The Oslo case study exploits the information provided by the TROPOMI instrument on Sentinel-5P (Schneider et al., 2021) among other things to improve the EPISODE urban-scale dispersion model (Hamer et al., 2020). To do so we first built an observation operator than can translate the model output into observation space, i.e. makes it comparable to the TROPOMI column observations (Figure 1). This capability is then used to calculate the biases between TROPOMI NO2 columns and modeled NO2 columns throughout the year (Figure 2). The result indicate that the biases between satellite and model correspond very well to the biases between model and measurements at air quality monitoring stations (Figure 3). As such, the satellite-derived model biases can be used to modify the underlying emissions and the model can subsequently be re-run using updated emissions. Comparing the updated model output to reference measurements indicates that the inclusion of S5P/TROPOMI-corrected emissions results in up to 20% higher accuracy of the model throughout the year (Figure 4). This “calibrated” model can subsequently be used for integrating surface observations from monitoring stations and low-cost sensor systems using data assimilation techniques (Figure 5) (Lahoz et al., 2014; Schneider et al., 2017; Mijling, 2020).

In addition, geostatistical downscaling (Stebel et al., 2021) has been used increase the spatial resolution of surface NO2 fields derived from S5/TROPOMI data, in order to make the product more useful for urban-scale monitoring (Figure 6).

This image has an empty alt attribute; its file name is image.png
Figure 1. Comparison of the tropospheric NO2 column from S5P/TROPOMI (left panel) and the corresponding EPISODE NO2 column (right panel) over the Oslo study region, here shown for 11 March 2019 at 11:25 UTC. Units given in 1015 molec./cm2. The label printed in each cell gives the exact NO2 column value.
This image has an empty alt attribute; its file name is image-2.png
Figure 2. Temporal evolution of the absolute difference between S5P NO 2 TVCD and EPISODE NO2 TVCD. The size of each marker indicates the number of individual retrievals (pixel) going into the calculation of each data point. The colour of the markers shows the average QA value of all the retrievals going on into the daily aggregation. The orange line shows a Loess fit to the data.
This image has an empty alt attribute; its file name is image-1.png
Figure 3. Scatter plot of relative average monthly bias of the EPISODE model versus air quality monitoring stations against the relative average monthly bias of EPISODE against S5P/TROPOMI. This shows that the model biases indicated by the satellite instrument agree well with the true model biases indicated by reference equipment.
This image has an empty alt attribute; its file name is episode_station_validation_emissions_orig_vs_global_scaled_monthly.png
Figure 4. Change in EPISODE performance measured against seven air quality monitoring stations equipped with reference instrumentation. Shown are a) mean bias b) RMSE and c) correlation (Pearson r) between model output and reference station for original emissions (red) and emissions scaled at a monthly level using factors derived from S5P/TROPOMI (blue).
This image has an empty alt attribute; its file name is image-3.png
Figure 5. Assimilation of reference station NO2 data into the S5P/TROPOMI-calibrated EPISODE model for a central part of Oslo. The original model output (“background”) is shown in the top panel (with the station observations as circular markers), whereas the result of the assimilation (“analysis”) can be seen in the bottom panel.
This image has an empty alt attribute; its file name is image-4.png
Figure 6. S5P/TROPOMI-derived surface NO2 over the Oslo study site for 2019-10-30 at original Level-2 pixel geometry (left panel) and the same data downscaled to 1 km spatial resolution using geostatistics (right panel).


Hamer, P. D., Walker, S.-E., Sousa-Santos, G., Vogt, M., Vo-Thanh, D., Lopez-Aparicio, S., Schneider, P., Ramacher, M. O. P., and Karl, M.: The urban dispersion model EPISODE v10.0 – Part 1: An Eulerian and sub-grid-scale air quality model and its application in Nordic winter conditions, Geosci. Model Dev., 13, 4323–4353,, 2020.

Lahoz, W. A., & Schneider, P. (2014). Data assimilation: Making sense of Earth Observation. Frontiers in Environmental Science, 2, 16.

Mijling, B. (2020). High-resolution mapping of urban air quality with heterogeneous observations: A new methodology and its application to Amsterdam. Atmospheric Measurement Techniques, 13(8), 4601–4617.

Schneider, P., Castell, N., Vogt, M., Dauge, F. R., Lahoz, W. A., & Bartonova, A. (2017). Mapping urban air quality in near real-time using observations from low-cost sensors and model information. Environment International, 106, 234–247.

Schneider, P., Hamer, P. D., Kylling, A., Shetty, S., & Stebel, K. (2021). Spatiotemporal Patterns in Data Availability of the Sentinel-5P NO2 Product over Urban Areas in Norway. Remote Sensing, 13(11), 2095.

Stebel, K., Stachlewska, I. S., Nemuc, A., Horálek, J., Schneider, P., Ajtai, N., Diamandi, A., Benešová, N., Boldeanu, M., Botezan, C., Marková, J., Dumitrache, R., Iriza-Burcă, A., Juras, R., Nicolae, D., Nicolae, V., Novotný, P., Ștefănie, H., Vaněk, L., … Zehner, C. (2021). SAMIRA-SAtellite Based Monitoring Initiative for Regional Air Quality. Remote Sensing, 13(11), 2219.