Contributors: Hans Hooyberghs (VITO), Julie Berckmans (VITO), Filip Lefebre (VITO), Koen De Ridder (VITO)
1. Overview
Users and stakeholders managing urban areas are particularly interested in the spatial pattern of heat exposure within a city, as such datasets can be used to identify cool islands and hot spots on an urban scale. In dialogue with the user community involved in the SIS health project, it has therefore been decided to produce high resolution (100 metres) current-climate urban climate data for 100 cities across Europe.
This report provides an overview of the urban climate indicators that have been developed within the scope of the SIS health project (C3S_422_Lot2_VITO). The next paragraphs provide an overview of the urban climate data composed in this SIS. We focus on the methodology (paragraph 2), provide details on the actual data and the Essential Climate Variables (ECVs) considered (paragraph 3), and visualize some example data (paragraph 4).
2. Urban climate data
2.1. Methodology
Urban-scale temperature data are produced using the urban climate model UrbClim, a model designed to simulate and study the urban heat island effect (UHI) and other urban climate variables (wind speed, humidity, …) at a spatial resolution of 100 metres (De Ridder et al., 2015). The model scales large-scale weather conditions down to agglomeration-scale and computes the impact of urban development on the most important weather parameters, such as temperature and humidity. UrbClim is composed of a land surface scheme describing the physics of energy and water exchange between the soil and the atmosphere in the city, coupled to a 3D atmospheric boundary layer module. The atmospheric conditions far away from the city centre are fixed by meteorological input data, while local terrain and surface data influence the heat fluxes and evaporation within the urban boundaries. The primary output consists of hourly air temperature and humidity maps with a spatial resolution of 100 metres. The simulated domain is dependent on the city under investigation, but it is always chosen such that it comprises the city and its immediate surroundings. The main advantage of the urban boundary layer climate model is that it is faster than high- resolution mesoscale climate models by at least two orders of magnitude, while it has a similar level of accuracy. Because of that, the model is well suited for long time integrations, in particular for applications in urban climate (adaptation) projections, and for a deployment involving a large number of cities. Note that a comprehensive description of the UrbClim model is provided in (De Ridder et al., 2015).
The UrbClim model has been subjected to exhaustive validation. Model results have been compared with hourly temperature measurements for London, Bilbao, Athens, Almada, Toulouse, Berlin, Antwerp, Ghent, Brussels, Augsburg and Paris (De Ridder et al., 2013, 2015; Kourtidis et al., 2015; Lauwaet et al., 2016; Sarkar & De Ridder, 2011; Verdonck et al., 2018; Zhou et al., 2016).
2.2. Input data
The model requires two types of input data: large-scale meteorological data (typically global reanalysis datasets), and a description of the terrain (land use, vegetation…) in the city. In the next paragraphs we focus on both.
Meteorological input
For the current set-up of the model, the input stems from the ERA5 reanalysis dataset. UrbClim requires several input variables of different natures, including surface variables and variables defined at pressure levels, see the list defined in Table 1. The data has been downloaded using the ECMWF-API for the period 2008 – 2017, and processed for inclusion in the UrbClim model by off- line tools.
Table 1: Meteorological input parameters of the UrbClim model.
Type | Variable | Time increment | Category |
Surface | Latent heat flux | hourly | Surface |
Surface upward sensible heat flux | hourly | ||
Downwelling short-wave radiation | hourly | ||
Downwelling long-wave radiation | hourly | ||
Surface air pressure | hourly | ||
Air temperature at surface | hourly | ||
U component of wind velocity at surface | hourly | ||
V component of wind velocity at surface | hourly | ||
Specific humidity at surface | hourly | ||
Sea | Sea surface temperature | hourly | |
Precipitation | Precipitation | hourly | |
Convective precipitation | hourly | ||
Soil | Soil temperature | hourly | |
Soil moisture content | hourly | ||
Vertical profiles | U component of wind velocity | hourly | Pressure levels |
V component of wind velocity | hourly | ||
Temperature profile | hourly | ||
Specific humidity profile | hourly |
Terrain input
UrbClim requires several input datasets describing terrain-related input. The three most important ones (which describe the actual terrain lay-out of every grid cell) are:
- Land use type: the land use in every grid cell. UrbClim uses 15 different classes (among which 4 urban types), hence a conversion from the input dataset to these 15 classes is required
- Soil sealing: the amount of sealed soil in every grid cell.
- Vegetation: the amount of vegetation in every grid cell, modelled using the NDVI-index. The NDVI is assumed to vary between the months.
In addition, UrbClim also requires:
- A digital elevation model (to take into account height effects)
- Anthropogenic heat flux: the amount of heat produced by the city and its inhabitants (from traffic exhaust, air-conditioning devices…)
An overview of the land use input is provided in Table 2. All datasets have been downloaded, and resampled to the 100 m UrbClim-grid covering the city under consideration.
