The work forms part of the OLADAPT project, funded by Spain’s Ministry of Science, Innovation and Universities, and is being conducted in collaboration with the Polytechnic University of Madrid, the University of Extremadura and the UNESCO Chair in Sustainable Development and Environmental Education at the EHU. The researchers aligned their approach with the guidelines of the Intergovernmental Panel on Climate Change, which frame climate risk as the product of hazard, exposure and vulnerability. In the case of urban heat, that framing matters enormously, because the hazard itself—the temperature a given area experiences—is only one ingredient in the equation. How a city is built, which areas are hardest hit by high temperatures, and how many people are exposed and how little protection they have, all combine to determine the real risk to human health, thermal comfort and the liveability of urban spaces.

“We are looking at how to conduct a multi-dimensional and multi-scale assessment of the risk posed by heatwaves in urban areas. The risk posed by heatwaves is not measured solely by temperature; it has multiple dimensions,” explains Irantzu Álvarez, a researcher in the SAREN group, which is led by Leire Garmendia. That multidimensional perspective is what distinguishes the new methodology from coarser city-level assessments. Rather than treating a municipality as a single unit of analysis, the team measured vulnerability at the level of individual buildings and open public spaces, generating fine-grained vulnerability maps that reveal hotspots of risk that would otherwise remain invisible in aggregate statistics.

Building the maps required assembling a large volume of data drawn from a variety of sources. The researchers incorporated socioeconomic variables describing the people who live and work in each location, detailed characteristics of the buildings themselves, and environmental variables capturing the conditions of the surrounding urban environment. This combination is crucial, because sensitivity to heat depends on both the physical fabric and the human context. “A building may be sensitive to heat, but its residents may have the financial means to protect their home from it,” the researchers explain. In other words, a thermally poor building occupied by wealthy residents may pose less of a welfare problem than a marginally better building occupied by people with no means to install cooling, improve insulation or otherwise shield themselves from extreme temperatures.

The study’s most consistent finding across all three cities is the weight of socioeconomic variables in determining vulnerability. “People on low incomes, older people, etc. are far more vulnerable,” the research group states. This pattern held in Bilbao, Madrid and Cáceres despite their very different climates, suggesting that the social gradient of heat risk is a structural feature of Spanish cities rather than an artefact of any particular climate zone. Just as importantly, the analysis showed that vulnerability is not uniform within a city, and may even vary substantially within a single neighbourhood. It is therefore not enough, the researchers argue, to declare one city more vulnerable than another; detailed, building-by-building analysis is needed to locate the specific hotspots where intervention will save the most people from harm.

The choice of the individual building as the unit of analysis is what gives the methodology its practical power for municipal decision-making. “This methodology shows which buildings or types of buildings should be renovated or adapted first,” say Álvarez and Laura Quesada, the two researchers who spearheaded the analysis. Quesada is an architect specialising in the restoration and conservation of historic heritage and works at the Vitoria-Gasteiz School of Engineering, while Álvarez is a Technical Engineer in Surveying with a degree in Geography, based at the Bilbao School of Engineering; both lecture on the Master’s Degree in Construction Engineering at the Bilbao School of Engineering. Their combined expertise reflects the interdisciplinary character of the work, which spans building physics, urban geography and social science.

A second layer of the methodology links the vulnerability of buildings to that of the open public spaces around them. The researchers measured the impact of the surrounding environment on each building, capturing factors such as green cover and shade. “Buildings may be highly vulnerable but have plenty of green areas and shade around them. Taking all of this into account means that adaptation measures can be implemented in different ways,” the team explains. A vulnerable building in a leafy, shaded street may call for a different response than an equally vulnerable building on a hot, exposed plaza. By treating buildings and open spaces as an interconnected system rather than as separate problems, the maps allow planners to design adaptation strategies that operate at the scale of the streetscape as well as the individual structure.

The practical output of the methodology is a set of urban vulnerability maps that clearly indicate where priority action should be taken to tackle heatwaves. “Adaptation measures can be implemented much more effectively with this type of study,” the researchers say. For municipalities operating under tight budgets, this kind of prioritisation is invaluable: rather than spreading resources thinly across an entire city, officials can target renovation programmes, shading installations, green infrastructure and social support towards the buildings and neighbourhoods where the combination of physical exposure and social vulnerability is greatest. The maps thus translate abstract climate risk assessments into concrete, spatially explicit guidance.

Demonstrating the approach in three cities with distinct climates also served as a test of its transferability. “The methodology is fully replicable; it can be used anywhere. This makes the methodology more robust. It is a genuinely useful tool for municipalities and cities,” the group states. The successful application in Bilbao, Madrid and Cáceres suggests that the framework could be adopted by other cities facing intensifying heat risk, both in Spain and further afield, provided comparable socioeconomic, building and environmental data are available.

