Integrated Urban Services for European cities: the Stockholm case

Accelerating urban population growth, especially in developing countries, has become a driving force of human development. Hydrometeorological events, climate change and air pollution have an increasingly significant impact on crowded, densely populated cities. In addition, the complexity and interdependence of urban systems augment the vulnerability of cities. A single extreme event can lead to a widespread breakdown of infrastructure, often through domino effects.

Many organizations, including WMO, recognize that rapid urbanization necessitates new types of services that make best use of science and technology. Such Integrated Urban Weather, Environment and Climate Services should assist cities with planning and in facing hazards like storm surges, floods, heatwaves and air pollution episodes, especially in changing climates. The aim is to develop urban services that meet the special needs of cities through a combination of dense observation networks, high-resolution forecasts, multi-hazard early warning systems and long-term urban climate projections at sub-urban scales for the design and planning of sustainable and resilient cities. Several recent international studies have been initiated to explore these issues.

In June, the WMO Executive Council adopted a conceptual and methodological approach in the Guide for Urban Integrated Hydro-meteorological, Climate and Environmental Services. However, many cities – for example, Hong Kong, Paris, Shanghai, Singapore and Toronto – have already started the implementation of Integrated Urban Services. Urban service requirements are city specific, and their configuration follows local stakeholder needs. Stockholm provides one of the best examples of such a service in Europe, with a focus on urban planning for the creation of an attractive and healthy city environment for future citizens.


Background and concept

Stockholm’s ambition is to be one of Europe’s leading green cities. It was selected as Europe’s first green capital in 2010 and is now working to be free of fossil fuels by 2040. Stockholm is a growing city with an official target to build 140 000 new homes by 2030. This will require significant changes in the urban infrastructure. A first cooperation on urban services between Stockholm City and the Swedish Meteorological and Hydrological Institute (SMHI) was achieved from 2010–2012 within the FP7 (the European Commission (EC) 7th Framework Programme for Research and Technological Development) project SUDPLAN (

SMHI later devised a Sectoral Information System named UrbanSIS, targeting infrastructure and health sectors operating in cities, as part of a Copernicus proof-of-concept project from 2015–2018. The objective was to develop, demonstrate and put into production a method to downscale to the urban level (1 × 1 km2) a set of atmospheric essential climate variables (ECVs). This was to enable calculation of impact indicators related to flooding events, heatwaves and air pollution episodes, which are of prime importance for end users in Stockholm (Table 1).  The cities of Bologna (Italy) and Rotterdam (Netherlands) were also participants.

The information was provided in three datasets, each one based on five years of hourly gridded 1 × 1 km2 data, representing:

  • A historical period of specific years: 2006, 2007, 2012, 2013, 2014
  • Five years of data taken from a climate scenario representing present conditions (1980–2010)
  • Five years of data taken from a climate scenario representing future conditions (2030–2065)

The quality of the downscaling was evaluated against observations. The two datasets representing present and future conditions were created assuming strong forcing (IPCC´s scenario RCP8.5). Note that present and future datasets should be interpreted as “representative” and not “true historical” years. A novelty of producing the gridded 1 × 1 km2 data was the use of a spatially high-resolution numerical weather prediction (NWP) model (1 km grid resolution) integrated for long periods of time (years).


Stakeholders in Stockholm and their requirements

UrbanSIS output
Table 1. UrbanSIS output in the form of 26 ECVs and 65 sectoral impact indicators. A full description of the data produced is given at

Workshops and interviews with stakeholders and future users of the UrbanSIS data were held in the early phases of the project, which were followed up in other projects (Swedish projects: HazardSupport and MUMS, and the EC’s Horizon 2020 project Clarity). Governmental agencies (Swedish Contingencies Agency, Swedish Transport Administration, National Board of Housing, Building and Planning, Stockholm County Administrative Board and Public Health Agency of Sweden), associations (Swedish Water and Wastewater Association), a private insurance company (Länsförsäkringar), local stakeholders (Stockholm City and Stockholm Vatten och Avfall) and consultancies (WSP, Tyréns and SWECO) were among the end users and stakeholders.

The definition of which ECVs and impact indicators to deliver on the urban scale was discussed and concluded through further workshops and interviews. The participants were urban climate and health experts from the University of Reading (United Kingdom), the University of Umeå (Sweden) and SMHI, as well as traditional urban end users representing consultants and city authorities from Stockholm and Bologna , another demonstration city. It was clear that there was a need for information representing the conditions of today, as well as information of what is likely to happen in the future, from the end-user viewpoint. Another finding of the dialogue was that some of the impact indicators should have city-specific characteristics.

