Quantifying the Effects of Climate Change on the Urban Heat Island Intensity in Luxembourg—Sustainable Adaptation and Mitigation Strategies Through Urban Design

Abstract

Rapid urbanization and climate change intensify the urban heat island effect. This study quantifies the UHI impact in Luxembourg’s Pro-Sud region and explores sustainable mitigation strategies. In situ and mobile measurements, EURO-CORDEX regional climate projections (RCP4.5), and the FITNAH-3D urban climate model were used considering also future building developments. The results reveal a significant UHI effect, with substantial temperature and thermal stress level differences between urban and rural areas. Regional climate projections indicate a marked UHI intensification under future scenarios. FITNAH-3D simulations show increased thermal stress levels, especially in densely built areas, and highlight green infrastructure’s importance in mitigating UHI effects. Recommendations for spatial unit-specific urban climate measures specifically for vegetation, unsealing, and optimized urban design and planning are provided. Our research emphasizes the urgent need for tailored urban planning, adaptation, and mitigation strategies to enhance urban climate resilience and address thermal stress.

1. Introduction

Rapid urbanization is a defining trend of the 21st century, with populations increasingly concentrated in dense cityscapes. This transformation has a significant impact on local climate, leading to the well-documented phenomenon of higher air temperatures in cities. Balchin and Pjye [1] introduced in 1947 the term urban heat island to describe the fact of an island-like urban overheating, characterized by warmer urban areas surrounded by cooler open spaces. Compared to the surrounding countryside, cities have on average higher surface temperatures, higher air temperatures as well as higher below-ground temperatures [2]. One of the primary reasons for the higher temperatures is the change in the ground surface and building materials used in urban agglomerations [3]. A large proportion of vertical surfaces (surface enlargement) in urban areas, coupled with lower albedo (reflectivity) of these surfaces, results in increased absorption and retention of heat [4]. Moreover, the greater sealing of surfaces, e.g., due to impervious surfaces such as concrete and asphalt, reduces the natural evaporative cooling potential. The higher surface roughness and heat storage capacity of urban materials further exacerbate the problem, trapping heat within the built environment [3]. Another additional important contributor to the UHI intensity is the anthropogenic heat emission from transportation, industrial processes, and operational energy use in buildings [4,5].

According to the Intergovernmental Panel on Climate Change (IPCC), human activities, primarily through the emission of greenhouse gases, have been the principal driver of global warming, and the global surface temperature has risen by 1.1 °C above the 1850–1900 baseline during the period of 2011–2020 [6]. The year 2024 was the warmest year in global temperature records going back to 1850. According to ECMWF, the global average surface air temperature of 15.1 °C was 0.72 °C above the 1991–2020 average and 0.12 °C above 2023, the previous warmest year on record [7]. Regional climate change projections suggest an increase in the mean annual air temperature values—depending on the emission scenario—between +1.3 K (RCP26), +2.1 K (RCP45), and +3.1 K (RCP85) [8]. The UHI effect is being intensified by climate change, posing significant challenges to urban sustainability and human health [9]. As global temperatures rise, cities are experiencing increasingly hotter and partly more humid conditions compared to their surrounding rural areas. Cities in Central European urban agglomerations are experiencing the intensifying impacts of climate change [8,10,11]. These cities, often with a history of industrialization, contribute significantly to greenhouse gas emissions, while simultaneously facing increased risks from extreme weather events. Rising temperatures lead to heatwaves, impacting public health, productivity, and energy consumption [8,12,13].

The understanding of the complex dynamics and interactions among urban design, urban planning, human activities, and environmental influences that contribute to the urban heat island effect is a prerequisite for the development of robust strategies to mitigate and adapt to the challenges posed by rising urban temperatures. Mobile measurements of the UHI effect are a suitable approach for the assessment of temperature variations between urban and rural environments. In contrast to fixed stations, mobile measurements utilize vehicles or other transport methods to capture temperature data across diverse locations within urban settings. This approach enhances the spatial resolution of the air temperature information and allows for the localisation of variations in the urban temperature field influenced by factors such as landscape features, human activity, and microclimates.

Table 1. Selected studies using different kinds of mobile measurements for the assessment of the urban heat island.

