Leveraging urban AI for high-resolution urban heat mapping: Towards climate resilient cities

Abstract

Urban heat island (UHI) effects are increasingly recognised as a significant challenge arising from urbanisation, leading to elevated temperatures within urban areas that pose risks to public health and undermine the sustainability of cities. Effective UHI management requires high-resolution and timely mapping of urban temperature patterns to guide interventions. Traditional methods for urban heat mapping often lack the spatial accuracy and efficiency necessary for detailed analysis, especially in complex urban environments. This study integrates Urban artificial intelligence (Urban AI) by presenting a U-Net model tailored for urban heat mapping within the metropolitan area of Adelaide, South Australia. Trained on high-resolution thermal and spatial data from the South Australian Government Data Directory, the model captures pixel-level temperature variations across diverse urban landscapes, including densely built areas, suburban zones, and green spaces. Achieving a low Mean Squared Error (MSE) of 0.0029 and processing each map in less than 30 seconds, the model demonstrates exceptional accuracy and computational efficiency. The U-Net model, as an Urban AI agent, offers a scalable tool for urban heat analysis, supporting real-time assessments and facilitating targeted UHI mitigation efforts. By bridging the gap between advanced geospatial modelling and practical urban planning, it enables data-driven decisions that enhance climate resilience, optimise green infrastructure, and improve public health in rapidly urbanising regions. This approach highlights the transformative potential of Urban AI in addressing urban heat challenges, delivering precise and actionable insights to support sustainable and climate-adaptive urban environments.

Introduction

The rapid pace of urbanisation and changing lifestyles have intensified urban heat challenges, posing serious risks to public health and well-being (Baum et al., 2007; Metaxiotis et al., 2010) and threatening the achievement of sustainable urban development (Yigitcanlar, 2010; Kamruzzaman et al., 2018). These risks are further exacerbated by climate change, necessitating a reassessment of urban infrastructure to mitigate adverse effects (Yin et al., 2023). Microclimates, where design and material choices create significant localised climatic variations, often result in notably higher temperatures and humidity compared to rural areas (Degirmenci et al., 2021; Vinayak et al., 2022). Managing such conditions demands a nuanced understanding and practical strategies (Cheval et al., 2024). Urban heat mapping is crucial in this context, providing reliable, high-resolution data to support diverse stakeholders in addressing these challenges (Dimitrov et al., 2024).

Despite extensive research, accurately predicting urban thermal environments remains complex due to the dynamic nature of variables and the technological limitations of traditional methods (Wang et al., 2022). Generating precise, georeferenced, and high-resolution data is challenging, as traditional techniques, grounded in empirical data and environmental simulation tools, often fail to process large-scale data efficiently or adapt to rapidly changing environmental conditions (Li et al., 2022; Shi et al., 2021). These shortcomings underscore the limitations of conventional heat mapping approaches and highlight the urgent need for advanced computational tools that can manage spatial complexity, improve accuracy, and support real-time decision-making in urban settings.

Urban artificial intelligence (Urban AI) offers a transformative approach to addressing the limitations of traditional geospatial analysis by leveraging advanced machine learning (ML) and deep learning (DL) techniques. By integrating large-scale urban datasets, including satellite imagery, sensor networks, and real-time environmental data, Urban AI enhances spatial accuracy, computational efficiency, and adaptability in complex urban environments (Yigitcanlar, 2024a; 2024b). This study introduces a novel U-Net convolutional neural network (CNN) model as an Urban AI agent to enhance urban heat mapping and microclimate analysis. Using high-resolution satellite imagery, the model effectively captures fine-scale spatial temperature variations, offering a robust tool for identifying urban heat island (UHI) effects. The U-Net architecture significantly reduces computational time and complexity, processing heat maps rapidly while maintaining high spatial accuracy. This enables the development of targeted mitigation strategies, such as optimising green infrastructure and managing anthropogenic heat sources (Irfeey et al., 2023; Mathew et al., 2019).

This study introduces a novel U-Net convolutional neural network (CNN) model as an Urban AI agent for high-resolution urban heat mapping. Trained on RGB satellite imagery, the model captures fine-scale spatial temperature variations, enabling accurate detection of urban heat island (UHI) effects. The encoder–decoder architecture of U-Net enhances spatial accuracy while maintaining computational efficiency, supporting timely and localised heat mitigation strategies such as optimising green infrastructure or reducing anthropogenic heat sources.