Table 2: Terrain input parameters of the UrbClim model.
Type | Source | Resolution | Link |
Land use | Copernicus (Corine land cover 2012) | 100 m | |
Soil sealing | Copernicus (Imperviousness 2012) | 100 m | |
Vegetation index (NDVI) | Copernicus (Proba V 2014 - 2017) | 300 m | |
Anthropogenic heat flux | AHF 2016 (see (Flanner 2009) for details) | 0.0416 | |
Digital elevation model (DEM) | U.S. Geological Survey (USGS) (Global Multi- resolution Terrain Elevation Data (GMTED) | 0.002083 |
3. Details of urban climate indicators
3.1. City selection
The 100 European cities for the urban simulations were selected based on user requirements within the health community. Furthermore, a high spatial distribution was aimed with specific focus on Eastern European countries that often lack access to relevant information (Table 3, Figure 1).
Table 3: Overview of selected European cities
Alicante | Edinburgh | Madrid | Rotterdam |
Amsterdam | Frankfurt am Main | Malaga | Sarajevo |
Antwerp | Gdansk | Marseille | Sevilla |
Athens | Geneva | Milan | Skopje |
Barcelona | Genoa | Miskolc | Sofia |
Bari | Ghent | Montpellier | Split |
Basel | Glasgow | Munich | Stockholm |
Belgrado | Goteborg | Murcia | Strasbourg |
Berlin | Graz | Nantes | Szeged |
Bilbao | Györ | Naples | Tallinn |
Birmingham | Hamburg | Newcastle | Tartu |
Bologna | Helsinki | Nice | Thessaloniki |
Bordeaux | Klaipeda | Novi Sad | Tirana |
Brasov | Kosice | Oslo | Toulouse |
Bratislava | Krakow | Padua | Trieste |
Brussels | Leeds | Palermo | Turin |
Bucharest | Leipzig | Palma de Mallorca | Utrecht |
Budapest | Liege | Paris | Valencia |
Charleroi | Lille | Pécs | Varna |
Cluj-Napoca | Lisbon | Podgorica | Vienna |
Cologne | Ljubljana | Porto | Vilnius |
Copenhagen | Lodz | Prague | Warsaw |
Debrecen | London | Reykjavik | Wroclaw |
Dublin | Luxembourg | Riga | Zagreb |
Dusseldorf | Lyon | Rome | Zurich |
The locations of the cities are shown on Figure 1.
Figure 1: Location of the cities considered for the urban use case. For visibility reasons, Reykjavik is not shown on the map.
3.2. ECVs
UrbClim provides output for several ECVs. The output for the four ECVs listed in Table 4 is ingested in the CDS.
Table 4: List of ECVs for the urban climate data
Standard name | Short name in CMIP5 | Units | Comments |
Near-surface air | tas | K | Reported at 2m |
Near surface relative humidity | russ | % | Reported at 2m |
Near surface specific | Huss | 1 | Reported at 2m |
Near surface wind speed | None | m/s | Reported at 2m, only scalar |
3.3. Auxiliary data
Two additional variables are provided to the dataset: a landseamask and (ii) a ruralurbanmask. They are derived from the CORINE land cover of 2012 (https://land.copernicus.eu/pan-european/corine-land-cover) and transformed to a mask that can be used for the application developed within SIS European Health. The landseamask presents a value of 1 for land surfaces and a NaN for water surfaces. The ruralurbanmask presents a value of 1 for rural surfaces and a NaN for urban surfaces. The latter is useful for calculating the Urban Heat Island.
3.4. Time format
Time Frame
For the current project, we have used the 10 year long time frame 2008 – 2017. This 10-year period has been selected based on data availability of the ERA5 data at the start of the simulations (June 2018), and computational constraints. Note that the first six hours of 2008 (2008-01-01T00:00 - 2008-01-05T00:00) are missing as the first ERA5 run of a year starts at 2008-01-01T06:00 UTC. These hours have been filled with NaN-values.
Frequency
We provide hourly output in UTC, in a format compatible with the common data format. The time format is thus unaffected by summer and winter time issues, and a shift is required to convert the data to local time.
3.5. Spatial extent
The size of the domain varies between the cities and is based on the size of the city (both the administrative boundaries and the true extent of the city). If possible, airports are included in the domain, as they usually coincide with the location of a temperature measurement station and could thus be used for validation. For each city, we have opted for a square domain. The centre and size of the domain are provided in Table 5.
UrbClim uses a projected coordinate system. For the current project, we have opted for the European Grid (EPSG:3035). To facilitate the ingestion of the data in the CDS, the final output is converted to a WGS84 (EPSG:4326) grid. Because of this, final maps use a skewed grid (as can be seen in the examples in Section 4).
