Version classiqueVersion mobile

The Mediterranean region under climate change

Jean-Paul Moatti
Stéphane Thiébault

Part 3. Adaptation, resilience, conservation of resources and prevention of risk

Sub-chapter 3.1.3. Observation systems and urban climate modelling

Elodie Briche, Nicolas Martin et Salem Dahech

Texte intégral

1Mediterranean cities are particularly vulnerable hot-spots of climate change in light of their spatial and geographical configurations (i.e. the proximity of the coastline, high water demand, a climate favouring summer drought and extreme meteorological events such as intense rains), and their continuous growth.

2Understanding the interlinkage between spatial and temporal scales is crucial in climatology if we are also to understand phenomena occurring at urban scales. The simulations run with Regional Climate Models (RCMs) show, under various scenarios, the frequency and the intensity of extreme meteorological events such as summer drought, which could reinforce Urban Heat Islands (UHIs), and enable us to anticipate the future without nevertheless providing solutions at local scales. It is therefore beneficial and necessary to develop models, instruments and measurements capable of simulating local climate phenomena.

3These local-scale approaches mark the first step towards adaptive methods for coping with climate change in the short and medium term. They could lead to the reconsideration of some urban policies (densification, land planning or energy policies), as well as architectural practices (orientation, use of materials). Planners will have to integrate the climate challenge within their practices and develop urban solutions in the Mediterranean to limit the negative effects of climate change (e.g. thermal discomfort, pollution), particularly in the summer period.

Observations adapted to the urban complexity of Mediterranean cities

4Specific measures are necessary in urban environments to spatialize meteorological parameters and therefore to compensate for the inadequacy of data recorded by conventional networks. However, at city scale, these measures require that particular consideration be given to the choice of measuring points and periods, and of instrument calibration. For instance, the UHI can be measured at three main altitude levels: surface, about 2 meters from the surface, and in the open air (Carrega, 2003).

Types of measurement in cities: stationary and mobile air temperature

5There are two types of measurement: stationary or mobile. In the first case, they are recorded under shelter within a network of meteorological sensors or stations (fig. 1) established for densifying a conventional network or for enabling climate observations at micro and local scales according to a given question: UHI, temporal and spatial variation of temperatures, etc. Precautions such as clearing of the environment, ventilation of the site throughout the day, and safety measures against power failure, vandalism and theft, must be taken when installing the equipment in cities.

6The duration of mobile measurements should be limited (max. 30-50 min) to avoid bias due to temporal variation. Depending on the travel distance, journeys can be made by car, on foot or by bike (box 1). Measurements can be recorded continuously or discontinuously with stops of no more than two minutes per measurement point. It is important to note that this type of field survey requires considerable physical and material efforts.

Application: measurement of the urban heat island in Sfax

7The UHI is studied using mobile air temperature measurements (at 2 meters above ground and, whenever possible, a minimum distance of 5 meters from the buildings) that are recorded with portable sensors. These measurements, carried out in July 2012, were only recorded at night during radiative conditions: they reveal spatial thermal disparities between the diverse types of surface in the city and its surrounding countryside. The measurements, which were obtained by three teams to ensure maximum coverage of the agglomeration, were conducted between 1 and 2am, with very light wind and fluid traffic.

Figure 1
Temperature and humidity monitoring systems under shelter during a summer period measurement campaign in Marseille. Briche, 2016.

8A temperature difference of 4 to 7°C is observed (fig. 4) between the city centre – the medina – and the surrounding countryside (a large agricultural area situated at more than 12 km from the medina). The amount of heat stored in the streets and buildings of the city centre is larger than that stored in the countryside. This amount of energy is returned at night. At the reference point, the medina, the largest temperatures are recorded because of the lack of ventilation, the wall effect caused by narrow streets (0.5 to 2 meters) and the density of buildings (100%) (Dahech, 2012). The coolest points are situated in fields at about 12 km from the medina. Indeed, vegetation turns part of the incident radiation into latent heat through evapotranspiration and therefore stores less solar energy. The mild environment generated by green spaces is undeniable. At night, the public park Touta (Sfax), is 2 to 3°C cooler than the surrounding dense areas of the city and provides a cooling breeze for the surrounding environment.

Is downscaling in modelling necessary?

9Climate models are the only tools that enable us to anticipate future climate conditions. Despite the uncertainties usually linked with numerical modelling (scenarios, complexity of physicochemical processes, downscaling), such models can provide a long-term vision at regional scale by taking into account the potential changes in global socio-economic conditions.