Local governments have long aimed to integrate urban technologies into planning and decision-making (Yigitcanlar, 2006; Yigitcanlar et al., 2023a, 2023b). To support these efforts, the proposed model of near real-time heat maps can guide policy and emergency response. Its high-resolution outputs are valuable for identifying heat-prone zones at street and neighbourhood levels, informing targeted interventions like tree placement and surface treatments. The model’s efficiency and low hardware requirements also make it suitable for widespread use, including in resource-limited contexts.

Overall, the study demonstrates how Urban AI can bridge the gap between geospatial modelling and actionable planning. By offering a scalable and accessible tool for heat monitoring, the model contributes to more adaptive, resilient, and liveable urban environments in the face of intensifying climate risks.

Literature background

The rapid growth of urban populations is driving the spatial expansion of cities, both horizontally and vertically (Wang et al., 2023). This urban sprawl is widely recognised as a contributing factor to numerous environmental and social challenges, including public health concerns (Mohan and Kandya, 2015), elevated pollutant concentrations (Ge et al., 2024; Sarrat et al., 2006), and increased energy consumption (Akbari et al., 2016). A significant consequence of this growth is the replacement of natural landscapes with impervious surfaces, leading to altered urban climates (McCarthy et al., 2010; Zhang et al., 2013) and notably higher temperatures, especially in the late afternoon and evening. This UHI effect, where urban areas are warmer than their rural counterparts (Voogt and Oke, 2003), is further intensified by global temperature increases (Li et al., 2024). Classified as an atmospheric hazard impacting a large proportion of the population (Wang et al., 2023), addressing the UHI effect requires coordinated local and national government strategies.

Urban heat mapping is an essential tool for informing these strategies and promoting sustainable urban growth. Traditional approaches to heat mapping include ground-based measurements from weather stations (Fabrizi et al., 2010), remote sensing to derive land surface temperatures (Zhou et al., 2018), and empirical models based on land use or topographic variables. However, these methods often face limitations, including low spatial resolution, limited coverage, time constraints, and inefficient data processing (Elmes et al., 2017). Additionally, their reliance on simplistic algorithms fails to capture the complexity of urban thermal climates (Shi et al., 2021; Sidiqui et al., 2022). The advent of high-resolution Earth observation satellites has improved this process, enabling more granular analyses of air temperature and microclimate variations (Coutts et al., 2016). Despite these advancements, traditional techniques still lack the adaptability and precision required for dynamic urban environments (Hung et al., 2006; Nichol et al., 2009).

In parallel, artificial intelligence (AI) has increasingly been employed to address diverse urban challenges, from transportation to climate risk management (Faisal et al., 2021; Ye et al., 2025). Urban AI offers a transformative approach to urban heat mapping by integrating machine learning with geospatial data, enabling more accurate and efficient analysis. Recent studies have used algorithms such as Artificial Neural Networks (ANN), Random Forests (RF), and eXtreme Gradient Boosting (XGB) for microclimate prediction (Guo et al., 2024; Li et al., 2023). However, these models often struggle to capture spatial hierarchies and contextual dependencies in complex urban environments (Bhakare et al., 2024; Li et al., 2024). In contrast, Convolutional Neural Networks (CNNs) offer pixel-level precision and spatial awareness, making them more suitable for geospatial applications (Nguyen et al., 2022; Ronneberger et al., 2015).

Building on this foundation, the U-Net model has gained prominence for its encoder–decoder structure and skip connections that retain both high- and low-level features. Originally introduced for biomedical segmentation (Ronneberger et al., 2015), U-Net is now widely adopted in urban applications for its ability to preserve spatial detail and delineate boundaries effectively (Ramani et al., 2024). These features are particularly important for urban heat mapping, where small-scale thermal gradients must be detected across varied urban morphologies. Comparative studies show U-Net’s consistent superiority over FCN, SegNet, and even advanced variants like Attention U-Net and Residual U-Net (Supritha and Murthy, 2023; Wu et al., 2022). Rampal et al. (2022) further confirm that simpler encoder–decoder architectures like U-Net offer more interpretable, flexible, and stable learning, especially in geospatial contexts (Afshari et al., 2023).

Recent studies validated the feasibility of using RGB satellite imagery for estimating land surface temperature (LST) via deep learning models. Although RGB data lacks thermal bands, it provides rich spatial and morphological features correlated with surface heat distribution. Zhang et al. (2018) demonstrated that convolutional neural networks can infer LST from RGB images by learning urban structural patterns. Similarly, Zhao et al. (2023) achieved high spatial accuracy in predicting LST using only visible bands. More recently, Abunnasr and Mhawej (2023) introduced a high-resolution downscaling model (HSR-LST) that predicted LST using RGB data with a mean absolute error of 0.88°C when validated against airborne thermal imagery. These studies support the use of RGB as a practical and scalable alternative, especially in data-constrained urban environments.

