Abstract. A description of the new air quality downscaling model – the urban EMEP (uEMEP) and its combination with the EMEP MSC-W model (European Monitoring and Evaluation Programme Meteorological Synthesising Centre West) – is presented. uEMEP is based on well-known Gaussian modelling principles. The uniqueness of the system is in its combination with the EMEP MSC-W model and the “local fraction” calculation contained within it. This allows the uEMEP model to be imbedded in the EMEP MSC-W model and downscaling can be carried out anywhere within the EMEP model domain, without any double counting of emissions, if appropriate proxy data are available that describe the spatial distribution of the emissions. This makes the model suitable for high-resolution calculations, down to 50 m, over entire countries. An example application, the Norwegian air quality forecasting and assessment system, is described where the entire country is modelled at a resolution of between 250 and 50 m. The model is validated against all available monitoring data, including traffic sites, in Norway. The results of the validation show good results for NO2, which has the best known emissions, and moderately good for PM10 and PM2.5. In Norway, the largest contributor to PM, even in cities, is long-range transport followed by road dust and domestic heating emissions. These contributors to PM are more difficult to quantify than NOx exhaust emission from traffic, which is the major contributor to NO2 concentrations. In addition to the validation results, a number of verification and sensitivity results are summarised. One verification showed that single annual mean calculations with a rotationally symmetric dispersion kernel give very similar results to the average of an entire year of hourly calculations, reducing the runtime for annual means by 4 orders of magnitude. The uEMEP model, in combination with EMEP MSC-W model, provides a new tool for assessing local-scale concentrations and exposure over large regions in a consistent and homogenous way and is suitable for large-scale policy applications.

Funding informationDirection Générale de l'Armement (France), Meteomodem and the Netherlands Organization for Scientific Research (contract 863.10.010 and project number SH-312-15), Physical processes relevant for the dissipation of thick, continental fog after sunrise are studied through observations from the SIRTA observatory and idealized sensitivity studies with the large-eddy simulation model DALES. Observations of 250 fog events over 7 years show that more than half of the fog dissipations after sunrise are transitions to stratus lasting 2 hr or more. From the simulations, we quantify the contribution of each process to the liquid water path (LWP) budget of the fog. Radiative cooling is the main source of LWP, while surface turbulent heat fluxes are the most important process contributing to loss of LWP, followed by the absorption of solar radiation, the mixing with unsaturated air at the fog top and the deposition of cloud droplets. The loss of LWP by surface heat fluxes is very sensitive to the Bowen ratio, which is importantly affected by the availability of liquid water on the surface; in a run without liquid on the surface, fog dissipation occurred 85 min earlier than in the Baseline simulation. The variability of stratification and humidity above fog top is documented by 47 radiosondes and cloud radar. Using DALES, we find that the variability in stratification has an important impact on the entrainment velocity; a three times more rapid fog-top entrainment enables the cloud base to lift from the ground 90 min earlier in weak stratification than in strong stratification in the model. With relatively dry overlying air, the fog evaporates faster than if the air is near saturation, leading to 70 min earlier dissipation in our simulations. Continuous observations of the temperature and humidity profiles of the layer overlying the fog could therefore be useful for understanding and anticipating fog dissipation.

Abstract. Radiative cooling and heating impact the liquid water balance of fog and therefore play an important role in determining their persistence or dissipation. We demonstrate that a quantitative analysis of the radiation-driven condensation and evaporation is possible in real time using groundbased remote sensing observations (cloud radar, ceilometer, microwave radiometer). Seven continental fog events in midlatitude winter are studied, and the radiative processes are further explored through sensitivity studies. The longwave (LW) radiative cooling of the fog is able to produce 40-70 g m −2 h −1 of liquid water by condensation when the fog liquid water path exceeds 30 g m −2 and there are no clouds above the fog, which corresponds to renewing the fog water in 0.5-2 h. The variability is related to fog temperature and atmospheric humidity, with warmer fog below a drier atmosphere producing more liquid water. The appearance of a cloud layer above the fog strongly reduces the LW cooling relative to a situation with no cloud above; the effect is strongest for a low cloud, when the reduction can reach 100 %. Consequently, the appearance of clouds above will perturb the liquid water balance in the fog and may therefore induce fog dissipation. Shortwave (SW) radiative heating by absorption by fog droplets is smaller than the LW cooling, but it can contribute significantly, inducing 10-15 g m −2 h −1 of evaporation in thick fog at (winter) midday. The absorption of SW radiation by unactivated aerosols inside the fog is likely less than 30 % of the SW absorption by the water droplets, in most cases. However, the aerosols may contribute more significantly if the air mass contains a high concentration of absorbing aerosols. The absorbed radiation at the surface can reach 40-120 W m −2 during the daytime depending on the fog thickness. As in situ measurements indicate that 20-40 % of this energy is transferred to the fog as sensible heat, this surface absorption can contribute significantly to heating and evaporation of the fog, up to 30 g m −2 h −1 for thin fog, even without correcting for the typical underestimation of turbulent heat fluxes by the eddy covariance method. Since the radiative processes depend mainly on the profiles of temperature, humidity and clouds, the results of this paper are not site specific and can be generalised to fog under different dynamic conditions and formation mechanisms, and the methodology should be applicable to warmer and moister climates as well. The retrieval of approximate emissivity of clouds above fog from cloud radar should be further developed.