Box 1
Mobile temperature measurements using a bike
Because of the high number of factors influencing temperature, its value changes rapidly over short distances in an urban fragment. The colour and coating materials (land use), the morphology of buildings (e.g. density, height, street widths), the wind and sun orientation of avenues, and the location of vegetated and water areas (from fountains to larger water bodies) are factors that modify the temperature since they have different impacts on the reflection and absorption of sunlight, and on the coupled process evaporation/condensation. Therefore, assessing temperatures in cities, using a bike for mobile measurements (a thermometer is attached to the handlebars at 1 metre above the ground, fig. 2) is one of the most efficient ways of capturing their spatial variation thanks to the speed and the fluidity of movement offered by this type of transport.

Figure 2
Bike with sensors for mobile temperature measurements.
Photo: Martin, 2015.

Several measurement campaigns were carried out between May and September 2015 in Nice over 54 days (approximately a day on two) in the middle of the afternoon. The aim of the study was to record maximum daily temperatures and, even though this was difficult to achieve due to temporal instability and the different timings of peak temperature that depend on the type of space, the resultant mapping remained close to the daily thermal maximum.
In urban and semi-urban areas, temperatures oscillate between 25.5 and 28°C (fig. 3). Sometimes, the variations observed are sensitive over short distances. This can be the consequence of a change in land use (e.g. from concrete coating to a grassy area of land with a loss of 1°C on average over few meters); on a steep slope, the combination of a paved road with a light-coloured bank that directly reflects the sunlight implies an increase in temperature (about 2 ° C); finally the orientation of the streets and the shadows cast by buildings make for streets that are more or less warm depending on the position of the sun in the sky (in the summer period, at around 3.30-4pm, streets that are orientated north-south are less exposed to the sunlight than those orientated east-west with a difference of 1°C on average).

Figure 3
Map and analysis of mobile temperature measurements in Nice in 2015. Aerial photo and 3D perspective: Google Earth.

Are models adapted to the urban scale?

10Numerous Mediterranean countries rely either on public organisations for modelling with regional climate models, or on collaboration with climate services to develop indexes generated from simulations, depending on specific applied questions (e.g. http://www.climrun.e/​).

11However, the RCMs allow for the prospective regional analysis of potential future conditions (e.g. identification of trends, increase in interannual variation), but do not provide answers at the intraurban scale.

12Other types of models that are better adapted to urban scales and that enable us to refine the analysis include LASER/F (Najjar et al., 2005) for the urban canyon and TEB/ISBA coupled with the mesoscale atmospheric model Meso-NH for the district and agglomeration. Validation of the model consists of the comparison between outputs and observed field data (Kastendeuch et al., 2010).

13Models are also characterized by differing methods of calculation: statistical and/or dynamic (box 2).

14The sensitivity of models depends on the temporal and spatial scales used, which explains the difficulty of disaggregating data at the urban “social construction” scale. Moreover, the more parameters there are, the larger the uncertainty margins become, because the interactions of climate phenomena are difficult to simulate in a realistic way.

Figure 4
Spatial distribution of night air temperature, at 2 meters above the ground (in °C), in the Sfax agglomeration (Dahech, 2012).

Application: simulation of the humidex index for the Nice region

15The study of thermal urban comfort, and therefore of the health impact of climate warming on cities, is possible thanks to the availability of local-scale temperature data. However, to best assess the health impact of this variable, it is necessary to weight it with the variable “humidity”. This variable considerably alters our perception and physical strength towards high temperatures. Indeed, the human body reduces its capacity to cool down when the air moisture increases because the perspiration cannot evaporate as readily.

Box 2
Global atmospheric circulation models feature a spatial resolution of several tens of kilometers or more, for instance 12 km in the ALADIN-Climat model (limited area version of the previous model). Such dimensions are however too large to be of use to decision-makers and planners as well as any institutions willing to take action based on fine spatial resolution data (Martin et al., 2013).
This gap between the need for fine spatial resolution and the outputs provided by models justifies the efforts that have been made for spatial downscaling. Several approaches exist to improve the spatial resolution of climate predictions. Dynamic logic differs from statistical logic. In the case of the former, an RCM will explicitly solve the physics and dynamics of a climate system with a resolution of 12 km at best (e.g. ALADIN-Climat). For the latter option, the improvement of the spatial resolution relies on the search for statistical relationships between local variables and each pixel of GCMs or RCMs’ outputs.
There are several methods to statistically refine the spatial resolution of a phenomenon. The environmental regression technique has been selected to substantially improve the spatial resolution of temperatures provided by climate simulations (resolution of 25m using a digital terrain model and up to a few meters with a digital elevation model for urban environments). This approach enables the creation of data from variables that are considered explanatory and that are involved in the studied phenomenon. This is at the interface between a deterministic approach and another based on spatial interpolation since the choice of environmental variables is not random but justified by strong statistical relationships of these data with the phenomenon to be modelled. For instance, the impact of oceanic thermal inertia on temperature can be integrated within a regression model using the “distance to the sea” variable. In urban areas, the evaporation generated by green spaces and/or water features, which provide fresh air, is also considered through the variable “distance”.