This study leverages the U-Net architecture within an Urban AI framework to generate high-resolution heat maps from RGB imagery, eliminating the need for direct thermal inputs. Traditional urban heat mapping depends heavily on thermal remote sensing, which often suffers from low spatial resolution, high acquisition costs, and limited frequency (Elmes et al., 2017). In contrast, RGB data is widely available and, when combined with U-Net’s capabilities, offers a scalable, cost-effective, and practical alternative for real-time urban heat analysis.

The model effectively captures fine-scale variations across urban typologies, including dense city centres, suburban areas, and vegetated zones, yielding consistent and interpretable results (Morgan et al., 2024). In addition to its accuracy, U-Net is computationally efficient and accessible for implementation in cities with limited technical resources. It supports urban planning by informing interventions such as green infrastructure, reflective surfaces, and cooling corridors.

Furthermore, its ability to generate heat maps in near real-time makes it valuable for emergency response during heat waves. By integrating Urban AI with planning systems, this study demonstrates a scalable and adaptive approach to urban climate resilience. As cities face escalating thermal risks, AI-driven heat mapping represents a critical step toward more sustainable, liveable, and climate-responsive urban development.

Research design

Study area and data collection

This study focused on urban heat mapping within the metropolitan area of Adelaide, South Australia (Supplemental Figure S1). Adelaide’s urban landscape, comprising densely populated city centres, suburban neighbourhoods, and surrounding green spaces, offers a dynamic and varied environment for analysing urban heat distribution. This diversity provides an ideal case study for assessing how different urban forms and infrastructure impact surface temperatures, making Adelaide a valuable model for understanding UHI effects in cities with similarly mixed landscapes. The selection of Adelaide allows for a detailed analysis of how densely built-up areas, green spaces, and suburban layouts each contribute uniquely to localised temperature patterns.

Data sources and preparation

High-resolution thermal and spatial data were obtained from Data SA, the South Australian Government Data Directory (SA Dataset, 2024). This dataset provides advanced satellite imagery capable of capturing detailed surface temperature variations across both urban and non-urban areas. The thermal data were analysed at a spatial resolution of 2m, where each pixel value corresponds to land surface temperature (LST). This fine spatial resolution enables the precise identification of localised thermal variations, a key requirement for understanding UHI patterns and their relationship with urban morphology and vegetation. High-resolution imagery was specifically selected to capture fine-scale variations in surface heat intensity, thereby enabling a precise examination of temperature distribution. This level of detail is essential for identifying UHI patterns, understanding how temperature gradients shift across different urban layouts, and supporting strategies for urban heat mitigation.

The dataset used in this study contained spatial and thermal information at a resolution of 1024 × 1024 pixels per image, offering a balanced combination of spatial granularity and computational feasibility, which is particularly suitable for deep learning applications. The dataset captures a single temporal snapshot during summer, with recorded temperatures ranging from a maximum of 33°C to a minimum of 16°C. While these conditions provide critical insights into peak urban heat scenarios, the static nature of the dataset limits its ability to account for seasonal variability and dynamic changes in urban heat patterns.

The dataset includes critical spatial and thermal details that allow the model to learn the impacts of urban morphology and vegetation on UHI patterns. High-density urban areas with impervious surfaces, such as roads and rooftops, retain more heat, resulting in localised hotspots. Conversely, green spaces, including parks and vegetation, mitigate urban heat through natural cooling processes such as evapotranspiration. The fine spatial resolution of 2 m enables the model to detect subtle thermal gradients and distinguish between densely built and vegetated areas. By leveraging these capabilities, the U-Net model captures complex interactions between urban structure, vegetation, and thermal dynamics, delivering actionable insights for targeted UHI mitigation strategies.

This study focused on high-resolution thermal mapping as the primary approach for urban heat assessment. Although variables such as humidity and wind influence urban heat dynamics, they were not included in this model due to constraints in data availability, computational efficiency, and the study’s focus on land surface temperature mapping.

Data normalisation

As a preparatory step, the values of each pixel in the satellite imagery were normalised to a continuous range [0, 1]. Normalising pixel values is particularly beneficial in deep learning, as it helps prevent issues such as vanishing or exploding gradients, thereby promoting stable training and faster convergence. This step is essential for achieving consistent results and improving the overall effectiveness of the model training process.

