The study focused on eight monitoring stations operated by India’s Central Pollution Control Board in Ahmedabad, one of the world’s most air-pollution-stressed cities. The team set out to predict the AQI twenty-four hours ahead at every station simultaneously, a task that naturally invites spatio-temporal models designed to exploit correlations between neighboring sensors. They compared two graph-based deep learning architectures, a Spatio-Temporal Graph Convolutional Network known as ST-GCN and a hybrid GCN-LSTM model, against four per-station approaches: Random Forest, XGBoost, Long Short-Term Memory networks, and Gated Recurrent Units. Every model faced the identical test set, evaluated across 31,831 usable station-hours, ensuring that differences in performance could be attributed to the models themselves rather than to favorable data splits or evaluation tricks.

One of the study’s most important contributions lies in how the target variable was constructed. Rather than computing a generic pollution score, the authors followed the CPCB’s official AQI methodology, which assigns pollutant-specific averaging periods, sub-index breakpoints, and strict data-completeness standards before a station’s daily index can be calculated. This matters because the AQI is the number that public health advisories are actually built on, and a model that excels at predicting raw PM2.5 concentrations may still fail when those concentrations are translated into the categorical scale that triggers warnings. By anchoring the prediction task to the regulatory definition, the researchers aligned their benchmark with the question that genuinely matters to city officials and residents: what will the official index be tomorrow?

The preprocessing pipeline also received unusual care. All data-dependent operations, including gap handling, outlier capping, and normalization, were fitted exclusively on the training segment of the data, eliminating a subtle but pervasive form of information leakage known as data snooping. In many published air quality studies, preprocessing statistics computed over the full dataset quietly leak information from the future into the training process, inflating reported accuracy. By restricting every transformation to training data and applying identical evaluation masks to all models on a shared test index, the authors built a benchmark whose results can be trusted, even when those results are unflattering to the fanciest architectures.

So which model won? The GCN-LSTM hybrid achieved the best pooled continuous performance, with a root mean square error of 35.84 and a coefficient of determination of 0.463 across the full test set. But here is the uncomfortable part: the persistence baseline, which requires no training at all and simply carries today’s value forward, came second, with an RMSE of 37.50 and an R-squared of 0.412. The ST-GCN followed closely at 36.18 and 0.452. The margin separating the most sophisticated graph neural network from a naive forecast of no change was slim, and bootstrap analysis, which resamples the test data to quantify uncertainty, indicated that the gains over persistence were rather slight. In other words, much of the apparent advantage of deep learning may fall within the range of statistical noise.

The authors did not stop at a single headline metric. They probed the robustness of their conclusions from several angles. Rolling-origin evaluation, in which models are repeatedly retrained and tested on successive time windows, assessed how performance holds up as the forecast period moves forward in time. Multi-seed experiments, in which each neural network is trained from several different random initializations, tested whether reported results depend on a lucky starting point. Matched-feature studies varied the input configuration and revealed that model rankings are sensitive to which predictors are available, a finding that complicates comparisons across studies that use different input sets. Perhaps most striking was the graph-structure ablation: replacing the distance-based connection scheme between stations with a trivial identity adjacency, in which each station is effectively isolated, produced no discernible difference in performance.

That last result deserves particular attention because it strikes at the core assumption of spatio-temporal graph models. These architectures are built on the premise that explicitly encoding spatial relationships, such as the physical distance between monitoring stations, allows information to flow between sensors and improves prediction. If an identity adjacency performs just as well on this network of eight Ahmedabad stations, it suggests that either the stations are not close enough for inter-station dynamics to matter at a twenty-four-hour horizon, or that the temporal signal at each station already contains whatever spatial information is useful. For a city-scale deployment, this raises a practical question about whether the added complexity of graph machinery is justified at all.

The picture grew darker when the researchers examined performance by AQI category. Rare high-pollution classes, precisely the episodes that matter most for public health, were predicted worse than common ones, and no model managed to correctly forecast the Severe category even once. This failure mode is well known in imbalanced classification but is often obscured by pooled metrics, which are dominated by the frequent moderate categories. A model can post a respectable RMSE while being systematically blind to the pollution episodes that send people to emergency rooms. The authors’ class-wise evaluation makes this blind spot explicit, and it is a practice that future air quality forecasting studies would do well to adopt as standard.

Why does this matter beyond Ahmedabad? Air pollution is linked to cardiovascular mortality, respiratory disease, and a wide range of economic and social costs, and roughly half the world’s population is exposed to increasing pollution levels according to global assessments. Cities across South Asia and beyond are investing in machine learning pipelines for next-day air quality advisories, often guided by literature that reports impressive accuracy figures without persistence baselines, leakage-free preprocessing, or uncertainty quantification. The Ahmedabad benchmark demonstrates that when those methodological safeguards are in place, the headline numbers shrink dramatically. An R-squared of 0.463 for the best model means that less than half the variance in next-day AQI is explained, leaving substantial room for meteorology, emissions variability, and atmospheric chemistry that the models do not capture.