The broader context for this research is the steady escalation of extreme weather under climate change. Heatwaves are becoming more frequent and more severe, with direct consequences for public health, thermal comfort and the overall liveability of cities. Because urban populations are concentrated and much of the built fabric is long-lived, the decisions cities make now about which buildings to retrofit and which public spaces to green will shape heat resilience for decades. By demonstrating that vulnerability is a local, block-by-block phenomenon shaped jointly by poverty, age, building quality and the surrounding environment, the SAREN team’s work offers city governments a rigorous, replicable way to direct adaptation resources where they will matter most—and a reminder that the people least able to protect themselves from extreme heat are often concentrated in the very places that need help first.

Subject of Research: Multiscale GIS-based assessment of urban heatwave vulnerability in Spanish cities

Article Title: Vulnerability to heatwaves varies greatly from one area to another within cities

Article References: Vulnerability to heatwaves varies greatly from one area to another within cities. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: heatwaves, urban vulnerability, climate adaptation, socioeconomic factors, buildings, public spaces, GIS mapping, Bilbao, Madrid, Cáceres, IPCC, urban resilience

The researchers set out to evaluate how feature importance and feature selection methods perform when the goal is rapid, reliable estimation of seismic risk levels for reinforced concrete buildings. Rather than treating machine learning as an opaque oracle, they deliberately interrogated it, asking which input parameters drive predictions and whether a trimmed-down set of inputs could match, or even beat, models fed the complete feature set. This dual focus on accuracy and interpretability places the work squarely within the growing field of explainable artificial intelligence, where the goal is not just prediction but understanding.

To build a trustworthy foundation, the team used a simulation-based dataset representing an existing building stock rather than a purely synthetic or idealized collection of structures. This choice matters. Rapid screening methods have a long history in earthquake engineering, from the FEMA P-154 rapid visual screening handbook in the United States to the P25 screening method developed in Türkiye and the Japanese Building Disaster Prevention Association’s evaluation standard. These approaches typically rely on a handful of visually observable parameters. Machine learning promises to go further by capturing subtle, nonlinear interactions among many parameters, but only if the underlying data reflect real buildings with realistic variability.

The study employed three machine learning models to classify seismic risk, allowing the authors to compare how different algorithms weigh and use the same information. Across their analyses, a consistent pattern emerged: combining feature importance measures with explicit feature selection improved prediction performance compared with using either approach alone. Feature importance methods, such as permutation importance, quantify how much each input contributes to a model’s predictions by measuring how performance degrades when that input is shuffled or removed. Feature selection methods, including correlation-based filtering rooted in classical statistical measures such as Pearson’s correlation coefficient, systematically prune redundant or uninformative variables before training.

The standout result came from LightGBM, a gradient boosting framework known for its efficiency on tabular data, when paired with normalized feature importance and correlation-based feature selection. This combination achieved a macro-F1 score of approximately 0.91 — a strong result for a multi-class classification task — while using fewer input parameters than models trained on the full feature set. The macro-F1 metric is particularly meaningful in seismic risk classification because it averages performance across all risk classes, preventing a model from looking good simply by mastering the most common category while neglecting rare but critical high-risk cases.

That a leaner model outperformed the full-feature version is not a paradox but a well-documented phenomenon in machine learning. Irrelevant and redundant features can act as noise, encouraging models to fit spurious patterns that do not generalize. By identifying and retaining only the most influential parameters, the selection process acts as a regularizer, sharpening the model’s attention on the structural characteristics that genuinely govern seismic behavior. For low-rise reinforced concrete buildings, such characteristics typically relate to geometry, structural system, material properties, and code-era design provisions — the same factors that decades of empirical screening methods have converged upon, but now quantified and weighted by data rather than expert judgment alone.

The interpretability dimension is central to the study’s practical value. Engineers and municipal officials cannot act on a black-box score; they need to know why a building is flagged as high risk and which interventions might matter. Frameworks like SHAP, introduced by Lundberg and Lee in 2017, and LIME, proposed by Ribeiro and colleagues in 2016, have made it standard practice to attach explanations to model predictions. By combining normalized feature importance with explicit selection, Yilmaz and Dirikgil provide a transparent chain of reasoning: here are the parameters that matter most, here is the evidence that they suffice, and here is the resulting risk classification. This aligns with a broader movement in structural engineering toward explainable machine learning, exemplified by recent work applying explainable frameworks to XGBoost-based seismic assessment.