Ground truth heat maps

Urban heat maps representing ground-truth heat intensity were converted into grayscale for single-channel processing, where pixel intensity represents relative heat levels across the urban landscape. This simplified format focuses the model’s learning on heat intensity distribution without colour channel complexities. Following prediction, the grayscale output was colour-mapped through a predefined lookup table (LUT) for enhanced interpretability. This colourisation aligns heat intensities with visually distinct colour bands, which is crucial for visual assessments.

Training-validation split

The dataset was split into training (80%) and validation (20%) subsets to evaluate the model’s generalisation capability. This ratio was selected to maximise the training set size, ensuring the model could learn robust spatial and thermal patterns while retaining sufficient validation data to assess performance on unseen samples.

To ensure the model generalised effectively, validation error was used as the primary indicator. The objective was to minimise training errors without compromising validation performance. If validation error began increasing while training error continued decreasing, it indicated overfitting, where the model memorised training data rather than learning meaningful patterns. To mitigate this, validation loss was continuously monitored, and early stopping was applied, halting training when validation performance began to deteriorate.

U-net model architecture

U-Net CNN architecture was employed owing to its strengths in pixel-wise image-segmentation tasks. Originally developed for biomedical image segmentation, U-Net’s encoder–decoder structure has proven effective in diverse applications where both fine and broad spatial details are essential to evaluate the output (Ronneberger et al., 2015). Its architecture allows the model to retain the spatial details of the input image, making it highly suitable for mapping tasks (Supplemental Figure S2). The U-Net model facilitates the detection of urban heat patterns by learning both local and global features, effectively balancing spatial accuracy with computational efficiency.

Encoder path

The encoder consists of four convolutional blocks, each performing two convolutional operations with ReLU activations and batch normalisation to standardise outputs. Each block includes a max-pooling layer to downsample the feature maps, successively capturing broader and more abstract spatial features as the resolution decreases. This hierarchical feature extraction process enables the model to capture complex urban thermal landscape patterns, including isolated hotspots and expansive heat-affected areas.

Bottleneck layer

The bottleneck layer, located at the centre of the model, contains 1024 filters that capture high-level feature representations of urban heat intensity. This dense feature layer compresses and abstracts the learnt patterns from the encoder, enabling the model to recognise global spatial dependencies. This plays a crucial role in bridging the encoded feature representations with the decoder, where these abstracted features are used to generate the final output, such as identifying heat-prone zones.

Decoder path with skip connections

The decoder path restores the spatial resolution of the encoded features, gradually transforming them back to the original image dimensions. Upsampling is performed through transposed convolutions, which learn to map lower-resolution feature maps to higher resolutions. Skip connections play a crucial role by linking each encoder layer with its corresponding decoder layer, allowing the decoder to access high-resolution features from the encoder. These connections reintegrate fine-grained spatial details, which enhances the ability of the model to accurately localise high-heat areas within urban grids. Skip connections help prevent information loss during down-sampling, ensuring that critical details such as edges and small features vital for UHI detection, are preserved across the network layers.

Output layer

The output layer consists of a single convolution with a sigmoid activation function, producing a grayscale heat map where each pixel value indicates relative heat intensity. Post-processing applies a colour LUT to transform grayscale predictions into a coloured heat map, making the model output visually accessible for planners and decision-makers. This intuitive colour coding facilitates quick identification of areas most affected by UHI.

Model training and optimisation

The U-Net model was trained using a GPU-based environment, which is necessary for handling large-scale data and complex architecture efficiently. Model training was guided by a range of hyperparameters optimised to ensure effective convergence and robust performance on unseen data.

Loss function

Mean squared error (MSE) loss function was selected, aligning with the continuous nature of the heat intensity target variable. Unlike binary segmentation tasks, in which the objective is to classify discrete classes, UHI mapping requires precise intensity prediction. MSE measures the squared difference between predicted and actual pixel values, penalising larger deviations and ensuring a focus on pixel-wise accuracy across the entire heat map.

Optimisation algorithm and learning rate

Adam optimiser was applied for its adaptability in handling large-scale, sparse gradients, which is an advantage in deep learning with high-dimensional data. An initial learning rate of 1e−4 was selected to balance convergence speed with stability. An adaptive learning rate scheduler further adjusted this rate, reducing it by a factor of 0.1 if validation loss did not improve after eight consecutive epochs, effectively fine-tuning the learning process allowing the model to refine its solution while avoiding overshooting.