Author	Year	Title	Abstract Summary
Batur, I., Markolf, S.A., Chester, M.V., et al. [14]	2022	Street-level heat and air pollution exposure informed by mobile sensing	Mobile sensors on public transportation vehicles were used to measure fine-scale urban heat and air pollution.
Bonn, B., von Schneidemesser, E., Andrich, D., et al. [15]	2016	BAERLIN2014—the influence of land surface types on and the horizontal heterogeneity of air pollutant levels in Berlin	The paper describes mobile urban heat island measurements using bicycle, van, and airborne platforms to quantify the impact of urban vegetation on air pollutant levels.
Brandsma, T., Wolters, D. [16]	2012	Measurement and Statistical Modeling of the Urban Heat Island of the City of Utrecht (the Netherlands)	The paper describes a mobile bicycle-based system for measuring urban microclimate data to study the urban heat island effect.
Brown, M.J., Ivey, A., McPherson, T.N., et al. [17]	2004	A study of the Oklahoma City urban heat island using ground measurements and remote sensing	The paper describes mobile urban heat island measurements with a vehicle in Oklahoma City and uses remote sensing to analyse temperature variations across land use types.
Chàfer, M., Tan, C.L., Cureau, R.J., et al. [18]	2022	Mobile measurements of microclimatic variables through the central area of Singapore: an analysis from the pedestrian perspective	Mobile measurements of microclimate variables in Singapore’s central area reveal the impact of urban morphology on the urban heat island effect.
Heusinkveld, B., Hove, B., Jacobs, C., et al. [19]	2010	Use of a mobile platform for assessing urban heat stress in Rotterdam	Mobile measurements using a cargo bicycle platform assessed the urban heat island intensity and the cooling effects of urban parks and greenery in Rotterdam.
Husni, E., Prayoga, G.A., Tamba, J.D., et al. [20]	2022	Microclimate investigation of vehicular traffic on the urban heat island through IoT-Based device	This paper investigates the impact of vehicular traffic on the urban heat island using IoT-based sensors and traffic data.
Kousis, I., Pigliautile, I., Pisello, A.L. [21]	2021	Intra-urban microclimate investigation in urban heat island through a novel mobile monitoring system	The paper presents a novel mobile monitoring (van) system for investigating intra-urban microclimate and urban heat island effects.
Kousis, I.,Manni, M.Pisello, A.L. [22]	2022	Environmental mobile monitoring of urban microclimates: A review	This review examines mobile monitoring systems using motorized and non-motorized vehicles to measure urban microclimates, air quality, light, and noise pollution.
Kousis, I., Pigliautile, I., Pisello, A.L. [23]	2021	A Mobile Vehicle-Based Methodology for Dynamic Microclimate Analysis	This paper presents a vehicle-based methodology for monitoring microclimate conditions in urban areas.
Machado, J.A., de Azevedo, T.R. [24]	2007	Detection of the urban heat-island effect form a surface mobile platform	The paper measures the urban heat island effect in Sao Paulo using a mobile platform with infrared thermometers.
Oke, T.R., Maxwell, G.B. [25]	1975	Urban heat island dynamics in Montreal and Vancouver	The paper used automobile traverses to measure urban heat island dynamics in Montreal and Vancouver.
Rodriguez, L.R., Ramos, J.S., Flor, F.J.S., et al. [26]	2020	Analyzing the urban heat Island: Comprehensive methodology for data gathering and optimal design of mobile transects	The paper proposes a methodology for conducting mobile urban heat island measurements using a vehicle.
Sharifi, E., Soltani, A. [27]	2017	Patterns of Urban Heat Island Effect in Adelaide: A Mobile Traverse Experiment	The paper conducted mobile urban heat island measurements in Adelaide, Australia using vehicle traverses.
Shi, R., Hobbs, B., Zaitchik, B., et al. [28]	2021	Monitoring intra-urban temperature with dense sensor networks: Fixed or mobile? An empirical study in Baltimore, MD	Vehicle-based mobile monitoring alone does not fully capture intra-urban temperature variability compared to a fixed sensor network.
Stewart, I.D. [29]	2011	A systematic review and scientific critique of methodology in modern urban heat island literature	This paper critically discusses the methodological quality of urban heat island studies in the period from 1950 until –2007.
Sun, C.Y, Kato, S., Gou, Z. [30]	2019	Application of Low-Cost Sensors for Urban Heat Island Assessment: A Case Study in Taiwan	This paper used low-cost sensors mounted on mobile vehicles to measure urban heat island effects.
Taha, H., Levinson, R., Mohegh, A., et al. [31]	2018	Air-Temperature Response to Neighborhood-Scale Variations in Albedo and Canopy Cover in the Real World: Fine-Resolution Meteorological Modeling and Mobile Temperature Observations in the Los Angeles Climate Archipelago	The paper used mobile temperature observations from vehicles to characterize the urban heat island effect and evaluate the cooling effects of increasing urban albedo and vegetation.
Voelkel, J., Shandas, V. [32]	2017	Towards Systematic Prediction of Urban Heat Islands: Grounding Measurements, Assessing Modeling Techniques	In this paper, vehicle-based temperature measurements were used to study urban heat island variation and develop statistical models for predicting urban heat.
Yin, Y., Grundstein, A., Mishra, D., et al. [33]	2020	A mobile sensor-based Approach for Analyzing and Mitigating Urban Heat Hazards	The paper presents a mobile sensor-based approach to analyse and mitigate urban heat hazards by collecting high-frequency temperature data from vehicle-mounted sensors.
Yin, Y., Hashemi Tonekaboni, N., Grundstein, A., et al. [34]	2020	Urban ambient air temperature estimation using hyperlocal data from smart vehicle-borne sensors	This paper describes vehicle-mounted sensors used to measure hyperlocal urban ambient air temperature variability and map urban heat hazards.
Zeynali, R., Bitelli, B., Mandanici, E. [35]	2023	Mobile data acquisition and processing in support of an urban heat island study	The paper shows mobile urban heat island measurements with a vehicle and uses various interpolation models to correct the mobile data using fixed station measurements.

contains selected studies of the assessment of the UHI based on different mobile measurement approaches.

