Horizontal Grid Length Research Articles

Machine learning bias correction and downscaling of urban heatwave temperature predictions from kilometre to hectometre scale

AbstractThe urban heat island (UHI) effect exacerbates near‐surface air temperature (T) extremes in cities, with negative impacts for human health, building energy consumption and infrastructure. Using conventional weather models, it is both difficult and computationally expensive to simulate the complex processes controlling neighbourhood‐scale variation of T. We use machine learning (ML) to bias correct and downscale T predictions made by the Met Office operational regional forecast model (UKV) to 100 m horizontal grid length over London, UK. A set of ML models (random forest, XGBoost, multiplayer perceptron) are trained using citizen weather station observations and UKV variables from eight heatwaves, along with high‐resolution land cover data. The ML models improve the T mean absolute error (MAE) by up to 0.12°C (11%) relative to the UKV. They also improve the UHI diurnal and spatial representation, reducing the UHI profile MAE from 0.64°C (UKV) to 0.15°C. A multiple linear regression performs almost as well as the ML models in terms of T MAE, but cannot match the UHI bias correction performance of the ML models, only reducing the UHI profile MAE to 0.49°C. UKV latent heat flux is found to be the most important predictor of T bias. It is demonstrated that including more heatwaves and observation sites in training would reduce overfitting and improve ML model performance.

Changes in freezing rain occurrence over eastern Canada using convection-permitting climate simulations

Freezing precipitation has major consequences for ground and air transportation, the health of citizens, and power networks. Previous studies using coarse resolution climate models have shown a northward migration of freezing rain in the future. Increased model resolution can better define local topography leading to improved representation of conditions that are favorable for freezing rain. The goal of this study is to examine the climatology and characteristics of future freezing rain events using very-high resolution climate simulations. Historical and pseudo-global warming simulations with a 4-km horizontal grid length were used and compared with available observations. Simulations revealed a northerly shift of freezing rain occurrence, and an increase in the winter. Freezing rain was still shown to occur in the Saint-Lawrence River Valley in a warmer climate, primarily due to stronger wind channeling. Up to 50% of the future freezing rain events also occurred in present day climate within 12 h of each other. In northern Maine, they are typically shorter than 6 h in current climate and longer than 6 h in warmer conditions due to the onset of precipitation during low-pressure systems occurrences. The occurrence of freezing rain also locally increases slightly north of Québec City in a warmer climate because of freezing rain that is produced by warm rain processes. Overall, the study shows that high-resolution regional climate simulations are needed to study freezing rain events in warmer climate conditions, because high horizontal resolutions better define small-scale topographic features and local physical mechanisms that have an influence on these events.

Numerical simulation of a violent supercell tornado over Vienna airport initialized and initiated with a cloud model

Supercell storm that occurred at the Vienna International Airport - “Schwechat” on July 10th, 2017, produced large hail and tornado with the strength of F2 on the Fujita scale, which corresponds to wind speeds between 118 km/h. and 180 km/h. (~33–50 m/s). Doppler Weather radar of the Austro Control Association has successfully captured the evolution of a supercell storm and the characteristic patterns representative for the appearance of tornado. The present study examines the main triggering factors responsible for initiation of supercell storm and the main ingredients for formation of tornado. The results from a number of sensitivity tests using the advanced research version of the Weather Research and Forecasting WRF-ARW model, indicate the existence of very unstable atmospheric conditions (e.g. large surface CAPE, enhanced wind shear and veering, high storm helicity index, differential heating induced by the strong local forcing environment) favorable for initiation and development of a tornadic supercell. A three-dimensional (3-D) cloud model simulation at a 500 m horizontal grid length quite well captured some important characteristics of a supercell storm dynamics (e.g., near surface convergence, strong rotating updrafts, vertical vorticity evolution and formation of mesocyclone) and cloud microphysics (evolution and transformation of a cloud hydrometeor fields, formation and fallout of precipitation). The radar reflectivity patterns identify a two downdraft cores, a hook-shaped echo at the rear side, and bounded weak echo region at a cloud mature phase. A tornado-like signature represents a quite useful ingredient for initializing and initiation of a tornado evolved from a supercell cloud at a small sub-domain (D2). The model runs at 50 m and 10 m horizontal and vertical grid spacing within a sub-domain (D3), respectively show evident advantage in simulation of a tornado-scale vortex, initiation, evolution and dissipation, relative to a coarser model runs. The sensitivity experiments using this novel technique revealed that such small-scale local hazards as tornadoes could be reproduced quite successfully with a close resemblance to observations. The initiation of a vortex tornado that evolves from a simulated supercell storm under open lateral boundary conditions allows for the selection of a smaller area and cloud simulation with a very fine resolution, capable of capturing smaller associated tornado vortices. The proposed method shows certain advantages over other models in terms of computing time savings and can serve as a reliable tool in various applications of short-term forecasting and local hazard warning.