Training duration and batch size

To capture detailed patterns within urban heat maps, the model was trained over 400 epochs with a batch size of 2. This batch size was selected to balance memory constraints imposed by high-resolution imagery and help avoid local minima. Prolonged training allows the model to learn intricate spatial and thermal patterns, with careful monitoring to prevent overfitting. Regularisation techniques and validation performance checks help ensure that the learnt patterns generalise well to diverse urban areas, thereby improving the ability of the model to generate accurate predictions on unseen data.

Training-validation split

A training–validation split of 80/20 was adopted in this study based on a sensitivity analysis conducted using different ratios, including 85/15 and 70/30. The 80/20 configuration provided the best balance between training data sufficiency and model generalisation performance. This setup aligns with typical practices in deep learning and ensures the robustness of the model’s predictive capabilities on unseen data.

Evaluation metrics

The efficacy of the model was assessed using a combination of quantitative and qualitative metrics, specifically designed to align with the requirements of urban heat mapping.

Quantitative metrics – mean squared error (MSE), peak signal-to-noise ratio (PSNR), and structural similarity index (SSIM)

MSE served as the primary quantitative metric, reflecting the pixel-wise accuracy of the heat intensity predictions. PSNR was used to evaluate the visual fidelity of the predicted heat maps in comparison to the ground truth, while SSIM assessed the model’s ability to preserve spatial patterns and structural consistency. Together, these metrics provide a comprehensive evaluation of the model’s effectiveness in replicating real urban heat distributions, where lower MSE and higher PSNR and SSIM values signify more substantial alignment with the reference data.

Qualitative analysis

In addition to quantitative metrics, qualitative visual inspections were performed to evaluate the model’s predictions. The predicted heat maps were visually compared against actual heat distributions, providing a practical assessment of spatial alignment and the model’s accuracy in localising high-heat areas. This step is essential in applications such as UHI mapping, where real-world relevance and spatial coherence are critical for decision-making.

Analysis and results

The following section presents a comprehensive evaluation of the U-Net model’s performance in predicting urban heat distribution, detailing quantitative accuracy metrics, visual consistency with ground truth data, robustness across diverse urban environments, and error sources.

Model performance and training metrics

The U-Net model was trained on high-resolution satellite imagery of urban areas, utilising an 85/15 split between training and validation data. The model’s training process captured both fine-scale and broad thermal patterns within urban heat distribution. Over 400 epochs, the model’s training performance was monitored using MSE as the primary loss function (Supplemental Figure S3). The training loss consistently decreased from an initial MSE of approximately 0.0013 to a final value of 0.0007, indicating effective convergence. The validation loss followed a similar decreasing trend, confirming the ability of the model to generalise to unseen data.

The model was optimised using the Adam algorithm with an initial learning rate of 1e−4. A learning rate scheduler dynamically adjusted the learning rate by a factor of 0.1 when the validation loss plateaued for eight consecutive epochs, refining the convergence process. A batch size of 2 was selected to balance memory requirements for processing 1024 × 1024 resolution images while capturing urban thermal patterns effectively. This combination of extended training, dynamic learning rate adjustments, and optimal batch size supported the model in learning complex spatial dependencies, ensuring robust performance across various urban environments.

Visual analysis of predicted heat maps

A comparison of satellite images, ground truth heat maps, and U-Net model’s predicted heat maps for four distinct urban areas (A, B, C, and D) is presented in (Supplemental Figure S4). Starting with Area A, the model successfully identified high-temperature zones within densely built-up regions, while cooler temperatures were accurately represented in nearby green spaces. This differentiation between urban infrastructure and vegetated areas closely aligned with the ground truth, highlighting the model’s sensitivity to variations in land cover.

In Area B, which features an open green space surrounded by residential buildings, the model predictions showed lower temperatures within the vegetative area and higher temperatures in the adjacent built environment. This result reflects the model’s capability to capture thermal gradients influenced by both natural cooling effects and nearby urban structures. In densely populated Area C, elevated temperatures appeared primarily along major roads and rooftops, where surface materials are known to absorb and retain heat. The predicted map closely mirrored the ground truth, demonstrating the efficacy of the model in identifying the typical thermal patterns associated with high-density urban zones.

Finally, Area D, with its complex layout of curved streets and varied building densities, further illustrates the model’s ability to generalise across intricate urban environments. Temperature gradients were accurately represented, with higher temperatures in dense building clusters and cooler values in vegetated zones. This consistency across diverse urban settings underscores the robustness and adaptability of the model in predicting thermal patterns across structured and complex landscapes.