Numerical urban climate models can serve as powerful tools to address this challenge, allowing for the simulation of the impacts of various factors and interventions on urban microclimates [36]. Furthermore, these models facilitate the exploration of scenarios that can include modifications in building materials, urban geometry, incorporation of green and blue urban infrastructure, and the consideration of realistic climate change scenarios [37]. This enables the planners to evaluate the effectiveness of these interventions in reducing the UHI effect. Finally, it is possible to integrate real-time data and advanced computational techniques, such as Internet of Things (IoT) sensors, artificial intelligence (AI, machine learning, and deep learning), and Geographic Information Systems (GISs), to improve the accuracy of predictions related to temperature distributions and heat retention within cities [38,39].

Relying on a modelling approach offers significant advantages over depending solely on fixed and mobile measurement campaigns. By closing the spatial and temporal gaps between measurements on a physically sound basis, additional meteorological variables and thermal stress indices such as the Physiological Equivalent Temperature (PET) and the Universal Thermal Climate Index (UTCI) can be calculated. By modelling comprehensive wind and air temperature fields, a better understanding of the complex urban climates can be obtained. This versatility positions them as fundamental tools for urban climatology research and informed decision-making in urban and landscape planning.

This paper is structured as follows. After the introduction, in Section 2, the material and methods, such as observational datasets; regional climate projections, including the bias-correction method; and the FITNAH-3D model for the bioclimatic assessment of the Southern part of Luxembourg (Pro-Sud region) are described. Section 3 presents the main results and findings, while Section 4 discusses these results in relation to other peer-reviewed international studies. Finally, this paper concludes with a short summary and recommendations for future mitigation and adaptation strategies.

2. Materials and Methods

2.1. In Situ Measurements

2.1.1. Long-Term Station Data for the Bias Correction of Climate Change Projections

Direct measurements were taken from the Findel airport SYNOP station (World Meteorological Organization station, WMO, station ID = 06590) and used as a regional reference time series for this study. The SYNOP station is located approximately six kilometres southeast of the city centre of Luxembourg (coordinates: 49°37′57.547″ N/6°13′58.543″ E) and is the only official WMO weather observation station situated within the Grand Duchy of Luxembourg. The daily air temperature records from the Findel station, spanning the 30-year period from 1 January 1971, to 31 December 2000, were used to bias-correct the output from the multi-model ensemble of the regional climate projections. This bias correction process is a paramount step in improving the reliability and accuracy of climate projections at the regional scale, ensuring that the projected data align more closely with the observed meteorological conditions in Luxembourg.

2.1.2. Special UHI Monitoring Network

For the assessment of the current urban heat island intensities in the Pro-Sud region, a dedicated measurement network was established. Measurements started in June 2023 and continued until June 2024. In three villages of different population sizes and extensions, weather stations with sensors for air temperature were installed (

Table 2. Measurement sites for the assessment of the urban heat island intensity in the Pro-Sud region.

Staton Name	Classification	Location	Height Above Sea Level
Esch-sur-Alzette I	city	5.98479 E|49.49418 N	293 m asl.
Esch-sur-Alzette II	rural	5.97786 E|49.48469 N	301 m asl.
Differdange I	city	5.88876 E|49.52401 N	304 m asl.
Differdange II	rural	5.87633 E|49.52224 N	347 m asl.
Kayl I	city	6.04073 E|49.48579 N	289 m asl.
Kayl II	rural	6.05161 E|49.48686 N	299 m asl.

).

Passive, naturally ventilated radiation shields were used to protect the sensors from direct sunlight exposure. At each site, one station was established in the city centre and a second one in the surrounding rural area. Temporal resolution of the measurements was set to 2 min intervals, and all data were aggregated to hourly values for the subsequent analysis.

2.1.3. Mobile Measurements

In addition to the stationary measurements, horizontal temperature transects can usefully supplement the meteorological data of the stationary measurement network. For this purpose, the air temperature and humidity were recorded along a previously defined route in Differdange and Esch-sur-Alzette using the specially equipped LIST environmental measurement vehicle (EMV). The EMV is equipped with two actively ventilated psychrometers for measuring dry and wet bulb temperatures (Figure 1).

Data were measured at 6 s intervals and stored for postprocessing and analysis. In addition, the measurements were geolocated with the help of a GPS, which was also included in the EMV. This enables the collection of area-wide information on the air temperature distribution in a city. The measurements were conducted during autochthonous weather conditions because the temperature differences between the city and the surrounding area are particularly pronounced under such conditions. The UHI effect is typically strongest at night. Therefore, the measurements were carried out between 22:00 and 05:00.

