How can the floor area types of a university campus mitigate the increase of urban air temperature?

The study area is located at the western Insubria region from Lago di Varese to Lago di Como. It ranges from 45.75°N 8.85°E to 45.82°N 9.05°E. Insubria has long been recognized as a distinct bioclimatic region, characterized by its mild climate [35]. This area is characterized by a sub-continental climate, with a mean annual precipitation of 1600 mm, occurring in two main periods (April–May and October–November) and a mean annual temperature of 10–11 °C [36]. There is a negative W-E precipitation gradient that affects the seasonal pattern of precipitation with summer rains usually in the West (Lago Maggiore) and summer drought in the East [37].

The Insubria university area is divided into 3 zones/cities, each of which resembles several university campuses: Bizzozzero and Ravasi for Varese, Manara for Busto Arsizio and Valleggio, Abbondio, Natta, Cavallotti and Oriani for Como (Fig. 1). Varese has 79,800 inhabitants and a population density of 1450 inhabitants per square km, Como has 84,700 inhabitants and a population density of 2280 inhabitants per square km, and Busto Arsizio has 84,000 inhabitants and a population density of 2700 inhabitants per square km.

To describe the weather conditions of the sites for all 2020, 3 automatic weather stations (AWSs) were selected to be the closest to the three cities. The three AWSs’ data were downloaded from the weather monitoring network of Lombardy Region (www.​arpalombardia.​it): (1) Busto Arsizio (45.6106°N, 8.8502°E, 222 m a.s.l.), (2) Varese (45.8325°N, 8.8236°E, 416 m a.s.l.), and (3) Como (45.8155°N, 9.0672°E, 201 m a.s.l.). The 3 AWSs were selected also to represent different urban settings in their neighborhood: more anthropized surfaces than green for Busto Arsizio, and half green and half anthropized for Como and Varese. In addition, a fourth AWS (Parco Pineta, PP) was chosen to represent the temperature difference between ground (at 10 cm of depth, under an herbaceous vegetated surface) and air (Centro Geofisico Prealpino, https://​www.​astrogeo.​va.​it/​meteo/​). The daily temperature differences were averaged with Varese AWS (which lays on an herbaceous mat) to have a general understanding of the annual heat exchange of the whole area (Fig. 2).

The fieldwork period, at the end of summer 2020, spanned from 28/09/2020 to 11/10/2020. This period was representative of the annual minimum variation of heat transfer between the surface and the air. Indeed, the annual standard deviation of air–ground daily temperature differences corresponds to 2.37 °C, whereas, during this study period, it is only 1.02 °C (Fig. 2a). Conversely, during summer and winter, the heat transfer can be amplified (Zajch and Gough 2021), with the occurrence of summer heat waves, soil moisture deficit (Whan et al. 2015), and high frequency of extreme events (Christidis et al. 2015). Moreover, photosynthesis and evapotranspiration can be reduced in summer due to stomatal closure (Zeiger et al. 1987). At these latitudes during spring, the frequent precipitation (Isotta et al. 2014) produced by the latitudinal oscillation of the North Atlantic Polar Front (Pinna 1996) is the main source of ground moisture which could alter the cooling effect of the green surfaces. The plant species and mean vegetation height (cm) for each type of lawn have been observed and measured in situ to assess the differences among the lawn types.

The choice of the surface types was taken by considering all the outdoor floor areas of the University Campus. In this sense, buildings were excluded from the study because vertical walls do not receive as much solar radiation as flat surfaces and roofs are difficult to access. Moreover, cool roofs are usually considered for the energy saving of buildings and not for outdoor floor areas that directly affect the human perception of the UHI. In addition, trees that cover a smaller total surface (21.5%) than grass lawns (78.5%) (example taken from Dunant, that is the zone with the highest number of trees) were also excluded. Each surface type was monitored once at the same period of the day (between 11:00 and 15:00 local time) to guarantee the zenith angle of the sun, and therefore the warmest moment of the day. Moreover, since it was not possible to investigate all the surfaces at once, additional monitoring days were selected on the basis of complete clear sky, no wind and after at least 2 consecutive dry days, therefore avoiding the precipitation occurring between 02/10/2020 and 05/10/2020 (Fig. 2b). A total of 87 measurements were conducted to guarantee at least 3 replicates for each surface type.