Although the model performed consistently well, it exhibited minor inaccuracies in specific urban environments, particularly in areas with high variability in building density, surface materials, and environmental context. In these heterogeneous settings, the model showed slight deviations from the ground-truth heat maps, particularly in regions where mixed land cover introduced complex thermal patterns.

In densely built urban centres with minimal vegetation, the model sometimes underestimated the intensity of UHIs. In contrast, in highly vegetated zones, the model occasionally predicted marginally higher temperatures than observed. These discrepancies were most prominent in urban areas with a combination of concrete and vegetative surfaces, where thermal variation was complex.

Quantitative evaluation metrics

The final model performance was assessed using MSE, metrics well-suited for continuous temperature prediction. On the validation set, the model achieved an MSE of 0.00383, reflecting its ability to produce pixel-level predictions that closely matched the ground-truth heat intensities. MSE provided insight into the model’s effectiveness in minimising larger errors.

These metrics were selected to assess the accuracy of model predictions in capturing real urban heat distributions. Due to the continuous nature of the output, segmentation metrics such as Intersection over union (IoU) and pixel accuracy were not applicable, as they are intended for categorical segmentation tasks. Nonetheless, MSE directly assessed the capacity of the model to predict fine-scale temperature gradients, affirming its suitability for urban heat mapping applications. In addition to its predictive accuracy, the model demonstrated efficient processing, with each image processed in under 30 seconds. This rapid processing time is advantageous for large-scale urban heat mapping.

To further evaluate the visual and structural quality of the generated heat maps, we also calculated Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM). A PSNR of approximately 18.7 dB indicates that the predicted heat maps are visually consistent with the ground-truth references, offering sufficient detail for spatial interpretation. Additionally, the SSIM value of up to 0.378 demonstrates the model’s ability to preserve spatial thermal patterns across different urban morphologies. Together, these complementary metrics confirm the model’s robustness in both numerical prediction and visual fidelity – attributes that are critical for practical urban analysis and planning applications.

Findings and discussion

This study demonstrates the efficacy of the U-Net model, an advanced Urban AI tool, in generating high-resolution urban heat maps, providing precise assessments of thermal distribution patterns across Adelaide, South Australia. The model’s high predictive accuracy, validated by low MSE value, underscores its ability to capture nuanced thermal variations within diverse urban landscapes. Its rapid processing time of under 30 seconds per image highlights its computational efficiency, making it an essential resource for large-scale urban heat mapping applications as well as emergency response during extreme heat events. These findings represent significant contributions to the broader field of urban heat assessment, addressing both accuracy and scalability – key elements for effective climate resilience planning.

Model performance and Implications

As an Urban AI solution, the U-Net model delivers high-resolution outputs that are particularly valuable for identifying microscale temperature variations in urban environments. This granular spatial detail enables urban planners to pinpoint heat-prone zones within dense infrastructure, facilitating targeted interventions such as green infrastructure expansion and the use of reflective materials, both of which are proven to reduce UHI intensity (Bowler et al., 2010; Gago et al., 2013; Shaamala et al., 2024). Furthermore, the model’s rapid processing capability aligns with operational requirements for extreme weather events, offering real-time decision-making support for mitigating heat risks (Chen et al., 2006; Johnson et al., 2012).

The U-Net model’s robust performance across varied urban morphologies highlights its adaptability. It accurately identified high-temperature zones in densely built areas, such as roads and rooftops, where impervious surfaces retain heat (Rizwan et al., 2008; Voogt and Oke, 2003). In suburban areas, it effectively captured lower temperatures associated with vegetation, reflecting greenery’s cooling effects (Bowler et al., 2010; Weng, 2009). However, discrepancies arose in mixed land cover areas, where complex interactions between concrete and vegetation create challenging thermal patterns. These findings emphasise the importance of integrating contextual environmental variables, such as shading and material reflectivity, to further enhance accuracy in heterogeneous urban environments (Gago et al., 2013; Middel et al., 2014).

Compared to traditional methods, the U-Net model exemplifies the transformative potential of Urban AI. Traditional techniques often rely on moderate-resolution satellite data (e.g. MODIS and Landsat) with extensive preprocessing, limiting their effectiveness for precise UHI analysis (Shi et al., 2021). In contrast, the U-Net’s encoder–decoder architecture, combined with skip connections, preserves critical pixel-level spatial features, enabling highly accurate identification of small-scale thermal gradients (Briegel et al., 2023; Ronneberger et al., 2015). This capability is indispensable for managing heterogeneous urban environments, where targeted interventions are vital for mitigating localised heat challenges (Diem et al., 2024; Elmarakby and Elkadi, 2024).