2.2. Climate Projections

We used climate projections from the CORDEX initiative for the European domain (EURO-CORDEX), which is part of the World Climate Research Programme, for the assessment of the future thermal stress levels in the Pro-Sud region. The EURO-CORDEX project aims to provide climate change information at the regional level. We extracted air temperature, humidity, surface wind speed, and global radiation from climate model simulations for the Representative Concentration Pathways 4.5 (RCP4.5) over the period 1971 to 2045. These data were used to force the FITNAH-3D model to simulate the future conditions in the area under investigation. The specific climate models used in the analysis are listed in

Table 3. Regional climate change projections with model abbreviations, the driving Global Climate Model (GCM), the Regional Climate Model (RCM) used for the dynamical downscaling as well as the institutions that are responsible for the model runs; temporal resolution: daily data; period: 1971–2045, RCP45 [8].

Model Abbreviation	Global Climate Model (GCM)	Regional Climate Model (RCM)
M1	CNRM-CERFACS-CNRM-CM5	CNRM-ALADIN53_v1
M2	CNRM-CERFACS-CNRM-CM5	RMIB-UGent-ALARO-0_v1
M3	MOHC-HadGEM2-ES	KNMI-RACMO22E_v2
M4	MOHC-HadGEM2-ES	SMHI-RCA4_v1
M5	MPI-M-MPI-ESM-LR	MPI-CSC-REMO2009_v1
M6	MPI-M-MPI-ESM-LR	SMHI-RCA4_v1a
M7	NCC-NorESM1-M	DMI-HIRHAM5_v2
M8	MOHC-HadGEM2-ES	CLMcom-CCLM4-8-17_v1
M9	CNRM-CERFACS-CNRM-CM5	SMHI-RCA4_v1
M10	IPSL-IPSL-CM5A-MR	IPSL-INERIS-WRF331F_v1
M11	CNRM-CERFACS-CNRM-CM5	CLMcom-CCLM4-8-17_v1
M12	ICHEC-EC-EARTH	KNMI-RACMO22E_v1
M13	IPSL-IPSL-CM5A	SMHI-RCA4_v1
M14	MPI-M-MPI-ESM-LR	CLMcom-CCLM4-8-17_v1

.

Since the output of a Regional Climate Model (RCM) is—in most cases—biased towards observed fields [40], we follow the example of previous impact studies [8,41,42] and correct the systematic biases of the RCMs with the data from the WMO station Findel based on a non-parametric quantile mapping technique. This method aligns the distribution of the model output with the observed data from the WMO station Findel. This method involves transforming the cumulative distribution function (CDF) of the model output to match the CDF of the observed data. This is performed by using empirical quantiles, without specifying a parametric form for the distributions [43]. The process involves the following steps: (i) sorting the model output and observed data separately in ascending order, (ii) calculating the CDFs of the model output and observed data based on their respective rankings, (iii) determining the quantile mapping function by matching the empirical quantiles of the model output with those of the observed data. This is completed by interpolating between the pairs of corresponding quantile values, and finally, (iv) applying the quantile mapping function to the model output to correct the bias. This flexible method can correct biases across the entire distribution of the model output, not just at specific quantiles. The analysis was conducted using the R programming language for statistical computing [44], with the support of the R-package “qmap” [45]. For this study, three different scenarios, including the foreseen construction activities, were defined (

Table 4. Different scenarios used in this study.

Scenario Name	Description
Status Quo	- FITNAH-3D modelling - Current climate (2019–2023) - Current building structure
Scenario I; weak climate change signal 2045	- FITNAH-3D modelling - Future climate (multi-model ensemble from CORDEX 2040–2050); 25th percentile from RCP45 (+0.7 K compared to Status Quo - Current building structure plus additional planned buildings
Scenario II; strong climate change signal 2045	- FITNAH-3D modelling - Future climate (multi-model ensemble from CORDEX 2040–2050); 75th percentile from RCP45 (+2.7 K compared to Status Quo - Current building structure plus additional planned buildings

).

2.3. Urban Modelling

To model the current and future thermal stress levels in the area under investigation, the FITNAH-3D (Flow over Irregular Terrain with Natural and Anthropogenic Heat sources) was used. It is a non-hydrostatic meteorological model designed for simulating wind fields in three dimensions. It was developed with the help of the German Research Foundation’s priority program and is a powerful tool for addressing environmental and meteorological issues in urban and landscape planning, complementing on-site measurements. The model has been applied in various cities in Switzerland, such as Zurich [46], Geneva [47], and Bern [48], or in Bonn and Berlin (Germany) [49,50].

FITNAH-3D, similar to the numerical climate and weather forecast models, is based on a set of fundamental equations, including the Navier–Stokes equation of motion (conservation of momentum), the continuity equation (conservation of mass), and the first law of thermodynamics (conservation of energy). The local climatic characteristics of the study area need to be accurately captured by the numerical model. A finer grid resolution provides more details and structures but also increases the computational requirements and need for input data. A compromise between necessity and feasibility must be found. In our study, the horizontal spatial mesh size used for the FITNAH-3D model is 5 m. The vertical grid is non-equidistant, with denser spacing near the surface to realistically capture the strong variations in meteorological variables. The lowest computing levels are at 2, 5, 10, 15, 20, 30, 40, 50, and 70 m above ground. The grid spacing increases with height, and the model top is located at 3000 m above sea level, where the disturbances from the underlying terrain and land use are assumed to have diminished. FITNAH-3D meets the standards defined in Verein Deutscher Ingenieure (VDI) guideline 3787, Bl.7 [51] for mesoscale wind field models in connection with dynamically and thermally induced wind fields. More details of the FITNAH-3D model are given by Gross [52] and Geo-Net [46,47,48].