Surface temperatures were recorded by taking 3 randomly distributed images per surface type with a thermal camera (FLIR E85, 384 × 288 pixels, 0.1 °C of resolution, 2.0 °C of accuracy), calibrated in situ (Shea and Jamieson 2011). The temperature minimum, maximum, and mean value were recorded for each image.

The material mean WC (%) was recorded through 3 randomly distributed measurements per surface using a handheld dielectric moisture meter (Voltcraft MF-100) with a resolution and an accuracy of 0.1%. This is an easy tool to assess the construction materials moisture (Eklund et al. 2013). It recorded the mean WC within a range of depth of 2–4 cm.

The albedo was calculated as the ratio of the reflected radiation of the incoming radiation on a surface with a portable pyranometer (DeltaOhm LP 9221 S2, spectral range of 450–950 nm, resolution of 0.1 W m⁻², accuracy of 1.0 W m⁻²). It was oriented perpendicularly to the surface at 50 cm above the surface. It was first directed to the surface, then directed in the opposite direction, 3 randomly distributed times per surface type.

Cost of the materials/surfaces has been expressed as the range of costs found on a) the Prezziario regionale delle opere pubbliche—2021 (www.​regione.​lombardia.​it), and b) Assoverde price list 2019–2021 (www.​assoverde.​it). They are comprehensive costs that include materials and/or the maintenance costs. It should be noted that the costs (€) per square meter refer to the entire amount. In this study, since it is not possible to forecast the descending price auction of the contractor company, we preferred to use the whole amount while keeping in mind that the real costs must be lower.

Social data were not considered in this analysis because the age of the University’s frequenters is not representative of the population of the 3 cities and, in turn, it would not resemble the real socio-economical habits of the citizens. However, the results will be the base for a future multidisciplinary research.

The data analysis aims to present the recorded data in a simple output that represent the potential of mitigation of each surface type. It is for this reason that, for UHI studies, it is the surface heat budget that governs the temperature perception, not only the surface temperature itself, as demonstrated by Eq. 1 (Takebayashi and Moriyama 2009):where R_n is the net radiation (W m⁻²), H the sensible heat flux (W m⁻²), G the conduction heat flux (W m⁻²), and lE the latent heat flux (W m⁻²). It has to be kept in mind that, at stable conditions (no wind, same material, constant ambient temperature), albedo is the discriminating characteristic of the absorption of shortwave radiation, surface temperature is the discriminating characteristic of H while surface moisture of lE (Takebayashi and Moriyama 2009). Due to this, we suggest to use these properties for the construction of a simplified model that resembles the capability of a surface to increase or decrease the near surface air temperature. Indeed, when a surface or pavement is light colored (high albedo), retains more water (high water content) and has a high emissivity (low surface temperature) is usually cooler that the others (Santamouris 2014). Therefore, we propose here an equation based on the three simple measurable parameters discriminating the different energy balance components (albedo, water content, and surface temperature) for the air temperature mitigation index (ATMI) of a surface:where WC is the volumetric water content of the material (%), A is the albedo of the surface (ranging between 0 and 1), and T is the mean surface temperature (°C). The formula can be unitless and represents the easiest and fastest way to assess the energy budget of a surface. Indeed, the parameters are easily measurable with portable instruments and no samples for laboratory tests (thermal conductivity, density, heat capacity) are needed. This method guarantees a quick assessment for spatialization analysis.

The single recorded data were inserted into the equation as spatial mean values and a standard deviation of each parameter was also provided. Moreover, the maximum ATMI of the surface types was calculated by substituting the mean surface temperature with the minimum surface temperature.

The precision and accuracy of the datasets would lay on the repeatability of the measurement and the instrumental error. However, even at stable atmospheric conditions, the positioning of the instrument could differ from one measure to the second and affect the final data. It is for this reason that the equipment used was either already chosen/calibrated in previous publications (Schnepfleitner et al. 2016; Ponti et al. 2021b), or tested in the field such as the portable pyranometer with a white paper of known albedo. Moreover, to reduce errors, the operator and his position with the instrument was kept constant for each measure and only the average value was used for the ATMI calculation.