16The humidex is an index developed by Canadian meteorologists to take into account this combination of factors. Each humidex interval corresponds with a level of comfort: between 30 and 39 a certain discomfort is perceived, above 40 a strong discomfort, and above 45 there is a threat with possible heat stroke, which is imminent when the humidex is greater than 54. Therefore, by using existing maps of past and future temperatures obtained by downscaling, assumptions on the relative humidity have been made to spatialize the humidex in the city of Nice and its surroundings. The maps show maximum daily values for the month of July (mid-afternoon) within the range of 32 to 41 at the end of the 20th century assuming 80% humidity, and within a range of 43 to 54 at the end of the 21st century under a pessimistic scenario using the same humidity assumption (fig. 5). From a certain discomfort to a strong one, the evolution of the climate will lead to potential or imminent heat stroke conditions.

Figure 5
Spatial modelling of the humidex index over a past and a future period for the Nice region.

17In Mediterranean cities, local initiatives are being carried out by various research teams to better understand the urban climate, in particular the spatial and temporal variation of the UHI. Emphasis is also given to studies of summer periods in view of Mediterranean climate characteristics and of the increasing probability of future droughts.

18Measuring, understanding and modelling climate at a fine scale enables action to be taken for the transformation of the city by proposing more thought-out urban settlements and structures. New climate change challenges suggest new urban planning methods, and require reliable fine-scale data on climate as well as on impact monitoring through the use of indexes (e.g. humidex) and indicators. These data are crucial to create a tool adapted to urban complexity and specificities. An interdisciplinary approach capable of taking into account the complexity of contemporary cities through co-actions between scientists and political actors is also needed.



Carrega P., 2003
Le climat aux échelles fines. Publication de l’Association Internationale de Climatologie, 15, 19-30.

Carrega P., 1994
Topoclimatologie et habitat. Revue d’analyse spatiale quantitative et appliquée, 35 et 36, 408 p.

Dahech S., 2012
Evolution de la répartition spatiale des températures de l’air et de surface dans l’agglomération de Sfax (1975-2010) et impact sur la consommation d’énergie durant la saison chaude. Climatologie, numéro spécial Climats et changement climatique dans les villes, 11-33.

Kastendeuch P., Najjar G., Lacarrere P., Colin G., 2010
Modélisation de lîlot de chaleur urbain à Strasbourg. Climatologie, 7, 21-35.

Martin N., Carrega P., Adnes C., 2013
Downscaling à fine résolution spatiale des températures actuelles et futures par modélisation statistique des sorties ALADIN-Climat sur les Alpes-Maritimes (France). Climatologie, 10, pp 51-72.

Najjar G., Kastendeuch P., Ringenbach N., Colin J.R., Stoll M.P., Nerry F., Bernard J., De Hatten A., Luhahe R., Viville D., 2005
Bilans radiatifs et énergie dans un cayon urbain. Annales de l’AIC, 2, 41-54.

Table des illustrations

Légende Figure 1Temperature and humidity monitoring systems under shelter during a summer period measurement campaign in Marseille. Briche, 2016.
Fichier image/jpeg, 148k
Légende Figure 2Bike with sensors for mobile temperature measurements.Photo: Martin, 2015.
Fichier image/jpeg, 239k
Légende Figure 3Map and analysis of mobile temperature measurements in Nice in 2015. Aerial photo and 3D perspective: Google Earth.
Fichier image/jpeg, 396k
Légende Figure 4Spatial distribution of night air temperature, at 2 meters above the ground (in °C), in the Sfax agglomeration (Dahech, 2012).
Fichier image/jpeg, 358k
Légende Figure 5Spatial modelling of the humidex index over a past and a future period for the Nice region.
Fichier image/jpeg, 232k


AMU, France
Climatologist, LPED (UMR 151)-Aix-Marseille University – IRD, France

Nice University, France
Geographer, ESPACE (UMR 7300), University of Nice, France

Sfax University, Tunisia
Climatologist, SYFACTE laboratory, University of Sfax, Tunisia

Le texte et les autres éléments (illustrations, fichiers annexes importés) sont sous Licence OpenEdition Books, sauf mention contraire.

Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search