The timing of this research is not incidental. The February 2023 Kahramanmaraş earthquake sequence in Türkiye, which included magnitude 7.8 and 7.6 events, caused catastrophic damage to vast numbers of reinforced concrete buildings and prompted intensive post-earthquake damage surveys. Those surveys documented recurring vulnerabilities tied to design era, detailing practices, and construction quality, reinforcing the urgency of pre-earthquake screening. Türkiye’s successive earthquake codes, from the 1975 provisions through the current Turkish Building Earthquake Code of 2018, mean that a building’s design vintage alone carries substantial predictive information. Machine learning models trained on simulation data grounded in these code generations can encode that knowledge and apply it consistently across an entire urban building stock.

What makes the proposed framework genuinely scalable is the economy of its input requirements. Traditional detailed assessment requires structural drawings, material testing, and nonlinear analysis — resources that no city can mobilize for every building. Rapid visual screening reduces the burden but introduces subjectivity and coarse granularity. A model that achieves roughly 0.91 macro-F1 with a reduced set of critical parameters offers a middle path: data-driven consistency at a fraction of the data collection cost. The authors frame this as enabling rapid and efficient large-scale building stock assessments, with direct applications to regional seismic risk evaluation and urban-scale vulnerability screening. For municipalities prioritizing retrofit budgets, the ability to rank thousands of buildings quickly and defensibly could redirect scarce resources to the structures that need them most.

The study also contributes methodologically to the machine learning literature on feature selection. Recent comparative analyses have emphasized that different selection methods vary in both performance and stability, and that the choice of importance measure interacts with the choice of algorithm. By systematically testing combinations of three models with multiple importance and selection strategies, the researchers provide a template that other domains facing high-dimensional structural data can adapt. Their finding that the combination of normalized importance and correlation-based selection with LightGBM yields the best results is an actionable recommendation, not merely an observation. The authors note that some supporting data are included in the article, with expanded datasets available from the corresponding author upon reasonable request, supporting reproducibility in a field where simulation pipelines are often proprietary. As cities worldwide confront aging building stocks and tightening retrofit budgets, the message from Kayseri is clear: predicting which buildings will fail in the next earthquake may require far less information than anyone assumed — provided you know exactly which information to ask for.

Subject of Research: Interpretable machine learning for seismic risk prediction of low-rise reinforced concrete buildings

Article Title: Interpretable machine learning and feature selection for seismic risk prediction of low-rise reinforced concrete buildings

Article References: Yilmaz, M., & Dirikgil, T. (2026). Interpretable machine learning and feature selection for seismic risk prediction of low-rise reinforced concrete buildings. Bulletin of Earthquake Engineering. https://doi.org/10.1007/s10518-026-02667-7

Image Credits: AI Generated

DOI: 10.1007/s10518-026-02667-7

Keywords: seismic risk prediction, machine learning, reinforced concrete buildings, explainable AI, feature selection, feature importance, LightGBM, macro-F1 score, rapid screening, earthquake engineering, urban vulnerability, Turkish building stock

The researchers worked in cooperation with the University of Manchester’s African Cities Research Consortium, selecting twelve cities chosen to reflect the continent’s diversity in geography, climate, economy, culture, and urbanisation history. The set includes heavily studied metropolises such as Accra, Nairobi, Dar es Salaam, and Lagos, alongside severely under-researched cities like Bukavu, Freetown, Maiduguri, and Mogadishu. After screening more than 3,100 search results with machine-assisted tools and manual review, the team coded each documented intervention by actor, spatial scale, implementation status, response type, and potential synergies or trade-offs, then contextualised the results against a composite index combining climate exposure, social vulnerability, and infrastructure deficit.

The imbalance between adaptation and mitigation is not accidental, the authors argue, but rooted in the architecture of international climate finance. Mitigation programmes globally receive roughly 30 billion US dollars compared with 16 billion for adaptation and a mere 2.7 billion for integrated strategies. More than half of adaptation programmes, 57 percent, are loan-financed, and disbursement ratios hover around 46 percent. The cost-benefit logic of climate finance privileges measurable emission reductions over the diffuse, context-specific benefits of adaptation, creating what the study calls a paradox: African cities contribute least to global warming, with average per capita emissions of around 2.5 tonnes per year compared with 27 tonnes in cities like Rotterdam, yet the financing system disadvantages precisely those cities for whom adaptation is the urgent imperative.

To quantify where co-beneficial measures might be falling behind, the researchers introduced what they call the Synergy Gap, calculated as the difference between the general implementation rate and the rate at which actions with cross-domain benefits are implemented. Positive values signal that singular interventions are outpacing synergistic ones, a pattern the authors attribute to siloed governance structures that treat adaptation and mitigation as separate policy domains. About 32 percent of all recorded actions carried potential synergies, and the largest share, 51 percent, fell into a category the study labels the NBS-Sequestration Nexus: nature-based solutions such as tree planting, mangrove restoration, and green infrastructure that cool neighbourhoods and buffer floods while also sequestering carbon.

