Urban heat island intensity maps and local weather types description for a 45 French urban agglomerations dataset obtained from atmopsheric numerical simulations

This article presents a dataset of spatial nocturnal Urban Heat Island (UHI) intensities for 45 French urban agglomerations, at a horizontal resolution of 250 m. The urban influence on air temperature at 2 m above ground level was obtained by coupling the mesoscale atmospheric model Meso-NH with the land surface model SURFEX-TEB. For each agglomeration, two specific local weather situations that favour the development of a strong UHI in summer are simulated and described in a specfic sheet. Simulation outputs have been postprocessed to 1) identify the time of day when the UHI is the most developed, 2) to merge information from both meteorological situations in order to obtain one synthetic UHI map and 3) a geographical analysis that allows to classify each city among five spatial UHI classes (Concentrated Very High Intensity; Concentrated High Intensity; Limited Intensity; Dispersed High Intensity; and Dispersed Cool Zones). This dataset can therefore be used for several purposes, from the analysis at the scale of a city to the comparison of the urban agglomerations among them.


a b s t r a c t
This article presents a dataset of spatial nocturnal Urban Heat Island (UHI) intensities for 45 French urban agglomerations, at a horizontal resolution of 250 m.The urban influence on air temperature at 2 m above ground level was obtained by coupling the mesoscale atmospheric model Meso-NH with the land surface model SURFEX-TEB.For each agglomeration, two specific local weather situations that favour the development of a strong UHI in summer are simulated and described in a specfic sheet.Simulation outputs have been postprocessed to 1) identify the time of day when the UHI is the most developed, 2) to merge information from both meteorological situations in order to obtain one synthetic UHI map and 3) a geographical analysis that allows to classify each city among five spatial UHI classes (Concentrated Very High Intensity; Concentrated High Intensity; Limited Intensity; Dispersed High Intensity; and Dispersed Cool Zones).This dataset can therefore be used for several purposes, from the analysis at the scale of a city to the comparison of the urban agglomerations among them.The UHI raw data was generated using the non-hydrostatic mesoscale atmospheric model Meso-NH-SURFEX and is described in Gardes et al. 2020 [1] .The method to identify the LWT is described in Jougla et al. 2019 [2] .The LWT identification is integrated in the enriched metadata of each shapefile and the LWT description is presented in a separate .pdfdocument for each urban area.The geographical patch analysis is described in Suher-Carthy, 2021 [3] .Data source location Laboratoire Interdisciplinaire Solidarités Sociétés Territoires (LISST), Université de Toulouse, CNRS, 5 Allée Antonio Machado, 31058 Toulouse, France.Data accessibility https://doi.org/10.5281/zenodo.8009879

Value of the Data
• The dataset contains the nocturnal Urban Heat Island intensity in 45 urban agglomerations in metropolitan France and the description of the most current Local Weather Types.• This collection can be used to investigate the meteorological context and the microclimate of a city at a horizontal resolution of 250 m. • The gathered data can be integrated and/or combined with other climate, geographical, or social data.• The data provided can be used in cross scientific, technical or operational studies that aim to assess the social or environmental consequences of the UHI.

Objective
The data was generated within the framework of two French research projects.These are the ANR-MApUCE "Applied Modelling and Planning Laws: Climate and Energy" and the ADEME-PAENDORA "Planning, Adaptation and Energy: Territorial data and support" projects.The general objective of the projects was to integrate quantitative urban microclimate, climate, and energy data into the most relevant urban policies and legal French documents 1 .Within this framework, an urban (MApUCE database), an architectural (Danube database), and a meteorological database (Local Weather Type classification) were developed to perform the mesoscale climatic simulations here presented.

Data Description
This data paper provides maps of UHI intensity at a resolution of 250 m, for 45 French urban agglomerations (Fr: "Unité Urbaine").In the French context, urban agglomerations are defined as "a municipality or a set of municipalities with a continuous built-up area (no distance of more than 200 meters between two buildings) and at least 20 0 0 inhabitants" [4] .
In this data collection, each urban agglomeration has its own shapefile layer, whose attributes are organized as displayed in Table 1 .The urban agglomerations are named after the most important city within them Table 2 .

Experimental Design, Materials and Methods
Four main steps were necessary to generate the dataset.The first step is the identification and description of two Local Weather Types (LWT) that favour the UHI formation.The second step is to simulate six days during the summer season for each LWT using the mesoscale atmospheric model Meso-NH coupled to the land Surface Externalisée, in French, and Town Energy Balance (SURFEX-TEB) surface models.The third step is the identification of the three-hour period during the night with the highest UHI intensity and the averaging of this period, and the fourth step is the cluster classification of UHI spatial pattern ( Fig. 1 ).

