Fog and low-level clouds, namely stratus, are of major concern mainly due to theiradverse effects on transportation. Visibility is drastically reduced under theoccurrence of fog conditions, thus affecting aviation and road traffic. For example,fog causes many troubles to scheduled flights, like cancelations or delays, whilecreating dangerous driving conditions, thus having significant socioeconomic impacts.Moreover, apart from affecting transportation, fog and low-level clouds also affect theradiation budget, specifically at the Earth’s surface. For all these reasons, thesimulation and forecasting of fog is important, especially when no observational tools,e.g. radars, are available.Ioannina is a middle-sized city (~120,000 inhabitants) situated in the Epirusmountainous region in Northwestern Greece. The city is located on a plateau (basin)with an average altitude of 500 m, surrounded by high mountains with altitudes higherthan 1500m. The airport of Ioannina lies in an area of the plateau which experiences ahigh yearly number of fog events. This is mainly due to: (1) the presence of thenearby Pamvotis lake (area 23 km 2 , average depth 4 m, maximum depth 10 m), whichlocally enriches the overlying air masses with water vapour and (2) the specificgeographical and topographical characteristics of the area, which generally favourcalm (low wind speed) conditions speed and high relative humidity levels, as well asthe creation of temperature inversions. Moreover, due to the local topography andmeteorological conditions, the city frequently suffers from wintertime air pollutionepisodes (smog) due to extensive biomass burning for domestic heating activities.Despite the frequent occurrence of fog and the induced air traffic problems, there arenot available tools for forecasting fog locally. The present work aims to fill this gapby implementing a numerical model, specifically the parameterized fog modelPAFOG along with the spectral cloud microphysics model MIFOG. The two modelswill be operated and evaluated as to their ability to simulate fog under differentconditions.The fog models will be initialized with available data from local meteorologicalstations supplemented by vertically resolved reanalysis and satellite data, due to thelack of radiosondes in the study area. Local information on aerosol particles, acting asCCN, will be implemented as well, enabling to investigate their role for the formationof fog. The performance of the two models will be assessed through comparisons toavailable METARs from the Ioannina airport. This study is a first step towardsimplementing fog models for a routine fog forecasting at the city/airport.

The main goal of this article is to test the long-term performance of the three-dimensional non-local turbulence (NLT (3D)) parameterization scheme at different grid sizes in the so-called gray zone between classical mesoscale modeling (Delta x) several km) and large eddy simulations (LES: Delta x< several 100 m). For this, NLT3D has been implemented in the numerical weather prediction Icosahedral Nonhydrostatic model (ICON) of Deutscher Wetterdienst (DWD). Results are compared with a one-dimensional version of NLT (3D) (NLT (1D)) and with two operational turbulence schemes available in ICON. Comparisons with observations from radiosondes, the operational surface synoptic (SYNOP) station network, and RAdar-OnLine-ANeichung (RADOLAN) radar data of DWD indicate that all turbulence schemes investigated perform reasonably well. Nonetheless, a more detailed study of the model results reveals several interesting differences between the turbulence parameterizations to be discussed in detail. Median absolute errors (MAE) from point-to-point comparisons between numerical results and SYNOP observations tend to be smaller than those from comparisons with averaging simulated fields over an environment around each station location. This behavior indicates an information loss caused by the averagingprocess.Forthe2-mtemperature(T-2m)andthehourlyprecipitationsums(Pr-1h),MAEs decrease with decreasing grid sizes, thus suggesting an information gainfor finer grids. The nighttime MAEs ofT(2m)andPr(1h)obtained with NLT(3D )andNLT1Dare similar to or lower than those of the operational turbulence schemesof ICON. Moreover, during a shallow warm-air intrusion, NLT(1D)and especiallyNLT(3D)yield a more realistic representation of the horizontal structures ofPr1hand, during nighttime stable boundary-layer situations, alsoT(2m). Radiosondeprofiles of the potential temperature confirm a reasonable vertical mixing asobtained with NLT(3D)and NLT1D.

&lt;p&gt;Deep moist convection comes in many different forms and degrees of organization while producing a wide spectrum of severe weather. It is long known that orography can influence the appearance of deep moist convection. In this work, which is based on the bachelor thesis of the first author, we examine whether and to what extent orography influences the intensity of deep moist convection. Therefore, two case studies with typical single cell environments (weak shear and little synoptic forcing) are simulated using COSMO-DE, a numerical weather prediction model provided by the German Weather Service (Deutscher Wetterdienst, DWD). The simulations were performed with a grid spacing of 2.8 km with a temporal resolution of 15 min. In both cases deep moist convection occurred over western Germany, Belgium and The Netherlands over low mountain ranges (below 1000 m) and over adjacent lowlands. In order to investigate the influence of the orography on the simulated deep moist convective cells, maximum updraft speed and maximum precipitation rate over the cells&amp;#8217; lifetime was analyzed. The intensity of the convective cells located over the low mountain ranges is compared to that of the convective cells located over the lowlands. The results show that convective cells over the mountain ranges differ to convective cells over the lowlands. However, these differences are small, if an additional initiation mechanism on the larger scale such as a convergence line is in place. In contrast, major intensity differences occur if there is no such larger scale initiation mechanism. In this case, the deep moist convection is far more intense over the low mountain ranges compared to the lowlands, which provides an important result for nowcasting and forecasting of deep moist convection in these regions and to assess whether convective cells might reach severe weather criteria or not. In addition to these results of the first author&#039;s bachelor thesis, recent simulations performed with the Cloud Model 1 (CM1) of the National Center for Atmospheric Research (NCAR) are also presented. Idealized simulations of the two cases with a higher spatio-temporal resolution were examined with regard to updraft speeds and precipitation rates of the simulated cells.&lt;/p&gt;