Since the surface temperature is a consequence of the surface energy balance, a generalized linear model (GLM) with a backward stepwise method was conducted to test which of the measured parameters (air temperature, albedo, air relative humidity, and material type) majorly affected the surface temperature. Moreover, a multiple analysis of variance (MANOVA) and a one-way ANOVA were run to highlight the variance differences of ATMI and the measured parameters for each material type. All the above-mentioned analyses were performed using the software STATISTICA^®.

The spatial analysis was conducted in ArcGIS 10.8 (ESRI, Inc., Redlands, CA, USA): each surface type was digitalized throughout the university zones to obtain their areal extension (m²). This permitted us to calculate a weighted average of the ATMI per university area, taking the sum of the product of the weights (surface types areas) times the ATMI value correspondent to each surface type divided by the sum of the areas.

A total of 26 surface types were identified and analyzed. Among the natural surfaces, 6 categories of green surfaces were identified, not only depending on the type of management, but also on the floristic composition and average vegetation height. The list of species differs among the lawn types; however, some of them (the most abundant) are widespread (Poa pratensis, Trifolium pratense). The lawns that usually undergo frequent mowings had an average height lower (15 cm) than those that undergo rare mowings (40 cm) (Table 1).

Overall, the surface temperatures were always greater than the air temperature recorded at the reference AWS, demonstrating that a sensible heat transfer toward the atmosphere was occurring, as assumed, except for the shrub lawn. This latter showed a mean surface temperature equal to the air temperature due to the shadow effect of the canopy. The highest temperature difference was measured at the interlocking block pavement made of vibrocompressed concrete2 (17.9 °C) and at the holed interlocking pavement made of concrete and gravel as filling (17.6 °C). Apart from the shrub lawn, the minimum temperature difference was recorded at the porphyry pavement made of pink porphyry small blocks (0.7 °C) and at the 4-mowing lawns (1.4 °C). The mean WC varied considerably among the materials with values between 33.8% (asphalt) and 97.3% (3-mowing lawn). Smaller variations were recorded at surface types with the same material but different structures or dimensions such as the vibrocompressed concrete that ranged between 57.0% (type 2) and 64.0% (type 3). Generally smaller variations were obtained for the albedo that showed a minimum of 0.14 for the asphalt and a maximum of 0.27 for the shrub lawn and the white fine gravel (Table 2).

Figure 3 illustrates an example among the different university examined areas. We selected this example because it is the largest examined area and because here, there is a prevalence of green areas in which the different managements led to great ATMI differences. Moreover, the distribution of the green areas is heterogeneous and respects the suggestions of Kong et al. (2014) for the mitigation property. Among them, the 4-mowing lawns are the best mitigating surfaces (ATMI 0.96), followed by other lawns in general (ATMI 0.52–0.84). The worst mitigating property is shown for asphalt paths (ATMI 0.15) and the holed interlocking pavement made of concrete and gravel (ATMI 0.20). The same trend of ATMI illustrated in Fig. 3b was found also at the other university areas.

The surface types also showed large differences of the ATMI in terms of means and ranges (Table 3). Among all, the 4-mowing lawn had the greatest mean ATMI (0.96), followed by the shrub lawn (0.86) and the 3-mowing lawn (0.84). Asphalt had the lowest mean ATMI (0.15), followed by concrete and gravel (0.20) and the pink asphalt (0.27). However, the range distribution did not perfectly match the mean ATMIs. Indeed, except for the 4-mowing lawn that had the largest range (0.87), the mixed gravel had the second largest range (0.59), followed by the 3-mowing lawn (0.51). The smallest ranges, instead, were recorded at the concrete and gravel (0.12) and the vibrocompressed concrete of type 1 and 2 (0.12). By grouping for the surface type category, the green areas had a mean ATMI of 0.69 and a mean ATMI range of 0.51, while the anthropized surfaces had a mean ATMI of 0.42 and a mean ATMI range of 0.24.