Table 5: Details of the central coordinate and domain size of the cities
City | Longitude (degree) | Latitude (degree) | Domain size (km2) |
Alicante | -0.48 | 38.36 | 324 |
Amsterdam | 4.90 | 52.36 | 900 |
Antwerp | 4.40 | 51.25 | 900 |
Athens | 23.73 | 37.98 | 900 |
Barcelona | 2.13 | 41.38 | 625 |
Bari | 16.84 | 41.12 | 400 |
Basel | 7.59 | 47.56 | 225 |
Belgrado | 20.40 | 44.81 | 900 |
Berlin | 13.41 | 52.52 | 1600 |
Bilbao | -2.94 | 43.29 | 625 |
Birmingham | -1.89 | 52.48 | 900 |
Bologna | 11.34 | 44.50 | 400 |
Bordeaux | -0.59 | 44.83 | 625 |
Brasov | 25.60 | 45.66 | 225 |
Bratislava | 17.11 | 48.14 | 400 |
Brussels | 4.37 | 50.84 | 900 |
Bucharest | 26.09 | 44.45 | 900 |
Budapest | 19.13 | 47.50 | 900 |
Charleroi | 4.44 | 50.42 | 400 |
Cluj_Napoca | 23.61 | 46.78 | 324 |
Cologne | 6.96 | 50.95 | 900 |
Copenhagen | 12.55 | 55.67 | 400 |
Debrecen | 21.62 | 47.52 | 225 |
Dublin | -6.26 | 53.33 | 900 |
Dusseldorf | 6.81 | 51.23 | 784 |
Edinburgh | -3.21 | 55.95 | 625 |
Frankfurt_am_Main | 8.66 | 50.11 | 900 |
Gdansk | 18.63 | 54.38 | 625 |
Geneva | 6.13 | 46.21 | 100 |
Genoa | 8.89 | 44.43 | 625 |
Ghent | 3.72 | 51.08 | 625 |
Glasgow | -4.25 | 55.85 | 625 |
Goteborg | 11.98 | 57.71 | 625 |
Graz | 15.44 | 47.05 | 400 |
Gyor | 17.66 | 47.69 | 400 |
Hamburg | 9.99 | 53.54 | 900 |
Helsinki | 24.94 | 60.21 | 900 |
Klaipeda | 21.16 | 55.71 | 225 |
Kosice | 21.26 | 48.71 | 400 |
Krakow | 19.96 | 50.05 | 900 |
Leeds | -1.55 | 53.80 | 484 |
Leipzig | 12.36 | 51.34 | 625 |
Liege | 5.58 | 50.64 | 484 |
Lille | 3.06 | 50.63 | 324 |
Lisbon | -9.14 | 38.70 | 900 |
Ljubljana | 14.51 | 46.06 | 400 |
Lodz | 19.46 | 51.76 | 625 |
London | -0.13 | 51.51 | 1600 |
Luxembourg | 6.13 | 49.61 | 225 |
Lyon | 4.83 | 45.76 | 900 |
Madrid | -3.70 | 40.41 | 900 |
Malaga | -4.42 | 36.72 | 784 |
Marseille | 5.43 | 43.30 | 625 |
Milan | 9.19 | 45.46 | 400 |
Miskolc | 20.75 | 48.10 | 400 |
Montpellier | 3.88 | 43.62 | 400 |
Munich | 11.55 | 48.16 | 900 |
Murcia | -1.13 | 37.99 | 400 |
Nantes | -1.55 | 47.23 | 400 |
Naples | 14.26 | 40.85 | 400 |
Newcastle | -1.60 | 54.97 | 625 |
Nice | 7.26 | 43.71 | 400 |
Novi_Sad | 19.84 | 45.27 | 400 |
Oslo | 10.75 | 59.89 | 625 |
Padua | 11.88 | 45.42 | 400 |
Palermo | 13.35 | 38.13 | 400 |
Palma_de_Mallorca | 2.65 | 39.57 | 400 |
Paris | 2.35 | 48.86 | 1600 |
Pecs | 18.24 | 46.08 | 324 |
Podgorica | 19.26 | 42.42 | 324 |
Porto | -8.56 | 41.18 | 625 |
Prague | 14.42 | 50.06 | 900 |
Reykjavik | -21.83 | 64.12 | 225 |
Riga | 24.12 | 56.97 | 625 |
Rome | 12.48 | 41.86 | 900 |
Rotterdam | 4.45 | 51.90 | 625 |
Sarajevo | 18.36 | 43.85 | 400 |
Sevilla | -5.98 | 37.39 | 400 |
Skopje | 21.43 | 42.00 | 400 |
Sofia | 23.33 | 42.69 | 400 |
Split | 16.46 | 43.51 | 144 |
Stockholm | 18.07 | 59.33 | 625 |
Strasbourg | 7.74 | 48.58 | 400 |
Szeged | 20.15 | 46.26 | 400 |
Tallinn | 24.76 | 59.43 | 400 |
Tartu | 26.73 | 58.37 | 100 |
Thessaloniki | 22.95 | 40.63 | 400 |
Tirana | 19.79 | 41.34 | 400 |
Toulouse | 1.43 | 43.59 | 400 |
Trieste | 13.79 | 45.64 | 225 |
Turin | 7.68 | 45.07 | 400 |
Utrecht | 5.08 | 52.09 | 400 |
Valencia | -0.39 | 39.48 | 400 |
Varna | 27.93 | 43.21 | 400 |
Vienna | 16.41 | 48.20 | 900 |
Vilnius | 25.28 | 54.68 | 900 |
Warsaw | 21.05 | 52.23 | 900 |
Wroclaw | 17.03 | 51.10 | 900 |
Zagreb | 15.97 | 45.83 | 484 |
