Tiny Urban Voids Could Cool Historic Cities by Several Degrees, Study Finds

In the dense, sun-baked historic centers of Mediterranean cities, planting a new park is often physically impossible. Streets are narrow, buildings are packed shoulder to shoulder, and nearly every square meter of open ground is sealed beneath asphalt, basalt paving, or concrete. Yet it is precisely these constrained fabrics that suffer most from the urban heat island effect, where densely mineralized areas can exceed surrounding rural temperatures by more than 8 to 10 degrees Celsius. A new study published in Environmental and Sustainability Indicators proposes a way around this impasse: instead of chasing large green spaces that cannot be built, researchers have developed a protocol that hunts for tiny, overlooked pockets of sealed land and converts them into micro green units capable of measurably lowering local temperatures.

The research, led by Floralba Pirracchio Massimino with colleagues Rui Alexandre Castanho, Inmaculada Gómez, and Javier Velázquez, tested its Sustainable Greening Upgrades protocol in Catania, a Sicilian city of roughly 296,500 inhabitants whose historic core was largely rebuilt after the 1693 earthquake in Baroque style and now forms part of the UNESCO-listed Late Baroque Towns of the Val di Noto. The team focused on a 2.22 square kilometer sample area within the historic center, a fabric they characterize as compact mid-rise: continuous building blocks, medium heights, extensive impervious surfaces, and very limited vegetation. The area houses about 18,432 residents and has a built-coverage ratio of 0.68, meaning more than two thirds of the ground is occupied by buildings, with most of the rest sealed in one way or another.

The urgency behind the work is stark. The June 2025 heatwaves tripled heat-related deaths in Europe, and Mediterranean cities are warming faster than many other European regions. Previous studies of Catania had already shown that land cover is the main driver of surface temperature differences observed by satellite, with gaps of 15 to 17 degrees Celsius between green and built areas both day and night. In Thessaloniki, the historical center records mean land surface temperatures of about 29 degrees with peaks above 31 degrees due to canyon effects, and similar constraints have been documented in Florence. What has been missing, the authors argue, is a practical bridge between diagnosis and design: maps of urban heat are abundant, but concrete, georeferenced micro-sites ready for intervention are not.

The protocol’s methodological core is a five-stage decision chain built entirely from open or routinely available spatial data. First, the study area is delimited morphologically. Second, satellite imagery is preprocessed to produce diagnostic layers of land surface temperature and vegetation. Third, deficits in urban green space are mapped using a 200-meter service-area buffer around existing parks larger than 500 square meters. Fourth, candidate urban voids are delineated and prioritized. Fifth, the hottest and most transformable sites are evaluated in a microclimate simulation. The team retrieved land surface temperature from Landsat 8 thermal imagery acquired on 15 July 2025, applying radiometric conversion, brightness-temperature calculation, and NDVI-based emissivity correction, then disaggregated the 30-meter thermal data to 10 meters using Random Forest regression with Sentinel-2-derived predictors, achieving an internal model fit of R-squared 0.91 with a root-mean-square error of 1.8 degrees Celsius.

The definition of an urban void is deliberately operational rather than poetic: a mineral, sealed interstitial space between 15 and 100 square meters. Candidate sites had to satisfy six screening criteria, including a Normalized Difference Vegetation Index below 0.2, a surface temperature above 45 degrees Celsius, public or semi-public ownership, physical accessibility, and a distance of more than 100 meters from an existing green space. Crucially, the geometry of these voids was not derived from the coarse satellite raster but mapped independently by vector photointerpretation of high-resolution orthophotography, with the satellite layers serving only as diagnostic covariates assigned to each polygon. This separation avoids implying sub-pixel precision that the 10-meter data cannot deliver.

Applying these criteria across the surveyed extent of the Civita and San Cristoforo sectors, roughly 0.70 square kilometers, the team identified 127 urban voids covering 6,784 square meters, or 0.98 percent of the surveyed area. The units ranged from 19 to 94 square meters with a mean of 53.4 square meters. Thermal classification placed 23 voids in the hottest priority class above 50 degrees Celsius, 59 in the 47-to-50-degree band, and 45 in the 45-to-47-degree band. Notably, the two sectors differed in thermal severity rather than in the overall density of voids: 27.5 percent of San Cristoforo’s voids reached the highest thermal class against 11.8 percent in Civita, a statistically significant difference. Sensitivity tests on the size criterion showed the 23-site priority subset was robust, remaining unchanged under most alternative size ranges tested.