The study’s deeper lesson is about scientific hygiene rather than any particular architecture. Solid baselines, matched comparisons, leakage-free pipelines, bootstrap uncertainty, rolling-origin validation, and multi-seed stability checks are not bureaucratic overhead; they are the difference between a result that generalizes and one that evaporates on contact with reality. The authors’ finding that GCN-LSTM edges out persistence, but only barely, and that graph structure itself appears dispensable, should prompt the air quality forecasting community to re-examine published claims made without such controls. For cities deciding where to spend limited monitoring and computing budgets, the message is pragmatic: before deploying a graph neural network, check whether simply assuming tomorrow will look like today, augmented with modest statistical modeling, gets you most of the way there. In Ahmedabad, at least, the answer appears to be yes.

Subject of Research: Comparative benchmarking of machine learning and spatio-temporal graph neural network models for 24-hour-ahead air quality index prediction at multi-station monitoring networks

Article Title: Comparative analysis of machine learning and spatio-temporal graph models for multi-station air quality index prediction

Article References: Yagnik, H., Gupta, R., & Shah, K. (2026). Comparative analysis of machine learning and spatio-temporal graph models for multi-station air quality index prediction. Theoretical and Applied Climatology, 157(11), Article 706. https://doi.org/10.1007/s00704-026-06623-y

Image Credits: AI Generated

DOI: 10.1007/s00704-026-06623-y

Keywords: air quality index, AQI forecasting, graph neural networks, GCN-LSTM, ST-GCN, persistence baseline, machine learning, Ahmedabad, CPCB, data leakage, bootstrap analysis, spatio-temporal modeling

The scale of transformation in Ahmedabad is striking. Multi-decadal analyses of land use and land cover report that the city’s built-up area expanded by more than 130 percent between 1990 and 2019, largely at the expense of farmland, peri-urban vegetation, and open ground. Long-term assessments stretching back to 1976 confirm a pattern of continuous peripheral growth, in which rural and semi-urban zones have been progressively absorbed into the metropolitan fabric. What was once a compact historic core has evolved into an increasingly dispersed and heterogeneous urban system, pushed outward in every direction beyond the Sabarmati River by industrial corridors, major road networks, and large-scale residential development. Manufacturing hubs, commercial centers, and institutional zones have accelerated land conversion across the vast peri-urban belt surrounding the city.

Jaipur’s story is different in shape but similar in substance. Studies of the Rajasthan capital document significant and steady growth in built-up land since the mid-1990s, consumed mainly at the cost of agricultural land, scrubland, and open spaces. Unlike Ahmedabad’s radial spread, Jaipur’s expansion has been strongly directional, concentrated in the western, southern, and northwestern sectors of the city. The review attributes this asymmetry to the development of major roads, industrial estates, institutional nodes, and large residential colonies in those directions, combined with the availability of flat, easily developable land and improving connectivity. Vegetation cover in Jaipur has shown a consistent declining trend across the periods reported in the literature. Although planned green areas such as parks and landscaped localities have increased slightly, these gains are confined to small geographic pockets and do not compensate for the loss of natural and agricultural cover, with negative consequences for the city’s ecological balance.

The thermal consequences of these land conversions are well documented in both cities. Satellite-based studies consistently find a positive correlation between the growth of built-up area and rising land surface temperature, or LST, across multiple temporal scales. The mechanism is straightforward physics. When permeable, vegetated ground is replaced with impervious materials such as concrete and asphalt, the surface absorbs more solar radiation and retains more heat. Urban surfaces typically have lower albedo than natural vegetation, meaning they reflect less sunlight and absorb more energy. At the same time, the loss of vegetation and open land reduces evapotranspiration, the process by which plants release moisture and cool their surroundings. Dense building clusters, narrow streets, and high-rise construction further obstruct airflow and natural ventilation, trapping heat within street canyons and degrading the city’s capacity to shed warmth after dark.

The review distinguishes carefully between related thermal phenomena. The urban heat island, or UHI, describes the situation in which air temperatures in an urban center exceed those of surrounding rural areas because of urbanization, vegetation loss, and waste heat from human activity. The surface urban heat island, or SUHI, is its surface manifestation, measured through satellite-derived estimates of land surface temperature rather than thermometer readings. In Ahmedabad, the interplay between these phenomena proves surprisingly complex. Surface moisture availability and the agricultural cropping cycle strongly influence SUHI intensity. During the pre-monsoon season, when soils are dry and vegetation cover is sparse, rural areas around the city can become extremely hot, sometimes producing lower or even negative daytime SUHI intensity, meaning the city surface is not dramatically hotter than its parched surroundings. Nighttime SUHI intensity, however, remains consistently positive, because urban materials store heat during the day and release it slowly after sunset. Jaipur, by contrast, shows an almost constant rise in land surface temperature driven by the steady expansion of urban surfaces.