An intercomparison between 10 single-column (SCM) and 5 large-eddy simulation (LES) models is presented for a radiation fog case study inspired by the Local and Non-local Fog Experiment (LANFEX) field campaign. Seven of the SCMs represent single-column equivalents of operational numerical weather prediction (NWP) models, whilst three are research-grade SCMs designed for fog simulation, and the LESs are designed to reproduce in the best manner currently possible the underlying physical processes governing fog formation. The LES model results are of variable quality and do not provide a consistent baseline against which to compare the NWP models, particularly under high aerosol or cloud droplet number concentration (CDNC) conditions. The main SCM bias appears to be toward the overdevelopment of fog, i.e. fog which is too thick, although the inter-model variability is large. In reality there is a subtle balance between water lost to the surface and water condensed into fog, and the ability of a model to accurately simulate this process strongly determines the quality of its forecast. Some NWP SCMs do not represent fundamental components of this process (e.g. cloud droplet sedimentation) and therefore are naturally hampered in their ability to deliver accurate simulations. Finally, we show that modelled fog development is as sensitive to the shape of the cloud droplet size distribution, a rarely studied or modified part of the microphysical parameterisation, as it is to the underlying aerosol or CDNC.

We present MISTRA-v9.0, a one-dimensional (1D) and box (0D) atmospheric chemistry model. The model includes a detailed particle description with regards to the microphysics, gas–particle interactions, and liquid-phase chemistry within particles. Version 9.0 (v9.0) is the first release of MISTRA as an open-source community model. A major review of the code has been performed along with this public version release to improve the user friendliness and platform independence of the model. The purpose of this public release is to maximise the benefit of MISTRA to the community by making the model freely available and easier to use and develop. This paper presents a thorough description of the model characteristics and components. We show some examples of simulations reproducing previous studies with MISTRA, finding that v9.0 is consistent with previous versions.

In numerical weather prediction (NWP) models atmospheric processes are usually split into grid scale and subgrid scale ones with the latter to be parameterized. Traditionally, all subgrid scale dynamics is classified as turbulence which is mainly active in mixing the planetary boundary layer and the coupling to the surface. Modeled turbulence is essential for the proper representation of the fluxes of tracers, energy and momentum in the lower atmosphere and, therefore, for the quality of weather forecasts.However, in modern numerical weather prediction models grid sizes keep on decreasing as affordable with increasing computational power. Thus, originally subgrid scale physical processes such as turbulence have become partially resolved on the model grid. After the classical K approaches, this first lead to the introduction of vertically non-local turbulence parameterization schemes to explicitly represent the vertical extent of the largest eddies in the PBL.Meanwhile also horizontal grid sizes in the order of 1 km have become small enough to also resolve at least the larger eddies extending over the whole PBL on the model grid. This is accounted for in the NLT3D scheme. It employs a transilient matrix formulation which has proven to be a flexible way to extent the non-local representation ofturbulent fluxes also to the horizontal direction. NLT3D has already been tested in the framework of the WRF model in several short term simulations over the North American prairie and Appalachian mountain range. It has now been implemented into the ICON model of Deutscher Wetterdienst to be analyzed in longer term simulations over Europe. Additionally, the impact of different grid size of the hosting model is analyzed over more inhomogeneous terrain.We will present results from ICON simulations including NLT3D and discuss comparisons with the results from the classical ICON model setup including its operational turbulence scheme. For validation also comparisons with observational data from field campaigns and with data of the operational station network of DWD will be shown.

