1. Introduction
Population growth and urbanization have intensified competing land use demands, placing serious constraints on the healthy and sustainable development of urban areas worldwide. The urban land use planning process is inherently complex, requiring the integration of social, economic, environmental, and political systems, all of which are central to professional planning practice. At the same time, rapid advances in artificial intelligence (AI) and big data technologies present planners with significant opportunities to critically reassess their approaches and shift toward data-driven strategies to enhance urban sustainability, accessibility, and spatial governance [
1,
2]. New forms of data collection, analysis, and modeling can augment planners’ understanding of places and help them anticipate futures with improved quality of life [
3]. AI and big data can provide access to richer, more timely information on travel patterns, energy consumption, land use, and environmental impacts, while also helping better integrate the complex systems that shape urban land use outcomes [
1].
The seven articles collected in this Special Issue of Land respond to these opportunities from a diverse set of geographic and methodological perspectives, spanning study areas in Sri Lanka, China, the United States, and Sweden and employing methods ranging from big data fusion and machine learning to spatial optimization and automated microclimate simulation. In this commentary, we briefly summarize the contributions, identify the major themes that unite them, and reflect on some areas for future research.
2. Overview of Contributions
Akalanka et al. (List of Contributions, 1) address a significant challenge for urban land use planning: accurately delineating urbanization patterns in countries where census data are infrequent, outdated, or conceptually limited. Using Sri Lanka as a testbed, they develop a big data fusion approach that integrates nighttime light imagery, points of interest (POIs), mobile network coverage, road network data, and normalized difference vegetation index (NDVI) data. Their findings challenge the conventional administrative-boundary-based understanding of Sri Lankan urbanization, revealing that a substantial share of officially designated urban areas are either over- or under-bounded. In contrast, many functionally urban places go unrecognized. The study demonstrates how open data sources can make urbanization measurement more dynamic, cost-effective, and transferable to other low- and middle-income country contexts where accurate land use information is most urgently needed.
Mortaheb et al. (List of Contributions, 2) evaluate how alternative planning frameworks shape commuting patterns, engaging the critical connection between urban land use and transportation. Their study integrates geographic information systems (GISs) with linear programming—specifically, a variant of the classic transportation problem—to assess commuting efficiency under a form-based code (FBC) regime in Orange County, Florida. By modeling journey-to-work trips across three workforce cohorts segmented by industry sector, they demonstrate that form-based codes hold the potential to improve commuting efficiency at the local level through strategies such as mixed-use development, transit-oriented densification, and juxtaposing residential and employment activities. However, their results also reveal that the FBC framework falls short of achieving an ideal job–housing balance at the regional scale, with approximately 76% of internal work trips remaining excessive even under the FBC scenario. These findings underscore the persistent difficulty of aligning land use and transportation planning across metropolitan areas.
Lu et al. (List of Contributions, 3) focus on the quality of children’s active school travel spaces (ASTS) in Lanzhou, China, linking urban land use analysis to transportation and the built environment. Using the Amap pedestrian route planning API, spatial syntax, and street view image recognition via deep learning, they evaluate walking routes and school gate areas across 151 public primary school districts on dimensions of access, safety, and comfort. Their geographic detector analysis reveals that population density and transportation convenience significantly increase child friendliness, whereas school district size and school centrality significantly decrease it. The study makes a notable methodological contribution by combining big data techniques, including geographic big data mining through API-based route construction and computer vision, with children’s subjective perceptions, as captured by survey-adjusted indicator weights, thereby bridging quantitative and qualitative assessment traditions in a way that is directly relevant to land use decision-making around schools and residential neighborhoods.
Zhai et al. (List of Contributions, 4) examine the spatiotemporal dynamics of high-quality development (HQD) and development coordination in cities along the Lower Yellow River from 2000 to 2020. Using a multi-source remote sensing framework built on the Google Earth Engine platform, they construct a three-dimensional assessment that incorporates agricultural, ecological, and urban functions, drawing on the production–living–ecology (PLE) analytical framework. Their analysis tracks how the functional composition of land has shifted across 149 counties and districts over two decades, finding that urban functions increased steadily while ecological functions declined before recovering, and that agro-ecological land types dominate under most administrative scales. By quantifying land use functions at fine spatial scales using remote sensing rather than statistical yearbook data, the study provides a spatially explicit approach to monitoring the human–land conflicts central to understanding sustainable urban development in resource-constrained regions.
Li et al. (List of Contributions, 5) present an AI approach to urban land use classification that integrates commercial area-of-interest (AOI), POI, nighttime light, and population distribution data. Using the categorical AOI data as supervised labels for deep learning models and comparing XGBoost, SVM, random forest, and multi-layer perceptron approaches, they refine land use classification into nine major categories and 21 subcategories across cities of varying scales and locations. The XGBoost model achieved the highest accuracy with a weighted average F1 score of 0.87. Combined with remote sensing imagery and transportation network data, the approach generates detailed land use maps that substantially increase the granularity achievable with conventional methods by addressing a core challenge that traditional remote sensing struggles to distinguish among functionally distinct yet visually similar built-up land categories.
Gao et al. (List of Contributions, 6) investigate the coupling between street hierarchy and urban functional distributions in Shenzhen through the lens of geographic big data mining and complexity science. Constructing natural streets from OpenStreetMap data and applying head/tail breaks to reveal hierarchical scaling structures, they demonstrate that POI distributions, which represent aggregated commercial, public service, and other urban land use functions, are systematically shaped by the multilevel configuration of the street network rather than being randomly dispersed. Power law analysis reveals that both street networks and hotspot clusters follow heavy-tailed distributions, forming nested living structures. The finding that form–function correlations are strongest at mid-to-upper hierarchical levels, weakening and becoming increasingly nonlinear toward finer levels, provides theoretical insight into how urban morphology and land use activity patterns co-evolve, with practical implications for understanding how street network investments shape functional land use outcomes.