The U-Net model’s operational efficiency further cements its role as a foundational Urban AI agent. Its ability to rapidly assess thermal patterns supports timely decision-making during extreme heat events, addressing climate resilience needs effectively. Unlike traditional LST approaches that face significant delays (Tomlinson et al., 2011; Zhou et al., 2018), the U-Net model ensures prompt, data-driven urban planning and climate-adaptive design (Almashhour et al., 2024; Kumar et al., 2024).

Future advancements could include integrating additional environmental variables, such as humidity, wind patterns, and solar radiation, to better contextualise temperature patterns and enhance sensitivity to complex urban environments (Ghorbany et al., 2024; Liu and Morawska, 2020; Tehrani et al., 2024). Expanding the training data to encompass cities with diverse climates and urban morphologies would strengthen the model’s adaptability and robustness, ensuring its usability in varied geographic contexts. Further exploration of hybrid models combining U-Net with temporal data analysis, such as recurrent neural networks, could refine its predictive accuracy for continuous seasonal assessments (Demiss and Elsaigh, 2024; Nazir et al., 2024).

Evaluating the U-Net model across multiple cities would validate its scalability and establish a framework for Urban AI-enabled heat mapping. Such a framework could provide tailored insights into diverse thermal profiles, enabling planners to design interventions for location-specific UHI challenges (W. Zhou et al., 2011; Li et al., 2017). By aligning with global urban resilience goals, this Urban AI-driven approach fosters sustainable urban development, improves liveability, and enhances climate adaptability (Degirmenci et al., 2021; Ramyar et al., 2021).

Urban heat resilience and vulnerability in Adelaide

This study highlights significant variations in urban heat resilience across metropolitan Adelaide, shaped by land use, urban morphology, and vegetation cover. High-resolution mapping using the U-Net model reveals that densely built areas, including the CBD, industrial precincts, and transport corridors, record the highest land surface temperatures (LSTs) due to impervious surfaces and heat-retaining materials (Supplemental Figure S5).

Conversely, suburban neighbourhoods with green spaces exhibit moderate resilience, benefiting from tree cover and open spaces that help regulate temperature extremes. However, new high-density residential developments with limited vegetation contribute to localised warming, underscoring the need for climate-responsive urban design.

Coastal zones, parklands, and peri-urban areas demonstrate the greatest heat resilience, with lower LSTs due to the cooling effects of vegetation, water bodies, and open landscapes. These findings are in line with Morgan et al. (2024) and reinforce the crucial role of green and blue infrastructure in mitigating urban heat stress and enhancing urban climate resilience.

Practical Implications for urban planning

Urban planners and local governments currently rely on a combination of coarse-resolution thermal satellite imagery (e.g. MODIS or Landsat), empirical models, and in situ measurements to inform their urban heat mitigation strategies. These include tree canopy targets, zoning incentives for reflective materials, and emergency heatwave plans. However, existing approaches often lack the spatial granularity and real-time responsiveness required for street-level interventions.

The proposed U-Net model addresses these limitations by generating high-resolution, near real-time heat maps from widely available RGB imagery, requiring less than 30 seconds per image. This enables time-sensitive applications, such as identifying high-risk zones during extreme heat events and supporting proactive planning decisions. For instance, city councils could integrate these maps into GIS platforms to prioritise tree planting on heat-vulnerable streets, evaluate zoning compliance, or assess the thermal impact of new developments.

Moreover, prior studies have validated the effectiveness of RGB-based LST estimation, confirming that models trained on morphological and spatial patterns can reliably predict thermal behaviour across different urban contexts – even in the absence of thermal infrared data (Zhang et al., 2018; Zhao et al., 2023). This makes RGB-based approaches especially valuable for cities with limited access to thermal sensors or commercial satellite products.

The model also complements existing strategies by offering a scalable, low-cost solution that supports a continuous assessment of urban thermal dynamics. Nonetheless, its current reliance on land surface temperature restricts its ability to capture air temperature or thermal comfort directly. Therefore, it should be seen as a complementary decision-support tool rather than a replacement for physical sensor networks. Future integration with real-time sensing and urban analytics platforms could further enhance its value and planning impact.

Limitations and future research

While the U-Net model demonstrated strong overall performance, certain limitations emerged in areas with extreme thermal dynamics, presenting unique challenges. For example, in highly urbanised zones with minimal vegetation, the model occasionally underestimated peak temperatures. This underestimation likely stems from the limited representation of high-density UHI conditions in the training dataset. Incorporating additional data that captures extreme urban heat scenarios could improve the model’s sensitivity, ensuring more accurate predictions in such contexts. Expanding the dataset to include nuanced environmental variables would enhance the model’s capacity to perform reliably in diverse urban environments.