Within the green areas, the 4-mowing lawn was the best in terms of mean ATMI (0.96) and range (0.87), while hedge2 had the lowest mean ATMI (0.39) and the 8-mowing lawn the smallest range (0.31). Within the anthropized surfaces, concrete1 and vibrocompressed concrete3 had the highest ATMI (0.63), while asphalt (0.15) and concrete and gravel (0.20) the lowest. The smallest range was recorded for the latter and the vibrocompressed concrete of type 1 and 2 (0.12) (Table 3).

The material surface temperature can be considered the result of the energy balance that is governed among all, not only by the material albedo and water content, but also by the air temperature and relative humidity (RH). The surface temperature, in turn, triggers a heat flux exchange able to increase or decrease the air temperature. It is for this reason that it is fundamental to understand what variable affects the material surface temperature. The GLM (R² = 0.78, p < 0.001) showed that the surface type is the only factor that statistically varies with the surface temperature variations (F = 8.64, p < 0.001) (Table 4). In a way, the surface type represents all of the material properties that drive the final surface temperature (even more than the external environmental conditions).

Since the surface temperature is driven by the interactions of the material properties (among which the albedo and WC), it is important to highlight the variability of the ATMI within the same materials (10 classes of surface types) to understand where interventions on the material characteristics are feasible to improve the ATMI. The MANOVA analysis showed a statistical difference among the 3 ATMI’s components (F = 5.69, p < 0.001, 27 degrees of freedom). This demonstrates that the ATMI is amendable by acting on diverse measured properties, depending on the material type. Moreover, the one-way ANOVA showed that within the same component, there is a statistical difference of variance depending on the material type (Fig. 4).

The largest variability of ATMI was recorded for lawns (1.1), while the smallest for pebbles (0.07). Similarly, the largest variability of surface temperature was recorded for lawns (21.0 °C), while the smallest for pebbles (1.0 °C). Differently, hedges had the greatest WC variability (56%), while pebbles the lowest (10%). Again, lawns showed the greatest variability of albedo (0.34), while dirt road the smallest (0.01) (Fig. 4). All the highest ATMI (favored) surface types per type of material are presented in Table 5.

Among all of the university zones, the largest surface corresponds to the 3-mowing lawns (72,445.1 m²), followed by the 8-mowing lawns (44,162.5 m²), while the most restricted surfaces are hedge2 (151.5 m²) and pebbles (167.4 m²). The weighted average of the ATMI according to its areal extension is Valleggio the best mitigating university zone (0.69), followed by Bizzozzero (0.67). Conversely, the worst mitigating areas are Oriani (0.30) and Ravasi (0.42). Overall, the whole university area accounted for an ATMI of 0.66. Maintenance and posing costs of such surfaces vary between 0.15 (lawn mowing) and 97.21 € m⁻² (granite slab). However, for equal extension, it is important to note that surfaces with reduced maintenance costs could be more expensive than surfaces with cheaper posing costs, depending on the number of mowings or prunings. In this study, the total annual maintenance costs for the university green areas range between 69,060 and 82,250 €, while the total posing costs range between 2,897,426 and 3,616,984 €. This suggest that currently the active costs for only the maintenance of green areas are 2.3% of the posing costs sustained. Therefore, green areas will overcome the posing costs sustained with the current management in 50 years or more (Table 6).

For green areas like lawns that obviously had the highest WC, albedo was the variable with the greatest variation able to affect ATMI. Indeed, depending on the vegetation species, hairy shrubs on lawns considerably increased the albedo and, in turn, the ATMI. Similarly, taller herbaceous species (as consequence of less frequent mowings) had a higher albedo than shorter vegetation due to the quantity of green surface. For hedges, of which the WC had the largest variation, the canopy structure or the leaf area index (LAI) could have influenced the air humidity within the canopy (Hardwick et al. 2015), and thus could have affected the ATMI.

