FMI-ENFUSER Modelling System

The FMI-ENFUSER (The Finnish Meteorological Institute's ENvironmental information FUsion SERvice, Johansson et al. 2022) is an urban scale dispersion modelling system (Gaussian puff & plume) that uses AQ-measurement -driven data assimilation. The novelty of the system is the continuous utilization of AQ measurement data facilitating a) real-time adaptation for better accuracy and b) longer-term learning mechanisms.

Inputs

The model has been coupled in real time with the regional chemical transport model (SILAM), a local AQ monitoring network, and several sources providing meteorological data (HIRLAM, ECMWF). In addition, the model uses high resolution Geographic Information System data (GIS) to describe the local environment with a 5m resolution, including a detailed description of the urban landscape, vegetation and ground elevation. OpenStreetMap (OSM) is used to describe the modelling area in detail. Existing available emission information can also be utilized in the modelling; as an example in Helsinki area, a residential heating inventory from local authorities is being used and local shipping emissions can be considered by accessing FMI-STEAM shipping emission data (Figure 1).

Figure 1: Illustration of FMI-ENFUSER modelling system, which utilizes recent and historical AQ measurements, meteorological data and regional air quality forecasts in the production of high-resolution output for the current and future air quality in urban locations. GIS-datasets and emission inventories together with archived concentration time series are used for the calibration of the model.

Outputs

The modelling system provides near real-time information on urban air quality (near surface concentrations at breathing height) with a resolution of 10m to 15m for the hourly concentrations of PM2.5, PM10, NO2 and O3. Additionally, based on these modelled concentrations the Air Quality Index (AQI) is also provided. Model predictions on pollutant species such as black carbon (BC), particle number concentration (PNC) and lung-deposited surface area (LDSA) may also be provided, depending on AQ measurement data availability.

Data and code availability

The model is operational in Helsinki and Turku areas. The results for Helsinki can be downloaded from the FMI open data service. They are also shown on the Helsinki Region Environmental Services website and on the info screens of local tram and metro lines. The data for Turku is continuously provided for the city’s use, but the data is not publicly available. For most modelling areas the modelled AQ data provided by the system is also made available on Amazon S3 cloud storage. Source code for the ENFUSER model is publicly available.

Figure 2: Example of air quality index visualizations provided for the public based on ENFUSER model data.

FMI-ENFUSER has been designed to be portable (e.g., by relying on global open access input) and can be setup in other regions outside of Finland. Abroad, the model has been utilized in China, (Nanjing, Langfang) and in India, Delhi.

Projects and previous work

The development history of FMI-ENFUSER dates to the year 2011. In EU/PESCaDO project a module for environmental information fusion was introduced. Previously, FMI-ENFUSER has been utilized in several projects including CLEEN-MMEA (in China), TEKES-INKA, TEKES-NAQT (in China) and in the CLIMOB project (in India). Commercial collaboration with Vaisala corporation has been conducted on several occasions.

In addition, we have participated in projects such as:

• UIA-HOPE (Healthy Outdoor Premises for Everyone, project no. UIA03-240)

• Smart & Clean HAQT (Helsinki Air Quality Test Bed)

• Business Finland CITYZER (Services for effective decision making and environmental resilience)

• The ACCC Flagship (The Atmosphere and Climate Competence Center, Academy grant no. 337552)

• The Urban Air Quality 2.0 project

• RESPONSE (EU, H2020)

• GIANT (Global trends in IAQ: Novel technologies, Competence and Business, Business Finland)

LES-assisted AQ modelling in Turku (RESPONSE-project)

In the RESPONSE project (funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 957751), ENFUSER was modified to utilize very high-resolution wind data produced by a large eddy simulation (LES) model. Operational usage of LES for wind prediction is currently not feasible due to the high computational burden of the models. For this reason, the building-resolving wind data at 4 meter and 8-meter accuracy was pre-calculated for different wind directions using the PALM model. These precalculated data are then utilized by ENFUSER to compute the desired dispersion characteristics in the area where the very high-resolution wind-data is available. Significant modifications to the ENFUSER code base were needed to accomplish this. Currently, the very high-resolution ENFUSER is in operational use in Turku for particulate matter together with the regular ENFUSER for other quantities. The use of building-resolving wind data improves the dispersion modelling especially within the densely built areas of the city.

Figure 3: Illustration of LES-assisted dispersion modelling near the center of Turku, Finland. Left: A snapshot of simulated ground level PM2.5 concentrations in the outer nest (8x8m). Right: corresponding concentration in the inner nest (4x4m). Black color corresponds to low concentrations, orange and yellow indicate high concentrations.

Use cases

In addition to the timely production of urban air quality forecasts and now-casts, there are other potential use-cases for the model, including:

• Air quality sensor benchmarking

  • The performance of AQ sensors can continuously be benchmarked against all other available measurement information during the data assimilation. Resulting information may be used to guide maintenance, and sensors can even be calibrated (online drift correction).

• Measurement site selection

  • FMI-ENFUSER can assist in the selection of optimal measurement locations for, e.g., complementary sensor network installation.

• Tool for decision making

  • Automatic and personalized alerts can be provided based on the current and forecasted air quality output. Informed decisions based on the data can be made and pre-emptive administrative actions can be carried out accordingly.

• Personal exposure estimation

  • The high-resolution modelling output can be used to predict personal exposure of citizens and provide tools to plan routes based on air quality (https://www.helsinki.fi/en/researchgroups/digital-geography-lab/green-paths).

• Averaged long term pollutant concentrations

  • The modelling system can be used to assess high resolution annual average concentrations. This information can be used to provide citizens with the tools to assess their personal long-term exposure in their location of interest.

Additional information

Johansson, L., Karppinen, A., Kurppa, M., Kousa, A., Niemi, J.V, Kukkonen, J., An operational urban air quality model ENFUSER, based on dispersion modelling and data assimilation, Environmental Modelling & Software, Volume 156, 105460, ISSN 1364-8152, doi:10.1016/j.envsoft.2022.105460, 2022.

Johansson, L., Epitropou, V., Karatzas, K., Karppinen, A., Wanner, L., Vrochidis, S., Bassoukos, A., Kukkonen, J. and Kompatsiaris I. Fusion of meteorological and air quality data extracted from the web for personalized environmental information services. Environmental Modelling & Software, Elsevier, 64, 143-155. 2015.

Johansson, L., Jalkanen, J.-P. and Kukkonen, J. Global assessment of shipping emissions in 2015 on a high spatial and temporal resolution. Atmospheric Environment, Vol. 167, p403-415, doi:10.1016/j.atmosenv.2017.08.042, 2017.

Kassandros, T., Bagkis, E., Johansson, L., Kontos, Y., Katsifarakis, K.L., Karppinen, A. and Karatzas, K., 2023. Machine learning-assisted dispersion modelling based on genetic algorithm-driven ensembles: An application for road dust in Helsinki. Atmospheric Environment, p.119818.