Another limitation lies in the model’s exclusive reliance on data from Adelaide, South Australia, which restricts its generalisability to regions with varying climates, urban layouts, and vegetation types. Although the model achieved commendable results within Adelaide, its application to other geographic areas may necessitate retraining or fine-tuning to account for unique environmental characteristics. Broadening the training dataset to include a wider range of urban forms, climates, and land-use types would significantly enhance the model’s robustness and adaptability, enabling it to deliver accurate results across diverse global landscapes.

Additionally, the static nature of the data used in this study constrains the model’s ability to account for seasonal variations in urban heat dynamics. Changes in solar radiation, atmospheric conditions, and vegetation states significantly influence surface temperature patterns, which are not captured in the current model. Incorporating seasonal variability into the dataset could provide a more dynamic perspective on urban thermal patterns, strengthening the model’s predictive power for year-round applications. Such enhancements would allow for more comprehensive analyses of heat distribution under varying environmental conditions.

Furthermore, while the model estimates land surface temperature, it does not capture air temperature or thermal comfort directly. This distinction is important, and future work could incorporate sensor-based measurements or predictive thermal comfort indices to enhance the relevance for human-centred heat risk assessment.

Addressing these limitations through data expansion, the integration of additional variables, temporal modelling, and computational optimisation would considerably improve the U-Net model’s adaptability, accuracy, and scalability. These advancements would enable the model to perform effectively across a broader range of urban settings, enhancing its role as an invaluable tool for sustainable urban planning and climate resilience. By refining its capabilities, the model could support more precise and actionable insights, empowering urban planners and policymakers to implement targeted interventions that mitigate UHI effects and promote climate-adaptive urban development.

Conclusion

This study demonstrates the U-Net deep learning model as a transformative tool for high-resolution urban heat mapping, offering an innovative approach to predicting land surface temperature (LST) from RGB satellite imagery. By eliminating reliance on direct thermal data, this method provides a cost-effective, scalable, and accessible solution for urban heat monitoring, particularly in cities where thermal imagery is limited. The encoder–decoder architecture with skip connections enables pixel-level precision, effectively capturing fine-scale thermal variations across Adelaide’s diverse urban landscape. This capability is particularly valuable for identifying localised heat-prone areas, allowing planners to implement targeted heat mitigation strategies, such as expanding green infrastructure or incorporating reflective materials. The model’s robust performance, as reflected in low MSE value, confirms its suitability for real-world urban climate analysis.

One of the U-Net model’s standout contributions is its ability to process high-resolution heat maps in under 30 seconds per image, a marked improvement over traditional thermal mapping methods, which often require extensive preprocessing and offer less spatial precision. This computational efficiency makes the model highly practical for large-scale and near real-time applications, particularly during heat emergencies when rapid decision-making is crucial for public health interventions. By bridging the gap between advanced geospatial modelling and practical application, this study supports dynamic urban planning, providing data-driven insights that policymakers can use to promote sustainable urban development and climate resilience.

While the U-Net model has demonstrated strong potential, opportunities exist for further refinement. Expanding the training dataset to include diverse urban morphologies, climates, and land-use types would improve generalisability and accuracy. Additionally, integrating environmental variables such as wind patterns, humidity, and solar radiation could enhance predictive capability by offering a comprehensive understanding of urban microclimates. Future research could explore hybrid models that combine U-Net with temporal analysis techniques, improving the ability to predict seasonal variations and adapt to extreme climate conditions.

In conclusion, this study highlights the potential of RGB-based Urban AI models in high-resolution urban heat mapping, providing an innovative solution to a critical urban challenge. This framework empowers planners and policymakers to mitigate UHI effects, optimise green infrastructure, and enhance public health strategies by delivering precise, rapid, and actionable insights. Its scalability and adaptability position it as a valuable tool for climate-responsive urban planning, supporting the transition towards sustainable, liveable, and thermally resilient cities.

Supplemental Material

Supplemental Material – Leveraging urban AI for high-resolution urban heat mapping: Towards climate resilient cities

Supplemental Material for Leveraging urban AI for high-resolution urban heat mapping: Towards climate resilient cities by Abdulrazzaq Shaamala, Niklas Tilly, and Tan Yigitcanlar in Environment and Planning B: Urban Analytics and City Science.