The geography of implementation is strikingly uneven. Nairobi, Dar es Salaam, and Lagos consistently show the highest levels of reported climate activity and implementation, while Addis Ababa, Freetown, and Mogadishu sit at the lower end. In Addis Ababa, roughly 90 percent of recorded actions consist of plans and recommendations rather than concrete implementations, despite strong problem awareness among city planners regarding drinking water, sewage, and flooding. Accra, together with Lilongwe, shows the largest Synergy Gap in the dataset, with promising policy frameworks on solid waste, energy, and transport undermined by weak cross-sectoral execution. In contrast, Nairobi, Freetown, and Maiduguri are the only three cities where more synergistic actions are implemented than singular ones, driven mainly by local and multi-actor initiatives.

One of the study’s most consequential findings concerns who actually gets things done. Households and communities achieve higher implementation rates than government-led measures across the twelve cities, and multi-actor collaborations also perform comparatively well. Autonomous actions were overwhelmingly coded at the household and community levels, with half of household actions being behavioural or cultural practices such as savings strategies and improved waste handling. Yet these are the least resourced actors in the climate finance landscape. The authors note that this partly reflects a research bias, since studies tend to document community practices only once they are already observable, while government planning is researched regardless of whether it is ever implemented. Still, the pattern underscores a structural tension: the potentially most effective actors are the least funded.

Informal settlements emerge as both the epicentre of climate vulnerability and a blind spot in the evidence base. Actions targeting informal settlements account for 37 percent of the sample, with the highest shares in Nairobi at 77 percent and Lilongwe at 67 percent, but only 3 percent in Addis Ababa despite its slum population share of around 72 percent. The study devotes considerable attention to the conceptual difficulties of informality, noting that the term is a malleable socio-political construct and that formal and informal sectors are so interpenetrated in many African cities that separating them is, as one cited scholar puts it, ridiculous. Data on informality, the largest dimension of most African cities, are rarely publicly available, and urban indicators often aggregate formal and informal dimensions in ways that obscure intra-urban inequality.

Data scarcity itself emerges as a methodological finding. For Bukavu, Freetown, Maiduguri, and Mogadishu, fewer than ten publications per city exist, and the authors caution that the apparent positions of Bukavu and Maiduguri in their comparative analyses very likely reflect data artefacts rather than actual activity levels. Mogadishu, despite facing among the most urgent climate pressures, remains one of the least researched capital cities in Africa, with minimal solar and wind adoption despite favourable conditions. The authors stress that reported action levels are as much a function of research attention as of actual implementation, an epistemic bias they link to the framing of Southern Urbanism, which rejects applying Euro-American urban theory to cities shaped by colonial and postcolonial dynamics.

Trade-off risks, though less frequent than synergies, cluster in worrying ways. Of 268 potential trade-offs recorded, representing 14 percent of all actions, half relate to social exclusion, followed by potential ecosystem harm at 26 percent. The authors warn that when technical and utilitarian values dominate nature-based interventions, the social and cultural dimensions of human-nature relations central to how African communities experience climate risk are marginalised. Excess rainfall is the most frequently recorded climate stressor across half the cities, associated with nearly 60 percent of preventative adaptation goals, while temperature extremes concentrate in Accra, Addis Ababa, and Dar es Salaam.

The study’s conclusions are deliberately framed as hypotheses rather than definitive rankings, given the exploratory nature of comparing cities with such uneven evidence bases. But its prescriptions are clear: integrated, multi-scalar approaches anchored in better data and dynamic frameworks are essential to align urban climate action with Africa’s heterogeneous urban realities. The authors call for flexible, co-produced typologies that can distinguish settlements appearing similar from above but diverging on the ground, dedicated funding windows for community-government coalitions, and governance reforms that overcome the compartmentalisation of adaptation and mitigation. As the scientific community prepares the IPCC’s forthcoming Special Report on Cities, the message from these twelve African cities is that integration is not simply another box to check, but a precondition for climate action that works.

Subject of Research: Integrated climate change adaptation and mitigation interventions, synergies, and co-benefits in African cities

Article Title: Lessons from integrated climate change adaptation and mitigation interventions for synergies and co-benefits in African cities

Article References: Bodenstein, C.-P., & Lwasa, S. (2026). Lessons from integrated climate change adaptation and mitigation interventions for synergies and co-benefits in African cities. Discover Cities, 3(1), Article 174. https://doi.org/10.1007/s44327-026-00359-0

Image Credits: AI Generated

DOI: 10.1007/s44327-026-00359-0

Keywords: climate adaptation, climate mitigation, African cities, synergy gap, co-benefits, nature-based solutions, informal settlements, urban climate finance, Southern Urbanism, climate governance, urban vulnerability, infrastructure deficit