With increasing resolution of numerical weather prediction (NWP) models, classical subgrid-scale processes become increasingly resolved on the model grid. In particular, turbulence in the planetary boundary layer (PBL) is vertically already partially resolved in contemporary models. For classical local PBL schemes, resulting up-gradient heat transports cannot be treated correctly. Thus, nonlocal turbulence schemes have been developed in the past. As the horizontal grid sizes of NWP models become smaller than a few kilometers, the large turbulence eddies in the PBL will also start to become partially resolved in the horizontal direction. A very flexible way to formulate nonlocal turbulent exchange is the transilient matrix method, which is used here to develop a new turbulence parameterization. The resulting NLT3D scheme applies transilient mixing matrices to subgrid-scale transports in all three dimensions. We compare results of WRF real-case simulations including our scheme, a classical local turbulence scheme (MYNN), and an existing nonlocal one-dimensional scheme (ACM2) with observations from field campaigns over homogeneous terrain (CASES-99) and complex terrain (CAPTEX). Over homogeneous terrain, all three schemes similarly well capture the observed surface fluxes and radiosonde profiles, whereas over complex terrain more differences become obvious. During a tracer release experiment (CAPTEX) over the Appalachian mountain region, the mixing and vertical extent of the PBL turn out to be decisive to reproduce the observed advection speed of the tracer-marked air mass. Deeper mixing not only accelerates surface winds but also enables tracer to travel faster at higher altitudes and then mix back to the ground. As results from a version of NLT3D with only standard horizontal Smagorinsky diffusion (NLT1D) demonstrate, simulating three-dimensional turbulence can be beneficial already at horizontal grid sizes of a few kilometers.

The spectral microphysics cloud scheme MiFog and the two-moment cloud scheme PaFog are implemented in a single-column model of the atmospheric boundary layer for the numerical simulation of a radiation fog event. Results of numerical sensitivity studies are presented where the influence of cloudiness on the evolution of the fog has been investigated. It is shown that the appearance of cloudiness in the morning hours triggers an efficient dissolution of the fog. During situations with weak turbulence, quasi-periodic oscillations of the liquid water content occur in the MiFog simulations. It turns out that these fluctuations are caused by the direct interaction of diffusional droplet growth and gravitational settling of the droplets. Comparison of the MiFog and the PaFog model runs reveals that most of the time PaFog is capable of simulating the bulk fog properties sufficiently well.

Abstract. We present MISTRA-v9.0, a one dimensional (1D) atmospheric chemistry model. The model includes a detailed particle description with regards to the microphysics, gas-particle interactions, and liquid phase chemistry within particles. Version 9.0 is the first release of MISTRA as an open-source community model. A major review of the code has been performed along with this public version release to improve the user-friendliness and platform-independence of the model. In the past 20 years, MISTRA has been used in over 25 studies to address a wide range of scientific questions. The purpose of this public release is to maximise the benefit of MISTRA to the community by making the model freely available and easier to use and develop. This paper presents a thorough description of the model characteristics and components. We show some examples of simulations reproducing previous studies with MISTRA, finding that version 9.0 is consistent with previous versions.

In this study, the Hybrid MAss flux Convection Scheme (HYMACS) is implemented in the ICOsahedral Non-hydrostatic (ICON) weather prediction model. In contrast to conventional convection parametrization schemes, the convective up- and downdraughts are solely treated as subgrid-scale processes in HYMACS, whereas the environmental subsidence is passed to the grid-scale dynamics of the hosting model. It is shown that the operational anisotropic divergence damping in ICON distorts the grid-scale dynamical response on the net mass transport parametrized by HYMACS. Thus, a revised numerical filter configuration is developed which focuses on both the compatibility to local mass sources (sinks) and the effective suppression of numerical modes inherent from the model's triangular grid. Evaluation of Jablonowski-Williamson dynamical core experiments reveal that the combination of an isotropic second-order divergence damping with a modified version of the fourth-order divergence damping outperforms against numerical filters based on diffusion. The obtained results are similar to the operational set-up indicating just a minor effect on the properties of the dynamical core. Moreover, a series of dry mass lifting experiments with the revised numerical filter confirms its compatability with HYMACS. The distortion of the grid-scale circulation is removed while gravity waves are still retained despite the potentially degenerative effect of the fourth-order divergence damping. Analyses of kinetic energy spectra confirm the effective suppression of checkerboard noise for a wide range of different situations. The present study may be understood as a base for future applications of HYMACS with a full cloud model in real-case studies.

A comparison between the spectral microphysics cloud scheme (MiStra) and the parametrized cloud scheme (PaStra) is presented. The main feature of MiStra consists of the treatment of aerosol particles, cloud droplets, and drizzle particles in a joint two-dimensional particle size distribution, whereas PaStra consists of a two-moment scheme for cloud droplets combined with a one-moment scheme for drizzle. Both cloud schemes have been implemented in a single-column model of the cloud-topped marine boundary layer. Numerical sensitivity studies are presented demonstrating that MiStra is capable of simulating in great detail the major cloud microphysical processes occurring in low-level stratiform clouds. While in MiStra no empirical parameter is available to tune the model, the empirical model parameters of PaStra have been tuned by means of the MiStra model results. By comparing the numerical results of PaStra with those of MiStra it is found that PaStra simulates the overall characteristics of the cloud-topped marine boundary layer quite well. At the same time, however, the effects of single cloud microphysical processes differ substantially from those of MiStra. Finally, it is shown that even in PaStra the inclusion of all major cloud microphysical processes is mandatory in order to obtain appropriate results.