The stakeholder input resulted in a portal with the large number of ECVs and indicators (Table 1). These data are delivered in a detailed way to technically advanced end users, such as consultants, urban engineers/scientists and health experts, to be used as input to specific local impact models. They are also in a format that can be directly used by urban planners for dimensional planning. Special effort was made to generate information that can be used for assessing and planning adaptation to urban hazards such as intense rainfall, heatwaves and extreme air pollution episodes.


Dynamic downscaling approach

General flow chart UrbanSIS
Figure 1. General flow chart representing the dynamical downscaling approach applied in UrbanSIS

The downscaling modelling chain consists of three numerical models as depicted in Figure 1. The meteorological/climate state was generated by the HARMONIE model (in two settings, as described below, depending on the period) and provided to the air quality model MATCH and the hydrological model HYPE.

For the historical period, lateral boundary data were provided to the NWP system HARMONIE-AROME (cycle 40h1.1) using UERRA-ALADIN reanalysis and surface observations. For present and future conditions, lateral and surface boundary data for the climate setting HCLIM-AROME were derived from the GLOBAQUA project.

High-resolution physiography data were generated by processing different open-access databases and products: (i) spatial coverage of land-cover types from Urban Atlas 2012 (Copernicus Land Monitoring Services), (ii) building polygons from OpenStreetMap, (iii) building heights from Lidar measurements (Swedish Forest Agency) and (iv) time series of leaf area indices from the Copernicus Global Land Service. The resulting 300 × 300 m2 grids were then interpolated by the surface/atmosphere scheme SURFEX to the final model grid at 1 × 1 km2 and combined with the default European ecosystem classification and surface parameters dataset ECOCLIMAP-II. Details of the model set-up and validation are given at

Image of the UrbanSIS portal mapping the UHI
Figure 2. Image of the UrbanSIS portal mapping the UHI over Stockholm in July 2014 and the time series of monthly mean T2m (in K) over two locations


Climate scenarios

The computational costs for high-resolution dynamical downscaling are considerable; therefore climate scenario data delivered was only for selected five-year periods. Moreover, urban downscaling was affordable for only one climate scenario realization, raising important challenges on how to communicate its representativeness and uncertainties.

There were user requirements to have extreme scenarios for the future climate data, so it was decided to use the RCP8.5 scenario. Output from a regional model was taken from the FP7 GLOBAQUA project, which offered three-dimensional data at a spatial resolution of 20 × 20 km2 for the periods 1980–2010 and 2030–2065. Within each 30–35-year window and for each city, five representative years were selected that encompassed combinations of cold/wet, cold/dry, warm/wet, warm/dry and “normal” summer seasons. The chosen seasons were selected to have episodes of extreme events.


Examples of outputs

Spatiotemporal air temperature gradients in Stockholm
Mean daily profile of T2m
Figure 3. Mean daily profile of  T2m in the summer of the five-year historical period over nature and urban tiles representing, respectively, the Observatorielunden park in Stockholm and the built-up area surrounding it (PCIavg is the average cooling promoted by this 4-hectare green area in summer conditions)

In UrbanSIS, the aim was to understand how climate is affected by a city’s morphology, and how this affects human comfort and health, especially during hot periods. The high-resolution urban climate data provided for Stockholm revealed a characteristic thermal fingerprint. This was expressed in the form of the urban heat island (UHI) or inner-city gradients associated with, for example, park cool islands. Figure 2 provides an example of the spatial coverage of the city’s UHI, its intensity and how it evolves with time.  

Analysis of the interaction of Stockholm’s heterogeneous surface with the atmosphere revealed cooling induced by urban parks, as shown in Figure 3, with distinct diurnal and seasonal cycles.

High-resolution dynamical downscaling of intense rainfall

One key motivation for dynamical downscaling, that is of very high resolution, is the extreme small-scale variability of rainfall. UrbanSIS was able to simulate intense small-scale rainfall events in a realistic way. Figure 4a shows a radar image taken during one of the most intense small-scale rainfalls observed in the evaluation period. Figure 4b shows the simulated rainfall at almost the same time. A perfect match is not attainable because of the chaotic nature of rainfall generation and uncertainties in the radar data. However, the simulated rainfall generally has a similar spatial extent and structure, as well as similar peak intensities, to the actual rainfall. The UrbanSIS simulations are realistic in statistical terms. Figure 4c shows the type of depth–duration–frequency (DDF) statistics that are widely used in urban hydrological engineering. The simulated statistics match well with the observed statistics from station and radar data, especially for the shortest durations (<1 h).