To separate thermal need from practical feasibility, the researchers introduced a transformability index that scores each site for usable size, operational accessibility, ownership regime, and utility interference, then multiplies by limiting factors for pavement removability and heritage restriction. The hybrid additive-multiplicative form reflects a key reality of protected historic fabrics: a single prohibitive condition, such as listed lava-basalt paving or direct appurtenance to a monument protected under Italian heritage law, can cancel feasibility entirely and would otherwise be averaged away. The 23 priority sites all scored at or above the 0.6 admissibility threshold, with archived scores ranging from 0.60 to 0.85 and a median of 0.70. The authors are careful to stress that these are expert-assigned scores and that heritage compatibility is a second-stage constraint, not a guarantee that every candidate site is immediately implementable.

The design scenario transformed the 23 priority voids into Micro Green Units through full depaving, sustainable drainage systems with a substrate albedo of 0.25, and one six-meter Celtis australis, the Mediterranean hackberry, per unit, with a leaf area density of 1.5 square meters per cubic meter. Simulations in ENVI-met V5.1.1 at 2-meter horizontal resolution compared the sealed baseline against the greened scenario over a 24-hour period on 15 July 2025, using hourly forcing data from the ARPA Sicily Catania Center station and an initial soil wetness of 30 percent. Outputs were averaged over the 14:00 to 16:00 UTC afternoon window at 1.5 meters above ground.

The modelled results are striking. Within the 23 greened units, mean land surface temperature fell by approximately 8.3 degrees Celsius, with a peak reduction of 12.1 degrees, and nearly 87 percent of internal model pixels showed reductions greater than 5 degrees. Air temperature at pedestrian height dropped by an average of 2.1 degrees, with a maximum of 3.8 degrees, while Physiological Equivalent Temperature, a comfort metric combining temperature, humidity, radiation, and wind, decreased by 4.6 degrees on average, shifting conditions from the very hot class above 41 degrees into the hot class of 35 to 41 degrees. A modelled cooling signal also extended roughly 20 meters into the surroundings, with a mean surface-temperature reduction of about 3.2 degrees, though the authors emphasize this is descriptive rather than a quantified cooling radius. These magnitudes sit comfortably within published ranges for pocket parks and tree-lined Mediterranean courtyards, lending contextual plausibility to the scenario.

The authors are unusually candid about the limits of their findings. The cooling figures are deterministic model responses under a single meteorological day, one soil-moisture condition, and an idealized full-depaving assumption with intermediate-stage vegetation; they are not field-validated predictions, and a paired pixel-wise statistical test was demoted to descriptive status because spatial autocorrelation undermines independence. Uncertainties from the one-day offset between satellite acquisitions, the downscaling procedure, and an unexplained uniform rescaling of the scenario polygons relative to the mapped inventory are all flagged rather than hidden. Yet the study’s strongest contribution is methodological: a workflow-reproducible chain that links open-data thermal screening to independently delineated, georeferenced micro-sites, separates thermal need from transformability, and connects the selected subset to an ex-ante design scenario. As a form of green acupuncture for compact historic fabrics, the approach reframes urban greening from an aggregate deficit problem into a hunt for specific, mappable, and potentially transformable hot spots, offering heat-stressed heritage cities a realistic path to turn their smallest forgotten spaces into climate assets.

Subject of Research: Micro-scale urban greening of sealed voids to mitigate the urban heat island in the historic center of Catania, Italy

Article Title: Turning urban voids into climate assets: Micro green units for cooling historic cities

Article References: Massimino, F. P., Castanho, R. A., Gómez, I., & Velázquez, J. (2026). Turning urban voids into climate assets: Micro green units for cooling historic cities. Environmental and Sustainability Indicators, 32, Article 101545. https://doi.org/10.1016/j.indic.2026.101545

Image Credits: AI Generated

DOI: 10.1016/j.indic.2026.101545

Keywords: urban heat island, urban voids, micro green units, Catania, land surface temperature, ENVI-met, nature-based solutions, historic cities, remote sensing, Landsat 8, Sentinel-2, green infrastructure