Within our study, we used specific input data for the numerical model FITNAH-3D to characterize the landscape of the study area. For the Pro-Sud region, a Digital Elevation Model (DEM) with 1 m spatial resolution was used. Land cover data were taken from the CORINE dataset, and in addition, orthorectified aerial photographs were used. The building and structural height as well as the degree of sealing were obtained from a 3D cadastre of Luxembourg. In cases when data were unavailable, the necessary information was derived from the land cover data. All input data were also compared regarding their plausibility using a current aerial photo dataset.

3. Results

3.1. UHI Intensity Based on Measurements

To supplement and validate the model results with in situ observations, both stationary and complementary mobile measurements of air temperature were conducted. In the municipalities of Kayl, Differdange, and Esch-sur-Alzette, air temperature sensors (enclosed with passive, naturally ventilated radiation shields) were installed at a height of two meters in green spaces on the urban periphery and in the sealed city centre (a total of six measurement points). These measurements were carried out over a period of one year (26 June 2023–22 May 2024) with a time resolution of two minutes. Figure 2 shows the measurement results (maximum difference of daily averages, maximum difference of hourly values, and hourly averages) exemplarily for the two special measurement stations in Esch-sur-Alzette. The maximum differences in daily averages range between 0.7 °C and 2.3 °C, and the differences in hourly values range from 0.7 °C to 5.3 °C. The temperature differences were lowest in the months of November/December 2023 and in the first half of January 2024. In contrast, from July until October 2023, as well as February, March, April, and May 2024, pronounced urban heat island effects can be observed due to higher air temperature differences between the city centre and the rural area (Ellergronn). Ellergronn is a natural reserve area with rich biodiversity and an important role as a vital ecological zone within the region. The highest difference was measured in early September 2023. The data collected in the municipalities of Differdange and Kayl revealed comparable results, however, with smaller observed differences in air temperatures. This was particularly evident in Kayl, where the air temperature difference between rural and urban areas was on average 60% lower than in Esch-sur-Alzette. This can be attributed to the substantially lower population density and corresponding lower urban sealing, which is only a quarter of that in Esch-sur-Alzette.

Mobile air temperature measurements are a fundamental method for the assessment of UHIs, as they can provide high-resolution spatial and temporal data that traditional fixed monitoring stations often cannot deliver. The mobility of measurement systems, such as vehicles equipped with portable sensors, enables the collection of temperature data across diverse urban microclimates, enabling the identification of temperature gradients influenced by land use, vegetation cover, and building materials. Studies employing mobile measurements have demonstrated their effectiveness in mapping UHI intensity and distribution, revealing localized heat hotspots that inform urban planning and climate adaptation strategies [53,54]. Figure 3 shows the results of a nighttime profile measurement in Esch-sur-Alzette on 1 February 2024 (10:12 p.m.–10:41 p.m.) with starting and ending points at the Ellergronn rural location. The maximum air temperature difference between the green area and the city centre is 4 °C. The air temperature difference between the start and end times was 0.3 °C. Therefore, the measured differences between the city centre and Ellergronn can mainly be attributed to the UHI effect.

3.2. CORDEX Regional Climate Projections

A comprehensive analysis of the future climate conditions in Luxembourg (49°36′ N, 6°7′ E) was conducted by extracting time series data reflecting daily mean, maximum, and minimum air temperatures under the Representative Concentration Pathway 4.5 (RCP45) scenario from the EU-CORDEX archive. The RCP45 scenario represents a moderately ambitious climate mitigation pathway, wherein the radiative forcing stabilizes at a level of 4.5 W/m² prior to the year 2100 because of the implementation of a variety of technologies and strategic measures aimed at reducing greenhouse gas emissions. The air temperature data utilized for this analysis were derived from a three-by-three grid box from the model output and spatially averaged. This averaging process did not incorporate a weighting function. For the forcing of the FITNAH-3D urban climate model, absolute temperature values are required. Before the output of the multi-model ensemble data could be used, a bias correction was applied to the data. This step ensures the alignment of the model’s input data with observed conditions and minimises discrepancies that might otherwise influence the FITNAH-3D results. Figure 4 illustrates the projected changes in the minimum, maximum, and mean air temperatures for the RCP45 scenario, highlighting significant trends in air temperature. The findings indicate a predicted increase in near-surface air temperatures by the end of this century: specifically, an increase of 2.1 K for both minimum and maximum annual temperatures, coupled with a 2.2 K rise in the mean annual temperature. These projections underscore the pressing need for adaptive strategies in urban planning and climate resilience. For the Pro-Sud project, it was decided that the forcing data for the subsequent FITNAH modelling would use the 25th and 75th percentiles of the temperature projections, rather than relying solely on the ensemble mean values.