The type of agreement shown in Figure 4 is not attainable in lower-resolution climate models. This suggests that future changes in local rainfall extremes are potentially more realistically estimated by UrbanSIS. Future projections in Stockholm indicate a stronger increase in local rainfall extremes than that estimated by lower-resolution climate models. This has strong implications for adapting cities and their water-related infrastructure to climate change.

Figure 4. Images of rainfall
Figure 4. Images of rainfall: (a) as observed by radar and (b) as simulated by UrbanSIS in the Stockholm domain around noon on 2013-06-09. (c) Observed and simulated 10-year DDF statistics for Stockholm.
Health risks of air pollution in the future

MATCH is a chemical transport model developed at SMHI. It has been used offline with the climate model in UrbanSIS to deliver gridded 1 × 1 km2 urban background concentrations of pollutants – nitrogen dioxide (NO2), ozone (O3), inhalable particulate matter (PM10) and fine particulate matter (PM2.5) – over Stockholm. The air quality simulations were performed in two steps, first on the European scale to obtain long-range contributions from outside the city and then as a nested high-resolution simulation over Stockholm.

The assumed emission changes from the present (1980–2010) to the future (2030–2065) for the pan-European simulations were taken from the ECLIPSE project (Figure 5a). Local emission development in Stockholm was provided by the municipality and represented the years of 2010 and 2030 (Figure 5b).

Figure 5. Total emissions
Figure 5. (a) Total emissions inside the pan-European domain, referring to 2010 (blue) and 2030 (orange) projected within the ECLIPSE project and (b) total emissions inside the Stockholm urban domain, referring to 2010 (blue) and 2030 (orange), as projected by the Stockholm municipality

Figure 6 shows the expected development of PM2.5 mean urban background concentrations from the present to the future. A general decrease of about 0.5 micrograms per cubic meter (µg m−3) from a present level of about 5 µg m−3 can be expected around the city, while the central parts will, due to local emission reductions, decrease by up to 1 µg m−3 from present levels of 7–8 µg m−3.

One air quality indicator is the estimated number of deaths in the age group 30+ due to long-term exposure to PM2.5. A relative risk factor of 1.062 per 10 µg m−3 was taken from the WHO HRAPIE project. UrbanSIS shows, for PM2.5 exposure over the Stockholm domain, 568 deaths per year during present conditions and 507 deaths for the future scenario.

This type of health indicator must be presented together with the assumptions made. In this case, the population size and the city design with residential areas remained constant from the present to the future. These assumptions are obviously not realistic. Reaction from stakeholders involved in urban planning indicated that more scenarios should be included, to isolate the impact of climate change or urban design on air quality. Such scenario assessments are being assessed for Stockholm within follow-up projects.

Figure 6. Annual mean PM2.5 concentrations over Stockholm
Figure 6. Annual mean PM2.5 concentrations over Stockholm for (a) present (~2010) and (b) future (~2030).


Guidance on uncertainties

As a guide for end users to judge the quality and uncertainties of UrbanSIS output, SMHI suggested a three-level quality colour scale:

  • Green = good quality: results can be used without considering certain limitations or restrictions (Go ahead!);
  • Yellow = medium quality: results are useful, but the user should be aware of certain limitations/restrictions (Caution!)
  • Red = poor quality: results can partly be useful, but the user must understand the limitations/restrictions (Warning!).

The quality scale is applied to three aspects:

  1. Downscaling model performance (Model)
  2. Determination of impact indicators (Indicator)
  3. Climate scenario uncertainties (Scenario).

The Model and Indicator aspects are classified individually for each city. For the Scenario” aspect, the ECVs are given the same classification for all European cities. (Further details are available at /C3S_D441.5.4.2_UrbanSIS_201711_Uncertainties_scalability_rev.pdf.)