Sensitivities of slantwise convection dynamics to model grid spacing under an idealized framework

AbstractAlthough the release of conditional symmetric instability (CSI) by slantwise convection is recognized as an important baroclinic process, the basic dynamics of these circulations and their representation in numerical models remain inadequately understood. To address this issue, a series of 2D idealized experiments of pure slantwise convection are performed in an initially statically stable environment using the non‐hydrostatic Weather Research and Forecasting model, with the horizontal grid lengths varying between 1 and 40 km. The results show that the larger‐scale feedbacks of the slantwise convection converge numerically when a cross‐band grid length (∆y) of 5 km is reached. The differences between the non‐converged and converged results tie closely to the release of a shallow layer of conditional instability that inevitably accompanies the early development of the slantwise circulation due to differential advection of saturation equivalent potential temperature (). The resolved small‐scale upright convection embedded within the slantwise band can energize the horizontal acceleration of the slantwise band at mid‐to‐upper levels by transporting low geostrophic momentum upward that results in localized inertial instability. The convective cell also enhances the large‐scale CSI neutralization by advecting high downward with strong downdraughts that orient more vertically than coarser‐gridded runs. Moreover, ∆y ≤ 5 km also better resolves the horizontal pressure gradients for cross‐band motions. This work suggests that global/climate numerical weather prediction models may not adequately resolve important characteristics of slantwise convection. As most cumulus schemes target only upright convection, the inclusion of parametrized slantwise convection may improve their performance.

A Physically Based Stochastic Boundary Layer Perturbation Scheme. Part I: Formulation and Evaluation in a Convection-Permitting Model

AbstractWe present a simple, physically consistent stochastic boundary layer scheme implemented in the Met Office’s Unified Model. It is expressed as temporally correlated multiplicative Poisson noise with a distribution that depends on physical scales. The distribution can be highly skewed at convection-permitting scales (horizontal grid lengths around 1 km) when temporal correlation is far more important than spatial. The scheme is evaluated using small ensemble forecasts of two case studies of severe convective storms over the United Kingdom. Perturbations are temporally correlated over an eddy-turnover time scale, and may be similar in magnitude to or larger than the mean boundary layer forcing. However, their mean is zero and hence they, in practice, they have very little impact on the energetics of the forecast, so overall domain-averaged precipitation, for example, is essentially unchanged. Differences between ensemble members grow; after around 12 h they appear to be roughly saturated; this represents the time scale to achieve a balance between addition of new perturbations, perturbation growth, and dissipation, not just saturation of initial perturbations. The scheme takes into account the area chosen to average over, and results are insensitive to this area at least where this remains within an order of magnitude of the grid scale.