3.3. FITNAH-3D Model Results

The area under investigation (Pro-Sud) covers a total area of 201 km² with a population of approximately 177,000 inhabitants. The future developments incorporated in this study indicate a significant increase in built-up areas, leading to profound changes in land use. The inhabited built-up area rises from 24 km² by 21% to 29 km², while the uninhabited built-up area increases from 13 km² by 28% to 17 km². Consequently, the share of the total built-up area in relation to the total area rises from 18.7% to a projected 23.1%. Since the total area remains constant, this comes at the expense of other land use classes. Green open spaces are particularly affected, decreasing from 96 km² by 8.2% to 88 km². Forest areas also show a decrease of 1.06% to 55 km². The changes in land use mainly result from the conversion of unsealed areas into sealed ones, particularly for settlement purposes. All results for the complete Pro-Sud area can be found in the public archive at geoportail.lu. The results for the air temperature at 4 a.m. for the present and future climate conditions (Scenario II) are shown in Figure 5 and Figure 6.

In green and blue are indicated in both figures the areas where new construction activities are foreseen within the next years. A strong increase in the air temperature between the Status Quo and Scenario II can be observed. Especially in the densely populated and heavily sealed areas, this increase is more pronounced.

For a more detailed analysis, air temperatures at 4 a.m. were calculated for six land use classes and the three different scenarios and are presented by boxplots in Figure 7. As expected, the lowest median air temperatures (15.1 °C) were observed for the Status Quo scenario and the land use class green vegetation/spaces. Sealed traffic areas show the highest median air temperatures of 17.9 °C for the Status Quo. For Scenario I, a mean increase of 0.7 K with the lowest values for the water bodies (+0.6 K) and highest values for the uninhabited settlement areas (+1.1 K) are observed. The same trend in the differences (median increase of +2.7 K) can be observed for Scenario II with an increase of 2.2 K for water bodies and +3.0 K for the uninhabited settlement areas.

For a more detailed analysis, the northern part of Esch-sur-Alzette was chosen (Figure 8). Taking the planned new buildings into account, the area expands from the current 60 km² by 10.04% to a projected area of 66 km². In addition to the development of new residential areas, there is also an increase in density within the already existing areas. Currently, 41% of the total area is designated as green open spaces, while inhabited built-up areas account for 21.3%. In the future, the green open spaces will be reduced by 38%, while the inhabited built-up areas will increase by 17.1%. The main reason for these opposing developments is that nearly all planned land use changes will result in an expansion of built-up areas.

3.4. Selected Adaptation and Mitigation Measures

A selection of adaptation and mitigation measures to reduce the urban heat island effect, categorized by their mechanism of action, their effect on the urban environment, and their spatial implementation possibilities, are presented in

Table 5. Recommendations for spatial unit-specific urban climate measures for the Pro-Sud region.

Measure	Explanation	Effect	Spatial Implementation
Indoor/backyard greening	- Vegetation and unsealing	- Reducing heat stress	- Inner and backyard courtyards
Creating public green spaces	- Small parks and green spaces in inner-city areas	- Reducing heat stress - Networking of green spaces	- Vacant lots, larger backyards
Climate-optimised design of outdoor surfaces	- Bright colours (especially roofs) and building materials that retain little heat	- Reducing heat stress	- Roofs, if applicable, roads, paths, squares, parking lots
Minimize unsealing/sealing content	- Lawns or partial sealing - Low number of above-ground parking spaces	- Reducing heat stress	- Roads, squares, parking lots, buildings, courtyards
Blue–green traffic space design	- Blue or green measures for increasing the proportion of vegetation in the traffic area and creating open water areas	- Reducing heat stress	- Roads, paths, squares, parking lots
Shading of outdoor recreation areas	- Trees or structural measures (awnings, roofing, sun sails)	- Reducing heat stress	- Roads, paths, squares, parking lots, buildings in the residential and working environment
Developing and optimising public green spaces	- Microclimatic diversity of green spaces (open meadows, trees, water areas, plantations)	- Reducing heat stress	- Green and open spaces - Roads, squares, parking lots
Maintaining and improving cold air production	- Protection of cold air generation areas and enhancement of areas with lower cold air production - Implement evaporative plants	- Reducing heat stress	- Green and open spaces
Protecting, expanding, and creating open, moving water surfaces	- Urban climate function of larger flowing and still waters areas - Low-roughness ventilation corridors that transport cold and fresh air	- During summer, bodies of water have a cooling effect on their immediate surroundings during the day	- Waters - Green and open spaces
Paying attention to the position of the building structure and spacing areas	- Building arrangement parallel to the cold air flow and/or sufficient (green) open spaces between the buildings	- Improvement of cold air flow - Reduction in heat accumulation	- New construction, building complexes
Avoidance of exchange barriers	- Avoid structural or natural obstacles running transversely to the direction of flow in the area influenced by cold air flows	- Protection of the air exchange system	- Green and open spaces, - Well-ventilated residential and commercial areas
Protection and networking of areas relevant to the cold air balance	- Keeping large green spaces free, preferably water-supplied and characterised by flat vegetation	- Protection against excessive overheating and deterioration of ventilation	- Green and open spaces
Green roof	- Extensive or intensive green roofs	- Improvement of the indoor climate - With large-scale implementation and low roof height, improvement of the immediately adjacent outdoor climate possible	- Flat roofs - If necessary, gently sloping roofs
Façade greening	- Ground- or system-bound façade greening	- Improvement of the indoor climate and the immediately adjacent outdoor climate	- Building
Shading of buildings by trees or structural measures	- Façade greening, trees, balcony design, structural measures such as external sun protection elements, reflective solar control glass	- Improvement of the indoor climate	- Building
Technical building cooling	- Adiabatic exhaust air cooling in which rainwater is used - Adsorption chillers powered by solar energy or waste heat	- Cooling of the interior of buildings through building air conditioning that is as sustainable as possible	- Buildings in which passive measures cannot be sufficiently applied