Lessons learned and future research and development

The approach of combining dynamical downscaling for meteorology, air quality and hydrology can provide useful and consistent climatological impact indicators on the urban scale. Access to high-resolution physiography, local emission inventory and local water channelling data is vital. These can partly be retrieved from European services, such as Urban Atlas. However, consistency among different data sources, for the urban and regional scales, has to be taken into account. Daily, monthly and seasonal profiles of local emission of chemical species can be fine-tuned using sector-related proxies (for example, traffic). In-depth sensitivity studies are needed to transform regional-scale emission inventories to the local scale. Additionally, local observations for different indicators during the historical simulation period are beneficial to validate and increase confidence in the results.

The considerable computational costs somewhat limit the domain size and number of simulated years. The small domain size resulted in a strong precipitation bias due to spin-up problems, especially during strongly forced winter conditions. The known resolution dependency of the coarser forcing data on the spin-up was not confirmed in this modelling set-up. Further research is needed to better understand and reduce this issue.

For urban simulation, precipitation bias meant that regional-scale simulated precipitation needed to be included, especially for river flow and accumulated snow. The limitation in number of years and future scenarios required careful selection and assessment of how the selected years and scenarios represented the climatological distribution for different indicators for present-day climate and future projections, including the assessment of their uncertainties. Downscaling specific historical extreme events, as worst-case scenarios, was considered and how these events could be affected in a future climate scenario examined.

So far, the aim of this approach has been to support long-term urban planning. However, the approach can also be adapted for short-term forecasts of weather and air pollution and early warning. For example, it could be used to carry out urban downscaling with a standard NWP and for running air quality model with Copernicus Atmospheric Monitoring Service (CAMS) results on the boundaries.

As a continuation of UrbanSIS, SMHI is now working with the city of Stockholm on the simulation of urban climate and effects on human comfort for distinct urban development scenarios: (i) the development plan for 2030; (ii) a strong increase of green infrastructure (“green scenario”); and (iii) sprawling and densification (“grey scenario”). This effort has demonstrated that downscaling the larger-scale climate information over a city provides new insights for urban planning and development, including landscape architecture and the use of nature-based solutions. It also delivers innovative and efficient solutions for the adaptation of cities to climate change.


Moving towards Integrated Urban Services

The  WMO concept of the Urban Integrated Hydro-meteorological, Climate and Environmental Services and demonstrated European experience, such as that of Stockholm, provide the following recommendations for WMO Members (National Meteorological and Hydrological Services (NMHSs), first of all) and interested cities:

  • Do not wait for a disaster to occur - Integrated Urban Services are already assisting decision-makers and end users (existing well-functioning urban services can be used as templates for development)
  • NMHSs should contribute to the promotion, development and coordination of Integrated Urban Services, including knowledge transfer
  • Ensure that legal and institutional frameworks are in place that clearly define government agency interactions and responsibilities to enable creation and maintenance of integrated services
  • Engage with relevant stakeholders (agencies, the public, NMHSs, city government, private sector and businesses) from the beginning, including raising awareness and getting feedback
  • Conduct further research, including multidisciplinary cross-cutting studies, to develop urban service capabilities
  • Encourage NMHSs to facilitate wider accessibility of data via influencing ownership issues and technical support
  • Showcase demonstration projects on urban services



The concept of an Integrated Urban Hydro-meteorological Climate and Environmental Service was proffered by WMO to meet the future needs of its Members, especially in meeting the United Nations Sustainable Development Goals. UrbanSIS in Stockholm is an excellent demonstration of an initiative integrating the various scientific disciplines in an innovative holistic way. Weather, air quality and hydrological models are used to produce spatial (1 km) and temporal (15 minutes to 1 hour) high resolution data for state-of-the-art eco-centric city planning and design.

The WMO initiative was undertaken cooperatively and collaboratively with other cities – Bologna and Rotterdam – to efficiently develop and generalize its capability. WMO is following up on the Guide for Urban Integrated Hydro-meteorological, Climate and Environmental Services, Part 1: Concept and Methodology with additional demonstration city examples with greater hazardous, geographic and economic diversity.



Jorge H. Amorim (, Christian Asker, Danijel Belusic, Ana C. Carvalho, Magnuz Engardt (Present address: SLB Analys, Stockholm City), Lars Gidhagen, Yeshewatesfa Hundecha, Heiner Körnich, Petter Lind, Esbjörn Olsson, Jonas Olsson, David Segersson, Lena Strömbäck - Swedish Meteorological and Hydrological Institute (SMHI)

Paul Joe and Alexander Baklanov - WMO Secretariat, Research Department

Share this page