Methodologically, the two cities have followed markedly different research trajectories, and this asymmetry is one of the review’s central findings. Ahmedabad has emerged as a methodological benchmark for predictive urban growth modeling in India. Its studies routinely integrate cellular automata–Markov chain models, artificial neural networks, transition probability matrices, and validation statistics such as kappa coefficients, weaving in population data, road proximity, and industrial growth indicators to simulate future expansion scenarios. This modeling sophistication gives planners a genuine capacity to evaluate alternative futures. Jaipur’s literature, in contrast, remains largely retrospective, relying on conventional supervised classification and change detection to document what has already happened, with few predictive modeling studies. Both cities exhibit comparable trends of vegetation loss and rising surface temperature, but the strength of evidence available to support forward-looking urban planning is considerably greater for Ahmedabad.

The review itself was conducted with unusual rigor for a narrative synthesis. The author searched Scopus, Web of Science, Google Scholar, and ScienceDirect from June to December 2025, using combinations of keywords covering land use and land cover, urban and surface heat islands, land surface temperature, remote sensing and GIS, urban growth modeling, and World Heritage City urbanization. The initial search returned 90 articles, which were screened down to 51 full texts and finally to 39 studies included in the qualitative comparative analysis, following a PRISMA-inspired framework to ensure transparency and reproducibility. Because the underlying studies used different satellite sensors, spatial resolutions, classification algorithms, and validation procedures, the review deliberately avoided direct numerical comparison, instead synthesizing evidence across common themes including growth pattern, thermal response, modeling sophistication, and planning relevance.

The stakes extend well beyond thermal maps. Rising urban temperatures elevate the risk of heat illness, dehydration, respiratory disease, and cardiovascular stress, with outdoor laborers, elderly residents, and economically disadvantaged communities facing the greatest exposure. Heat stress is also linked to declining labor productivity in construction, transport, and informal work sectors that dominate rapidly growing semi-arid cities. Climate change compounds these pressures by intensifying heatwaves and droughts and by reducing water availability, while vegetation loss, shrinking water bodies, and groundwater depletion erode urban ecological resilience. Recent Ahmedabad studies have begun integrating land surface temperature, evapotranspiration, vegetation indices, and groundwater parameters into compound analyses of urban growth and climate variation, though comparable integrated research in Jaipur remains scarce. For cities whose historic urban fabric, from Jaipur’s planned eighteenth-century grid to Ahmedabad’s walled old city, is the very asset that earned global recognition, unmanaged heat and sprawl threaten both environmental sustainability and heritage conservation.

The review identifies clear research gaps and a path forward. Most existing work concentrates on biophysical change detected through remote sensing, while socioeconomic, institutional, and governance drivers, including census-based demographics, migration, and land-use policy, remain underexplored in both cities. Peri-urban transition zones, where land conversion is fastest and least controlled, deserve far more detailed spatial assessment. The author calls for integrating land change analysis with urban climate modeling, heat stress mapping, and land-atmosphere interaction studies, and for greater use of machine learning techniques such as random forest, support vector machines, deep learning, and hybrid cellular automata approaches. Heat vulnerability analysis, combining thermal exposure with population density, public health indicators, and access to green space, is flagged as essential for identifying at-risk communities. On the policy side, the prescriptions are concrete: expand green infrastructure, protect urban ventilation corridors, promote permeable surfaces and nature-based solutions, and embed heritage-sensitive climate adaptation into planning. As the review concludes, the future sustainability of India’s semi-arid World Heritage Cities depends on treating land use, urban climate, and cultural heritage as a single, inseparable planning problem rather than three separate bureaucracies.

Subject of Research: Land use transformation and urban heat island dynamics in the UNESCO World Heritage Cities of Jaipur and Ahmedabad, India

Article Title: Land use transformation and urban heat dynamics in India’s world heritage cities Jaipur and Ahmedabad

Article References: Land use transformation and urban heat dynamics in India’s world heritage cities Jaipur and Ahmedabad. (n.d.). https://doi.org/10.1007/s44327-026-00361-6

Image Credits: AI Generated

DOI: 10.1007/s44327-026-00361-6

Keywords: land use land cover, urban heat island, land surface temperature, Jaipur, Ahmedabad, remote sensing, GIS, urbanization, World Heritage Cities, CA-Markov, surface urban heat island, semi-arid cities