. These measures, ranging from the micro-scale (in-door greening) to the macro-scale (large-scale green spaces), are crucial for enhancing the resilience of urban areas to rising temperatures and thermal stress driven by climate change. The effectiveness of these measures depends on several factors including the local climatic conditions, the urban morphology, and the scale of implementation. While some measures can be implemented individually by residents or businesses (e.g., in-door greening), others require more extensive, city-wide planning and investment (e.g., developing and optimising public green spaces). These measures provide a framework for urban planners and policymakers to strategically select and prioritize measures that best suit the specific needs and conditions of their communities. The selection of measures should be guided by a detailed assessment of the existing urban environment, considering factors such as building density, land cover types, prevailing wind patterns, and the availability of resources.

4. Discussion

This study investigated the urban heat island effect in Luxembourg’s Pro-Sud region, applying an integrated approach that combines field measurements, the FITNAH-3D urban climate model, and regional climate projections sourced from the EURO-CORDEX initiative. Figure 2, displaying daily mean air temperatures and temperature differences between urban and rural sites in Esch-sur-Alzette, reveals a distinct diurnal cycle and seasonal variations in UHI intensity. The larger temperature differences during warmer months are consistent with findings from numerous studies worldwide that demonstrate increased UHI effects during periods of high solar radiation and ambient temperatures [55,56]. The observed reduced temperature differences during the colder months also align with typical UHI behaviour, where the effect is often less pronounced due to reduced solar input and altered thermal properties of urban materials [57]. Our findings reveal a pronounced UHI intensity, with air temperature differences between urban and rural areas reaching values of 5.3 °C during peak periods (Figure 2). These results strongly corroborate the well-established phenomenon of elevated temperatures in urban settings compared to their surrounding rural counterparts [3,13,58,59]. Several contributing factors drive this UHI intensity, including the extensive sealed surface area of urban structures, the reduced albedo resulting from the dominant use of concrete and asphalt surfaces [2], and the substantial anthropogenic heat emissions from transportation, industrial processes, and energy consumption within the built environment.

UHI intensity, ideally measured by mobile measurements, offers essential insights into the spatial and temporal variations of urban temperatures compared to rural surroundings. These measurements are valuable because they capture real-time data across diverse urban landscapes, providing detailed mapping of temperature discrepancies caused by factors like building density, vegetation cover, and anthropogenic heat sources. Furthermore, these data sets combined with the long-term in situ measurements are a perfect data set for model validation. Figure 3 depicts the results of a night-time mobile measurement transect with the mobile air quality lab of LIST (Figure 1), showing the spatial heterogeneity of the UHI effect within Esch-sur-Alzette. This observation of a significant temperature gradient of more than 4 K between rural and urban areas is typical in UHI studies, reflecting the influence of land cover and surface properties on local temperature distributions. The magnitude of the temperature difference observed (4 °C) falls within the range reported in comparable studies conducted in other European cities [26,58]. Incorporating regional climate projections from EURO-CORDEX, specifically those associated with RCP4.5, allowed an assessment of projected future UHI conditions.

The RCP4.5 emission scenario is a moderate path for future greenhouse gas emissions and related climate change effects. It aims to stabilize the amount of energy trapped in the atmosphere at around 4.5 watts per square meter by 2100 through technological advancements, policies, and changes in behaviour to reduce greenhouse gas emissions. Strategies include increased energy efficiency, a shift to renewable energy sources, carbon capture and storage, sustainable land use, and reforestation. Our analysis revealed a substantial intensification of UHI effects under future climate scenarios, projecting a temperature increase of 2.1 K to 2.2 K for minimum and maximum annual temperatures, respectively, by the end of the 21st century for Luxembourg (Figure 4) [8]. This projected warming significantly exacerbates the UHI effect, thereby underscoring the imperative to implement robust adaptation and mitigation strategies within the urban landscape. Furthermore, the use of the 25th and 75th percentiles of the projected temperature data, rather than the ensemble mean, provides a more conservative yet realistic perspective on the potential range of future impacts.