Sub‐km scale numerical weather prediction model simulations of radiation fog

AbstractThe numerical weather prediction (NWP) of fog remains a challenge, with accurate forecasts relying on the representation of many interacting physical processes. The recent Local And Non‐local Fog EXperiment (LANFEX) has generated a detailed observational dataset, creating a unique opportunity to assess the NWP of fog events. We evaluate the performance of operational and research configurations of the Met Office Unified Model (MetUM) with three horizontal grid lengths, 1.5 km and 333 and 100 m, in simulating four LANFEX case studies. In general, the subkilometre (sub‐km) scale versions of MetUM are in better agreement with the observations; however, there are a number of systematic model deficiencies. MetUM produces valleys that are too warm and hills that are too cold, leading to valleys that do not have enough fog and hills that have too much. A large sensitivity to soil temperature was identified from a set of parametrisation sensitivity experiments. In all the case studies, the model erroneously transfers heat too readily through the soil to the surface, preventing fog formation. Sensitivity tests show that the specification of the soil thermal conductivity parametrisation can lead to up to a 5‐hr change in fog onset time. Overall, the sub‐km models demonstrate promise, but they have a high sensitivity to surface properties.

Convective initiation and storm life cycles in convection‐permitting simulations of the Met Office Unified Model over South Africa

Convective initiation is a challenge for convection‐permitting models due to its sensitivity to sub‐km processes. We evaluate the representation of convective storms and their initiation over South Africa during four summer months in Met Office Unified Model simulations at a 1.5‐km horizontal grid length. Storm size distributions from the model compare well with radar observations, but rainfall in the model is predominantly produced by large storms (50 km in diameter or larger) in the evening, whereas radar observations show that most rainfall occurs throughout the afternoon, from storms 10–50 km in diameter. In all months, the modelled maximum number of storm initiations occurs at least 2 hr prior to the radar‐observed maximum. However, the diurnal cycle of rainfall between the model and observations compares well, suggesting that the numerous storm initiations in the simulations do not produce much rainfall. Modelled storms are generally less intense than those in radar observations, especially in early summer. In February, when tropical influences dominate, the simulated storms are of similar intensity to observed storms. Simulated storms tend to reach their peak intensity in the first 15 min after initiation, then gradually become less intense as they grow. In radar observations, storms reach their peak intensity 15 min into their life cycle, stay intense as they grow larger, then gradually weaken after they have reached their maximum diameter. Two November case studies of severe convection are analysed in detail. A higher resolution grid length initiates convection slightly earlier (300 m as opposed to 1.5 km) with the same scientific settings. Two 1.5 km simulations that apply more subgrid mixing have delayed convective initiation. Analysis of soundings indicates little difference in the convective indices, suggesting that differences in convection may be attributed to the choice of subgrid mixing parameters.

Mesoscale Variation in Diabatic Heating around Sumatra, and Its Modulation with the Madden–Julian Oscillation

Abstract Diabatic heating in the Maritime Continent region is controlled by a unique blend of mesoscale variability associated with steep topography and complex coastlines and intraseasonal variability associated with propagating planetary-scale disturbances. In this study, the diabatic heating from a 10-yr austral summer simulation over the Maritime Continent with a 4-km horizontal grid length is analyzed with respect to diurnal, spatial, and intraseasonal variations. Results are compared, where possible, to analogous estimates from the TRMM precipitation radar. We show that the heating budget is largely a balance between latent heating and vertical advection, with rays of heating and cooling extending upward and outward from the coast evident in the advection terms, consistent with the gravity wave representation of the tropical sea breeze. By classifying rainfall into convective and stratiform components, it is shown that simulated convective heating over Sumatra peaks in MJO phases 2 and 3, while simulated stratiform heating peaks in phase 4. Similarly, spectral latent heating estimates from the TRMM Precipitation Radar show that stratiform heating peaks in phases 3 and 4, while convective heating peaks in phases 2 and 3. It is also shown that stratiform precipitation plays a greater role in offshore precipitation during the night, albeit with embedded convective cores, than over the land during the day. These results emphasize the importance of achieving a realistic representation of convective and stratiform processes in high-resolution simulations in the tropics, both for total rainfall estimates and for realistic latent heating.