Zurich | 8.53 | 47.39 | 625 |
4. Example data
In this section, we provide some example output for four cities: Rome (Italy), Budapest (Hungary), Antwerp (Belgium) and London (United Kingdom). Figure 2 shows the mean daily minimal and maximal temperature for the entire period under consideration (January 2008 – December 2017). In Figure 3, we focus on the time profile of the monthly mean of the daily minimal rural and urban temperature. The rural temperature is thereby defined as the mean temperature over all the grid cells with a predominantly natural typology, while the urban temperature is defined as the mean temperature over all the grid cells with predominantly urban typology1. The land use classification is taken from the Copernicus Corine Land Cover. Note that both the maps and time series take height effects into account, which for instance explains the pattern in the western part of the Budapest domain.
Rome | |
Budapest | |
Antwerp | |
London |
Figure 2: Mean daily minimal and maximal temperature for the period January 2008 - December 2017 for four selected cities.
Rome | |
Budapest | |
Antwerp | |
London |
Figure 3: Time series of the monthly mean of the daily minimal rural and urban temperature for four selected cities. We refer to the main text for the exact definitions of the rural and urban temperature.
5. References
De Ridder, K., J. Angel Acero, D. Lauwaet, W. Lefebvre, B. Maiheu and M. Mendizabal (2013): RAMSES project report D4.1: Validation of agglomeration-scale climate projections. Retrieved from http://www.ramses-cities.eu/fileadmin/uploads/Deliverables_Uploaded/ramses_deliverable4.1_final.pdf
De Ridder, K., D. Lauwaet and B. Maiheu (2015): UrbClim – A fast urban boundary layer climate model, Urban Climate, Vol. 12, pp. 21–48. https://doi.org/10.1016/j.uclim.2015.01.001
Flanner, M. G. (2009): Integrating anthropogenic heat flux with global climate models, Geophysical Research Letters, Vol. 36, L02801, pp. 1-5
Kourtidis, K., A.K. Georgoulias, S. Rapsomanikis, V. Amiridis, I. Keramitsoglou, H. Hooyberghs, B. Maiheu and D. Melas (2015): A study of the hourly variability of the urban heat island effect in the Greater Athens Area during summer, Science of the Total Environment, Vol. 517, pp. 162– 177.
Lauwaet, D., K. De Ridder, S. Saeed, E. Brisson, F. Chatterjee, N.P.M. van Lipzig, B. Maiheu and H. Hooyberghs (2016): Assessing the current and future urban heat island of Brussels. Urban Climate, Vol. 15, pp. 1–15. https://doi.org/10.1016/j.uclim.2015.11.008
Sarkar, A., and K. De Ridder (2011): The Urban Heat Island Intensity of Paris: A Case Study Based on a Simple Urban Surface Parametrization, Boundary-Layer Meteorology, Vol. 138 (3), pp. 511–520. https://doi.org/10.1007/s10546-010-9568-y
Verdonck, M.-L., M. Demuzere, H. Hooyberghs, C. Beck, J. Cyrys, A. Schneider, R. Dewulf and F. Van Coillie (2018): The potential of local climate zones maps as a heat stress assessment tool, supported by simulated air temperature data, Landscape and Urban Planning, Vol. 178, pp.
183–197. https://doi.org/10.1016/J.LANDURBPLAN.2018.06.004
Zhou, B., D. Lauwaet, H. Hooyberghs, K. De Ridder, J.P. Kropp, D. Rybski (2016): Assessing Seasonality in the Surface Urban Heat Island of London, Journal of Applied Meteorology and Climatology, Vol. 55, 493-505.