Spett et al. (List of Contributions, 7) address a practical bottleneck at the intersection of urban land analysis and environmental simulation: the labor-intensive process of constructing site models for microclimate simulation software such as ENVI-Met. They present an automated method for generating urban models from readily available cadastral data, RGB + infrared orthophotos, and digital elevation models. Their application, released under the open source MIT License, uses algorithms to analyze grass, trees, buildings, and roads and extract and configure spatial models compatible with microclimate simulation. The work demonstrates how remote sensing data on the physical characteristics of urban land can be efficiently translated into environmental simulation models, supporting the kind of integrated analysis needed to understand how land use decisions affect microclimate conditions and human well-being in urban areas.
3. Major Themes
The most prominent theme across the volume is multi-source and dynamic data integration. Nearly every article integrates multiple heterogeneous data sources, such as remote sensing imagery, POI data, nighttime lights, population data, street networks, and other open or commercial datasets, to construct richer representations of urban land use. Akalanka et al. fuse five open data sources; Li and Zhu combine AOI, POI, nighttime light, and population data; and Zhai et al. draw on land cover, NDVI, population, POI, nighttime light, and building height datasets. This reflects a maturing recognition that no single data source can capture the multidimensional and dynamic character of urban land use systems, and that the resulting insights exceed the sum of individual inputs.
A second theme is the expanding role of AI and machine learning. Li and Zhu’s XGBoost-based land use classification, Lu et al.’s deep learning and computer vision for street view segmentation, and Akalanka et al.’s adaptive thresholding algorithms all illustrate how machine learning is being deployed to automate and scale tasks that were previously manual or imprecise. Gao et al.’s computational complexity science approaches similarly leverage algorithmic analysis to identify structures in urban form that would be invisible to traditional methods. These applications suggest that AI is moving toward practical tools for land use planning, though questions of model interpretability and training data bias remain important.
Third, the land use–transportation connection remains a significant research focus. Mortaheb et al. directly model the commuting efficiency implications of alternative land use frameworks; Lu et al. connect land use configuration around schools to active-transportation outcomes; and Gao et al. demonstrate that street network hierarchy shapes functional land use distributions. These contributions reinforce the long-standing planning insight that land use and transportation are inextricable, while showing how big data and spatial analytics highlight this relationship with new precision.
Finally, the collection of articles in this Special Issue is notable for its emphasis on open data and global transferability. Akalanka et al. exclusively use open data with Python; Zhai et al. build on Google Earth Engine; Gao et al. rely on OpenStreetMap; and Spett et al. release their tool under an MIT license. This orientation toward accessibility is particularly significant for the Global South, where proprietary tools can be prohibitive barriers. It supports the reproducibility essential for building a cumulative evidence base across diverse urban contexts.
4. Research Opportunities and Directions
While the contributions represent important advances, this collection also reveals areas warranting further attention. Most notably, the articles are more oriented toward characterizing current urban conditions than toward detecting, modeling, or projecting land use change over time. While Zhai et al. track functional changes over twenty years, studies employing time-series change-detection algorithms or scenario-based simulations to explore how urban land use might evolve under different policy or climate conditions are absent. Given the accelerating pace of urban transformation, predictive and simulation-based approaches represent a critical frontier.
Another research area representing opportunities for future research related to big data and urban land use planning is social equity and environmental justice. While Lu et al.’s child-friendly focus and Mortaheb et al.’s workforce segmentation touch on distributional concerns, questions of whose neighborhoods are accurately represented in open data, how algorithmic classifications might encode existing biases, and how analytics can support environmental justice outcomes remain largely unaddressed. Relatedly, although Spett et al.’s microclimate work and Zhai et al.’s ecological assessment provide partial connections, direct links between urban land use patterns and energy consumption, carbon emissions, or climate adaptation are not explored, which is an important gap given the central role that land use decisions play in determining environmental outcomes.
We also expect that real-time monitoring and predictive capabilities can distinguish contemporary urban informatics from traditional analysis, and will be explored further in future research. Similarly, the rapid emergence of large language models, generative AI, and agentic AI has created new possibilities for planning applications, from automated plan review to natural language spatial data queries, that merit sustained scholarly attention. Finally, the articles primarily focus on analytical methods, with relatively limited discussion of how big data and AI outputs might be integrated into actual planning, governance, and decision-making processes, which is an essential step for realizing the practical promise of the field.
5. Conclusions
The articles in this Special Issue collectively explore big data and AI applications in urban land use planning, which have matured into a methodologically diverse and geographically expansive field. From fusing big data to redefining urbanization measurement in Sri Lanka and automating microclimate model generation in Sweden, the contributions illustrate both the breadth of land use problems that can be addressed and the range of computational methods available. The recurring emphasis on multi-source data integration, open source tools, and AI-assisted analysis reflects a field increasingly capable of producing knowledge that is both analytically rigorous and practically relevant.
As might be expected, plenty of new research frontiers will emerge, and as these types of analytical tools become more powerful, the questions of purpose, inclusion, and implementation become more consequential. The most impactful future work will likely be that which not only advances methodological sophistication but also engages with the normative and institutional dimensions of urban land use planning while asking not just what we can measure and classify but also how these capabilities can be deployed to address the human–land conflicts that define the contemporary urban condition.
We are grateful to the authors who contributed their scholarship to this volume and to the reviewers whose efforts strengthened each article. We hope this collection serves as both a useful resource for researchers and practitioners working at the intersection of big data, AI, and urban land use planning and as a catalyst for the next wave of inquiry into how emerging technologies can help build more sustainable and equitable urban futures.