Quantifying the Effects of Horizontal Grid Length and Parameterized Convection on the Degree of Convective Organization Using a Metric of the Potential for Convective Interaction

Abstract The organization of deep convection and its misrepresentation in many global models is the focus of much current interest. A new method is presented for quantifying convective organization based on the identification of convective objects and subsequent derivation of object numbers, areas, and separation distances to describe the degree of convective organization. These parameters are combined into a “convection organization potential” based on the physical principle of an interaction potential between pairs of convective objects. This technique is applied to simulated and observed fields of outgoing longwave radiation (OLR) over the West African monsoon region using data from Met Office Unified Model simulations and satellite observations made by the Geostationary Earth Radiation Budget (GERB) instrument. The method is evaluated by using it to quantify differences between models with different horizontal grid lengths and representations of convection. Distributions of OLR, precipitation and organization parameters, the diurnal cycle of convection, and relationships between the meteorology in different states of organization are compared. Switching from a configuration with parameterized convection to one that allows the model to resolve convective processes at the model grid scale is the leading-order factor improving some aspects of model performance, while increased model resolution is the dominant factor for others. However, no single model configuration performs best compared to observations, indicating underlying deficiencies in both model scaling and process understanding.

Estimating collision–coalescence rates from in situ observations of marine stratocumulus

Precipitation forms in warm clouds via collision–coalescence. This process is difficult to observe directly in situ and its implementation in numerical models is uncertain. We use aircraft observations of the drop‐size distribution (DSD) near marine stratocumulus tops to estimate collision–coalescence rates. Marine stratocumulus is a useful system to study collisional growth because it is initiated near the cloud top and the clouds evolve slowly enough to obtain statistically useful data from aircraft. We compare rate constants estimated from observations with reference rate constants derived from a collision–coalescence box model, the result of which is termed the enhancement factor (EF). We evaluate two hydrodynamic collision–coalescence kernels, one quiescent and one including the effects of small‐scale turbulence. Due to sampling volume limitations, DSDs must be averaged over length‐scales much greater than those relevant to the underlying physics, such that we also examine the role of averaging length‐scale with respect to process representation. Averaging length‐scales of 1.5 and 30 km are used, corresponding roughly to the horizontal grid lengths of cloud‐resolving models and high‐resolution climate models, respectively. EF values range from 0.1 to 40, with the greatest EFs associated with small mode diameter cases and a generally decreasing trend with drop size. For any given drop size or averaging length‐scale, there is about an order of magnitude variability in EFs. These results suggest that spatial variability on length‐scales smaller than 1.5 km prevents accurate retrieval of rate constants from large‐scale average DSDs. Large‐scale models must therefore account for small‐scale variability to represent cloud microphysical processes accurately. The turbulent kernel reduces EFs for all drop sizes, but can only account for at most half of the calculated EFs in marine stratocumulus.

A 10-Year Austral Summer Climatology of Observed and Modeled Intraseasonal, Mesoscale, and Diurnal Variations over the Maritime Continent

Abstract The Maritime Continent is one of the wettest regions on the planet and has been shown to be important for global budgets of heat and moisture. Convection in the region, however, varies on several interrelated scales, making it difficult to quantify the precipitation climate and understand the key processes. For example, the diurnal cycle in precipitation over the land varies substantially according to the phase of the Madden–Julian oscillation (MJO), and the diurnal precipitation cycle over the water is coupled to that over the land, in some cases for distances of over 1000 km from the coast. Here, a 10-yr austral summer climatology of diurnal and MJO-scale variations in rain rate over the land and sea over the Maritime Continent is presented. The climatology is based on mesoscale model simulations with a horizontal grid length of 4 km and satellite precipitation estimates. The amplitude of the observed diurnal precipitation cycle is shown to reach a maximum just prior to the MJO active phase, with a weaker secondary maximum after the MJO active phase. Although these two maxima also exist in the modeled diurnal precipitation cycle, there is less difference between the maxima before and after the MJO active phase than in the observations. The modeled sea-breeze circulation is also shown to possess approximately equal maxima just before and just after the MJO active period, suggesting that the asymmetry of the diurnal precipitation cycle about the MJO active period is related more to moisture availability than kinematic forcing.

