion.

6. Conclusions

This paper reviews advancements in the context of land use planning and optimization methodologies. Unlike most other studies that pick very specific issues, such as one or a bunch of similar methods or application of a single concept, this paper covers a broader and a more generic approach to assess the conceptual models/theories shaping contemporary optimization-based land use planning research and the state of methodological environment.

Before diving into the exploration of developments in the land use planning concept and the advancements in methods, this paper examined the feasibility of utilizing bibliometric method as a standalone method of exploring real knowledge developments. The results of the analysis indicate that statistical summary-based citation information is less likely to explore real developments in knowledge. More than 85% of the time, citation is for the purpose of general background information rather than for the core content of the articles being cited. It has been observed that review articles give due consideration to the core contribution of sources instead.

Through the traditional text reading method, conducted on selected 25 articles, a clear development track has been recorded in rural/regional land use planning within the framework of sustainable development. In rural land use optimization, overall planning has been approached as two sub-models. The planning for land demand often involves the optimization of ES values using metaheuristic algorithms or other prediction methods. The quantitative land use structure is then optimized for layout configuration by local optimizers, which apply defined transition rules for land use changes and further adjustment for suitability zoning. The suitability zone analysis involves both vertical (between different uses/departments) and horizontal interactions. This approach can be adapted to urban land use planning to solve local-level land use conflicts. Furthermore, the concept of mutual feedback between ecological change and human socio-economic activities is an emerging conceptual framework potentially initiating spatial allocation optimization in response to changing ES dynamics. Rural land use planning also takes into account uncertainties, including temporal, input, output, and process dimensions that can be adopted in the urban sector.

In urban land use planning, a shift away has occurred away from the basic spatial development theories and drivers of urban morphological structures. These classic theories, such as the bid–rent theory and economic geography, as well as utility models such as accessibility, residential quality preference modes, workplace assignment/labor allocation theories, etc. rarely appear in land use optimization. On the other hand, certain elements in land use optimization, such as compactness and contiguity objectives and density-based constraints, appear to contradict other aspects of features of the sustainable built environment such as mixed-use and overall city size compactness. More importantly, such mathematical representations are in sharp contrast to the basic understanding of sustainable cities that assumes overall city size compactness rather than functional differentiation that ultimately leads to spatial separation. These gaps and contradictions have to be questioned. Another limitation in urban land use planning is the identification of factors that drive land use change and corresponding degrees of influence. In this regard, lessons can be drawn from regional/rural land use planning.

Considering that a significant portion of urban land is already developed, a shift in perspective is also required. Optimization of flow resource and activities within the existing spatial structure may be a more applicable and scientifically supported policy direction. Assessing the performance of the existing built environment is also crucial in drawing lessons for planning expansion or an entirely new a city.

In terms of methodological advancements, land use planning has seen progress in terms of diversity and rigor. The current state-of-the-art implementation method involves coupling two or more algorithms, combining global optimizers and local simulators to varying degrees of integration. It ranges from a simple sequential coupling to a semi-parallelization and a more integrated approach. Some of frontiers include advancing the integration to the degree that beneficial parts of multiple algorithms combine to form new aimlessly unified platforms. Diversifying the hybridization horizontally is not an exhaustive approach either. More importantly, there is a need for an extensive, detailed, and objective assessments of the performance of hybrid land use optimization methods. Current evaluation methods often rely on general information in the literature, or on characterizing the scientific nature of the algorithms independently.

The overall implication of this literature study is the possibility for a reiteration of classical land use allocation theories/conceptual models and the associated physical morphology of the built environment within optimization-based land use study. It is possible that this iteration may make optimization-based land use planning more rational from a policy perspective, and more scientifically sound. It also implies the importance of standardizing methods of land use optimization. In this regard, the PRISMA prototype can be developed into a land use study guideline protocol so that comparison across the wider literature is facilitated, with a possibility of conducting a meta-analysis, which is lacking thus far. Regarding the implementation method, this review implies that the development of an environment for the production of seamlessly unified algorithms has already occurred.

Finally, we suggest that the development of a land use planning study protocol and a land use planning test problem take priority, in order to make land use planning studies stronger in both the scientific context and in methodological rigor.