Numerical urban mesoscale models such as FITNAH-3D are useful tools to answer urban climatological questions that can hardly be addressed by pure measurement campaigns. They can close the spatial and/or temporal gaps between the measurements on a physically sound basis, calculating additional meteorological variables and determining wind and temperature fields as well as thermal stress indices such as PET. The model calculations also offer the advantage that planning variants and compensatory measures can be studied in terms of their effect and efficiency, and optimized solutions can be found in this way [60]. An intercomparison of four different urban climate models including FITNAH-3D was recently published by Burger et al. highlighting the capabilities of FITNAH-3D to reproduce strong small-scale urban temperature gradients [61]. Figure 5 and Figure 6 show FITNAH-3D model results under current (Status Quo) and future (Scenario II) climate conditions and visualize the spatial patterns of air temperature and the projected intensification of the UHI under future climate scenarios. Forcing the FITNAH-3D model [62] with current and future climate boundary conditions as well as with the current and future building structure provided a detailed spatial resolution analysis of the UHI effect, showing clear temperature variations across diverse land use categories. The model results are consistent with many studies that have shown the importance of incorporating both urban morphology and climate change projections to accurately predict future UHI characteristics [63,64,65,66]. The pronounced increases in UHI intensity observed in the model simulations under Scenario II highlight the need for proactive urban planning and mitigation measures. The observed enhancement of UHI intensity under projected climate scenarios emphasizes the crucial need for tailored urban planning interventions at the local spatial scale. These interventions must focus on sustainable design principles and effective mitigation strategies to mitigate the increasing thermal stress experienced within urban areas. The boxplots in Figure 7 show air temperatures at 4 a.m. for six different land use categories and the three scenarios (Status Quo and two future scenarios), highlighting the differences in the modelled air temperature. The lower temperatures observed for green spaces compared to sealed surfaces concur with previous research [67], underscoring the importance of green infrastructure in mitigating UHI effects. The increased temperatures under future scenarios highlight the compounded effect of urbanization and climate change. The magnitude of the increase for water bodies being lower than that for sealed areas further supports the findings of previous studies [19]. Figure 8 visually demonstrates the changes in land use and built-up areas in a selected commune (Esch-sur-Alzette), showing an expansion of built-up areas and a reduction in green spaces for the two future scenarios. The impact of such land use changes is directly reflected in the increased UHI intensity with corresponding higher PET values.

Key strategies, including the incorporation of extensive green and blue infrastructure [68], the strategic selection of building materials with higher albedo values [69], and concerted efforts to reduce anthropogenic heat emissions [70], are listed in Table 5.

Their implementation, however, in the urban planning process requires a shift towards more sustainable practices that integrate ecological principles with efficient urban designs. Aspects of vegetation cover and its effects on the urban heat islands and environmental benefits are broadly discussed in the literature. Urban vegetation can cool cities between and has positive social, health, and economic effects. Temperature reductions depending on vegetation type and density between 1 °C and 5 °C are possible [71,72]. Several studies have shown that green roofs can reduce roof surface temperatures by up to 25 °C and ambient temperatures in surrounding areas by 1–3 °C during peak summer months [73,74,75,76]. Related to the suggested measures to increase the presence of water bodies (Table 5), a recent study by Qiao et al. highlighted the importance of the relationship between water body characteristics and their cooling effects [77]. Especially due to the projected changes in future precipitation patterns with related increasing drought conditions, the importance of blue infrastructure within urban areas will constantly increase [78,79].

Our study provides a valuable contribution to understanding the UHI intensity in Luxembourg under current and future climate conditions. Nevertheless, some limitations warrant acknowledgement. Due to computational constraints, only one RCP could be used for the forcing of the urban climate model, and due to the same reason, an ensemble approach was not suitable for the urban modelling approach.

The time frame for the in situ measurements covered only one year, limiting the range of captured seasonal and inter-annual variability in UHI behaviour. Nevertheless, a short comparison with the time series from an official meteorological station in Luxemburg (WMO ID = 06590) showed that 2024 was no extraordinary year in terms of thermal stress levels. Future research should include extended field campaigns spanning multiple years, incorporating diverse climatic conditions and expanding the geographical scope of analysis to encompass a broader range of urban settings within Luxembourg and other geographical regions with similar climatic characteristics.

5. Conclusions

Our study comprehensively assessed the UHI effect in Luxembourg’s Pro-Sud region, integrating field measurements, urban modelling, and climate projections. The results show a significant and intensifying UHI, with substantial thermal stress differences and projected increases under future climate conditions. The integrated approach provided a more complete understanding of UHI dynamics than single methodologies. While limitations still exist (e.g., measurement duration), this research offers valuable insights for developing adaptation and mitigation strategies emphasising green infrastructure and sustainable urban planning. Further research should investigate (a) how artificial intelligence could support the selection of tailored urban design features (e.g., building density, material properties, and green spaces) and influence the spatial and temporal variability of the UHI effect and (b) the long-term socio-economic impacts of an intensified UHI and how can these be addressed in a more efficient way.