Exploration of the impact of nearby sources on urban atmospheric inversions using large eddy simulation

The Indianapolis Flux Experiment (INFLUX) aims to quantify and improve the effectiveness of inferring greenhouse gas (GHG) source strengths from downstream concentration measurements in urban environments. Mesoscale models such as the Weather Research and Forecasting (WRF) model can provide realistic depictions of planetary boundary layer (PBL) structure and flow fields at horizontal grid lengths (Δx) down to a few km. Nevertheless, a number of potential sources of error exist in the use of mesoscale models for urban inversions, including accurate representation of the dispersion of GHGs by turbulence close to a point source. Here we evaluate the predictive skill of a 1-km chemistry-adapted WRF (WRF-Chem) simulation of daytime CO2 transport from an Indianapolis power plant for a single INFLUX case (28 September 2013). We compare the simulated plume release on domains at different resolutions, as well as on a domain run in large eddy simulation (LES) mode, enabling us to study the impact of both spatial resolution and parameterization of PBL turbulence on the transport of CO2. Sensitivity tests demonstrate that much of the difference between 1-km mesoscale and 111-m LES plumes, including substantially lower maximum concentrations in the mesoscale simulation, is due to the different horizontal resolutions. However, resolution is insufficient to account for the slower rate of ascent of the LES plume with downwind distance, which results in much higher surface concentrations for the LES plume in the near-field but a near absence of tracer aloft. Physics sensitivity experiments and theoretical analytical models demonstrate that this effect is an inherent problem with the parameterization of turbulent transport in the mesoscale PBL scheme. A simple transformation is proposed that may be applied to mesoscale model concentration footprints to correct for their near-field biases. Implications for longer-term source inversion are discussed.

Numerical modeling of a downwind-developing mesoscale convective system over the Masurian Lake District

Meteorological data concerning the severe convective system from the 21 August 2007 are analyzed in this study. Compiled information allows to understand the reason for the storm development and to identify its fundamental convective mode. Next, the EULAG model is utilized to perform an idealized test that shows a downwind–developing storm growth in an environment comparable to the one that was observed on the 21 August 2007 in the Masurian Lake District. Finally, the COSMO numerical weather prediction model is applied to reconstruct the storm development. The experiment is carried out for various computational grids having the horizontal grid length between 7.0 and 0.55 km. It turns out that the COSMO model is capable in simulating storms of that type. Since the model is used for operational weather forecasting in Poland the evaluation of this skill contributes to the increase of public safety.

Orographic drag on islands in the NWP mountain grey zone

The drag due to resolved and subgrid flow blocking and gravity waves is examined in simulations of flow over mountainous islands at ‘grey zone’ resolutions, in which the processes are neither well resolved or fully subgrid‐scale. Simulations of flows over narrow (South Georgia) and broad (New Zealand) mountain barriers are used to determine the behaviour of resolved and parametrized pressure drag and gravity‐wave momentum fluxes as a function of horizontal grid length. Analysis of the pressure drag in spectral space suggests that contributions from wavelengths shorter than 8–10 grid lengths are poorly resolved and that parametrization schemes should compensate for the missing drag from processes on these scales, rather than strictly the subgrid scales. The model performance for South Georgia is such that the total (resolved plus parametrized) pressure drag is approximately invariant across a range of resolutions. The parametrized (resolved) drag increases (decreases) as the grid length increases and the resolved and parametrized drag become comparable when the characteristic island wavelength is approximately 8 times the grid length. This is not true when the same tuning of the drag parametrization is applied to New Zealand, in which case the parametrized drag increases too rapidly with increasing grid length. However, satisfactory results are obtained when the scheme is re‐tuned and it is shown that the optimal tuning for the two islands is consistent with scaling the subgrid orographic heights to be representative of the orography on longer length‐scales, of up to 10 grid lengths. The results suggest that drag schemes require orographic information on these longer scales, rather than only the subgrid scale. At the coarsest resolutions, where the island wavelength and grid length are similar, the parametrized drag is too small because the grid boxes are increasingly treated as sea points. Possible solutions to this problem are discussed.