Other studies showed that the irrigation of anthropized surfaces highly mitigate extreme temperatures (Yamagata et al. 2008), especially if the WC is retained for a longer time (Nakayama and Fujita 2010). The fact that lawns showed such high WC values (71.3–97.3%) indicates not only the presence of moisture within the soil matrix but also within the organic matter tissues such as litter or roots that convey moisture to the surface and increase evaporation (Qin 2015). In a way, the presence of roots acts as a water reservoir so that the vegetated surfaces permit a twofold cooling process via evaporation of soil moisture and leaves’ evapotranspiration (Tan et al. 2015).

For vegetated surfaces like lawns, the correct management is the reduction of annual mowings (Townsend-Small and Czimczik 2010; Gu et al. 2015) that increase the photosynthetic total surface, thereby augmenting evapotranspiration and ground shading (Tan et al. 2015; Mariani et al. 2016). Shading properties of vegetation against the UHI effect have been demonstrated (Tan et al. 2015, 2021; Park et al. 2022) and confirmed here in relation to taller vegetation. However, the fact that we recorded a higher ATMI for 4-mowing lawns than 3-mowing lawns is related to the difference of surface temperatures. At the time of measurement, the 3-mowing lawns underwent a mowing not being representative. Therefore, the vegetation height of the 3-mowing lawns produced less shadows, thus higher temperatures (Mariani et al. 2016). For hedges, it is recommended to choose non-compacted canopies, not only to increase the shading properties (Tan et al. 2015), but also to increase the air RH.

For our point of view, to extend the correct practices in urban contexts, it is necessary to not only rely on the good appearance of the green areas, neither on total abandonment, but rather a compromise of a functional vegetation-soil system (high shading, LAI, water retention) that needs little annual management.

Generally mitigation strategies on anthropized surfaces focus on the increase of reflectance or the water retention, that is affected by the texture of the material (Santamouris 2013). In this study, anthropized surfaces’ ATMIs are mostly ameliorable by increasing the WC of the material. The surface temperature was only more variable than WC and albedo for porphyry and vibrocompressed concrete, probably due to the increased heat flux exchange through the different type of block inter-connections. This is similar to the different characteristics of seam-fillings (Starke et al. 2011). Surfaces with the lowest WC values like asphalt (33.8%) or pink asphalt (39%) are impervious materials that are the worst choice for the temperature mitigation (Santamouris 2013; Bozdogan Sert et al. 2021). Indeed, not only is such material porosity very low (Nakayama and Fujita 2010), but it is also a homogeneous surface with no interruptions like block grouts that, in the areal complexity, can be more impervious than granite blocks. Therefore, the dimensions of the concrete blocks increase the number of grout as connections with deeper and moister layers (Starke et al. 2011; Qin 2015). We, therefore, suggest that the porosity of the anthropized materials is the only factor responsible for ATMI variability.

Other types of materials displayed low temperature ranges and low ATMI ranges such as asphalts, vibrocompressed concretes 1 and 2, and the holed interlocking pavement. Asphalt is a homogeneous impervious surface, while blocks of vibrocompressed concrete2 showed higher mean temperatures than type 1 and 3 at similar WC because of the reduced presence of grouts (larger dimensions of blocks) that could increase the total surface water retention capacity, namely evaporation (Collins et al. 2010; Starke et al. 2011; Qin 2015). Conversely, the higher permeability of the holed interlocking pavement resulted in low WC due to its low water retention capacity (Qin 2015) (which results in less evaporation).