First Step: Local Weather Types Favourable to a Strong Urban Heat Island Effect
The UHI intensity depends on the local meteorological conditions.In order to identify the local meteorological situations favourable to a strong UHI, for each urban unit, LWT are identified and described according to the method of Hidalgo et al. (2014) [5] and Hidalgo and Jougla (2018) [6] .The selected LWT are based on the PAM classification (Partitioning Around Medoids) that considers daily values of temperature, humidity, precipitation, and wind direction.This method was applied by Jougla et al. (2019) [2] to 50 French cities using a re-analysis conducted with the operational high-resolution (0.025 °horizontal resolution, that is about 2.5 km in France metropolitan) AROME model for the period 20 0 0 to 2009.The UHI data provided in this article is obtained from the combination of numerical simulations for six days and two LWT the most favourable for the development of a strong UHI during the summer season [1] .

Second Step: Numerical Simulations of the Nocturnal Urban Heat Island Intensity
For each selected LWT, numerical simulations are performed with the mesoscale atmospheric model Meso-NH [7] .It uses four horizontal domains with horizontal grid resolution of 8 km (D1), 2 km (D2), 1 km (D3), 250 m (D4).Their size and position are adapted to the urban agglomeration's specific spatial characteristics [1] .
Meso-NH is coupled with the urban canopy parameterization Town Energy Balance (TEB) [8] which simulates the urban surface energy balance according to the meteorological conditions simulated by Meso-NH.Vertical profiles of meteorological parameters in the urban canopy layer are calculated with the Surface Boundary Layer (SBL) parameterization [9] .The second level of the TEB-SBL is located in 2 m a.g.l.; the simulated air temperature from this level is used as urban air temperature (T2M URB ).Simulations are repeated by replacing the urban land cover with a rural land cover type that is characteristic of the city surroundings.An SBL scheme similar to the one of TEB is employed in the rural areas; the non-urban air temperature (T2M NOURB ) is derived from this simulation.The difference between the urban and non-urban simulation yields the UHI intensity.For the two selected LWT, the UHI of the day following the 6 LWT days is analysed since UHI depends on the weather conditions of the previous day [10] .

Third Step: Maximum Urban Heat Island Intensity
The raw data from the numerical simulations are in the form of a data frame containing grid point values for the 24 hours of the day.The three-hour periods with the maximum UHI intensity during both LWT are first identified.Secondly, the data is rasterized and the average over those 3 hours is calculated for each LWT.The final raster is obtained combining both rasters (maximum value per pixel) to reflect the one with highest exposure to UHI intensity [3] .

Fourth Step: UHI Cluster Typologies
The maximum UHI for the urban agglomerations is analyzed according to their intensity and spatial distribution -to identify common UHI spatial patterns that would allow them to be grouped into clusters.For this objective, the ArcGIS PRO Patch Analyst2 tool to identify shared typologies between cities, and cluster categorization was executed in Rstudio with the {factoex-tra} package and Partitioning Around Medoids method.At the end of the treatments, 5 categories of UHI typology were found: ( 1 ) Concentrated High Intensity, ( 2 ) Limited Intensity, ( 3 ) Dispersed High Intensity, ( 4 ) Dispersed Cool Zones, ( 5 ) Concentrated Very High Intensity, cluster specific to the Urban Unit of Paris [3] .The types of UHI intensity constituting these clusters can be understood and described through the UHI classes provided in Fig. 2 of the next section.
1.The cluster 1 concerns strong exposure concentrated at the urban cores generally.Often, strong exposure zones in highly compact forms are crossed by rivers, passing through the cores of the cities. 2. The cluster 2 illustrates significant exposure with limited expansion and is sometimes dispersed in the urban space.3. The cluster 3 includes strong exposure highly fragmented and dispersed in the place.In this cluster, strong UHI exposure class is dominant.4. The cluster 4 is marked by the presence of cool zone area that causes heterogeneous forms of UHI.All the urban agglomerations of this cluster are coastal and are located in southern France. 5.The cluster 5 is only connected to the agglomeration of Paris.This city is a cluster in itself because of these very strong UHI exposure.The large urban structures in this city make it impossible to compare it to other urban areas in France.

Methodological Summary
The methodology for obtaining the data previously described in 3.1 to 3.3 is summarized in Fig. 1 .Fig. 2 gives an example of a cartographic representation of the data provided, for the

Fig. 1 .
Fig. 1.The methodology employed in the French ANR-MAPuCE project to produce data on nocturnal UHI intensity in French cities.

Fig. 2 .
Fig. 2.An example of a UHI map of the urban agglomeration of Toulouse.
© 2023 The Author(s).Published by Elsevier Inc.This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ )

Table 1
Attribute structure of the provided Shapefile layer.

Table 2
Table of urban agglomerations names.