By grouping the surface types by materials, we can highlight their variability of ATMI and speculate on the proper management/choice for the air temperature warming mitigation. For concrete, it is important to provide as many varieties of grout as possible by reducing the slab/block dimensions (Fig. 5c,d), and to use light colored or pervious concrete (Takebayashi and Moriyama 2012; Qin 2015). Others demonstrated the utility of using reflective concrete, for example adding white components to the mixture (Levinson and Akbari 2002) or increasing its permeability (Matsuo et al. 2005). For gravel, it is important to include different grain sizes to find an equilibrium between drainage and water retention (Qin 2015). This balance is likely obtained with the addition of a soil matrix or mixed grain sizes (Fig. 5e,f), and by utilizing light-colored rocks. It was observed that the age of pink vibrocompressed concrete (Fig. 5a,b) slightly affects the deterioration of material properties by worsening the surface temperature like with thermochromic coatings (Santamouris 2013). Therefore, a cleaning of the surface should be considered both for maintaining a high albedo and an effective infiltration (Mullaney and Lucke 2014). Regarding vibrocompressed concrete and porphyry blocks, similarly to concrete slabs, it is recommended to use small blocks to guarantee an increase of grout (Collins et al. 2010; Mullaney and Lucke 2014), rather than the change of the material color that would change their albedo. Similarly, among asphalts, pink asphalt seemed to be the best solution; not due to its color (albedo), but rather due to its different porosity that led to different WC.

Overall, asphalt is the worst material for the mitigation (Mullaney and Lucke 2014). However, some improvements of the mitigation characteristics of the asphalt can be guaranteed by a better circulation of water within the asphalt grains (De Bondt and Jansen 2006) or the use of light-colored aggregates (Anak Guntor et al. 2014).

After these considerations, we would suggest the choice of small pavements’ units, overlaid to properly selected grain sizes that favor the retention of water, able to percolate from the high number of grouts (Mullaney and Lucke 2014). This is a solution that is accessible in multiple urban contexts.

The areal extension of the University campus sectors shows that the Valleggio and Bizzozzero sectors contributed to the highest ATMI. This is due to the large vegetated area extension in Bizzozzero and Valleggio, even though some of these areas have recently been converted to parking areas in Valleggio. Conversely, the absence of consistent vegetated areas in Oriani and the large asphalt area in Ravasi made them the worst University zones for climate change mitigation. This is in line with other authors that found it is extremely important to provide green areas in urban landscapes (Takebayashi and Moriyama 2009; Lee et al. 2010; Kong et al. 2014; Mariani et al. 2016; Yao et al. 2019; Tan et al. 2021), especially if they are numerous, large, or fragmented (Kong et al. 2014; Chen and You 2020). This management also helps the mitigation of air temperatures of surrounding surfaces (Kong et al. 2014).

Few works listed the prices of materials for urban building and green planning, probably because the costs are very susceptible to the local economics, and thus highly variable. Lee et al. (2010) listed some general costs for American surfaces that is quite dated. Also Bretz et al. (1998) listed prices for higher albedo materials despite them being dated and focused on the light additives of roofs (which is not the object of this study). Interestingly, Pomerantz (2018) claimed that there is very little monetary energy savings from high albedo materials. The current literature shows no posing costs or green maintenance costs in Italy that include the benefits of natural and anthropized surfaces in urban areas.

In this research, the overall maintenance costs of the green surfaces account for ca. 69,060–82,250 € per year which is 2.3% of the posing costs of the anthropized surfaces. It is known that green surfaces are the best mitigation choice in every case; however, we found other synthetic surface types with interesting ATMIs: concrete1, granite block, pink porphyry small block, vibrocompressed concrete3, and white fine gravel. For example, if we take the average cost of the best mitigating materials (concrete1: 28.9 € m⁻² and vibrocompressed concrete3: 30.1 € m⁻²) and substitute the 3 to 8-mowing lawns with these materials, we would obtain a total cost of 11,986,177 € which is equal to ca. 180–197 years of lawn management. This suggests that it is a cheaper solution of management technique, but it is up to the University to decide the future territory planning. On one hand, certain anthropized material (cool pavements) guarantees no future management, but on the other hand, green management is less expensive but requires maintenance (Takebayashi and Moriyama 2009). Probably, the most reasonable choice would be to use concrete in small blocks and as much white as possible (Lee et al. 2010) to maintain the same cost. These could be used for parking areas, pathways or any trampled surface. The use of lawns would also be suitable for recreative or decorative large areas.

Both for green areas and anthropized surfaces, we propose feasible and common solutions to be spread to any urban context, from which local institutions could sustainably save funds. The reduction of green maintenance with suitable vegetation can decrease the annual costs and the choice of smaller pavement’s units does not affect the posing costs much more than the choice of another material type.