Idealized Large-Eddy and Convection-Resolving Simulations of Moist Convection over Mountainous Terrain

Abstract On summertime fair-weather days, thermally driven wind systems play an important role in determining the initiation of convection and the occurrence of localized precipitation episodes over mountainous terrain. This study compares the mechanisms of convection initiation and precipitation development within a thermally driven flow over an idealized double-ridge system in large-eddy (LESs) and convection-resolving (CRM) simulations. First, LES at a horizontal grid spacing of 200 m is employed to analyze the developing circulations and associated clouds and precipitation. Second, CRM simulations at horizontal grid length of 1 km are conducted to evaluate the performance of a kilometer-scale model in reproducing the discussed mechanisms. Mass convergence and a weaker inhibition over the two ridges flanking the valley combine with water vapor advection by upslope winds to initiate deep convection. In the CRM simulations, the spatial distribution of clouds and precipitation is generally well captured. However, if the mountains are high enough to force the thermally driven flow into an elevated mixed layer, the transition to deep convection occurs faster, precipitation is generated earlier, and surface rainfall rates are higher compared to the LES. Vertical turbulent fluxes remain largely unresolved in the CRM simulations and are underestimated by the model, leading to stronger upslope winds and increased horizontal moisture advection toward the mountain summits. The choice of the turbulence scheme and the employment of a shallow convection parameterization in the CRM simulations change the strength of the upslope winds, thereby influencing the simulated timing and intensity of convective precipitation.

A Validation of the Antarctic Mesoscale Prediction System Using Self-Organizing Maps and High-Density Observations from SNOWWEB

This study compares high-resolution output (1.1-km horizontal grid length) from twice-daily forecasts produced by the Antarctic Mesoscale Prediction System (AMPS) with a dense observational network east of Ross Island. Covering 10 000 km2, 15 SNOWWEB stations significantly increased the number of observation stations in the area to 19 during the 2014–15 austral summer. Collocated “virtual stations” created from AMPS output are combined with observations, producing a single dataset of zonal and meridional wind components used to train a self-organizing map (SOM). The resulting SOM is used to individually classify the observational and AMPS datasets, producing a time series of classifications for each dataset directly comparable to the other. Analysis of class composites shows two dominant weather patterns: low but directionally variable winds and high but directionally constant winds linked to the Ross Ice Shelf airstream (RAS). During RAS events the AMPS and SNOWWEB data displayed good temporal class alignment with good surface wind correlation. SOM analysis shows that AMPS did not accurately forecast surface-level wind speed or direction during light wind conditions where synoptic forcing was weak; however, it was able to forecast the low wind period occurrence accurately. Coggins’s regimes provide synoptic-scale context and show a reduced synoptic pressure gradient during these classes, increasing reliance on the ability of Polar WRF to resolve mesoscale dynamics. Available initialization data have insufficient resolution for the region’s complex topography, which likely impacts performance. The SOM analysis methods used are shown to be effective for model validation and are widely applicable.

Simulating the 20 May 2013 Moore, Oklahoma tornado with a 100‐metre grid‐length NWP model

AbstractSince 2013, the Met Office have run a 2.2 km horizontal gridlength version of the Unified Model (MetUM) as part of the National Oceanographic and Atmospheric Administration's Hazardous Weather Testbed Spring Forecasting Experiment. In this study, we perform high resolution MetUM simulations of the 20 May 2013 Oklahoma tornado outbreak at horizontal gridlengths between 2.2 km and 100 m. Here we present results showing that at 2.2 km gridlength the MetUM is able to simulate supercell‐like storms whereas at O(100 m) gridlength it is able to simulate realistic‐looking supercells with tornado‐like vortices. This opens up the opportunity for using such simulations to highlight areas of enhanced tornado risk ahead of time.

Controls on phase composition and ice water content in a convection-permitting model simulation of a tropical mesoscale convective system

Abstract. Simulations of tropical convection from an operational numerical weather prediction model are evaluated with the focus on the model's ability to simulate the observed high ice water contents associated with the outflow of deep convection and to investigate the modelled processes that control the phase composition of tropical convective clouds. The 1 km horizontal grid length model that uses a single-moment microphysics scheme simulates the intensification and decay of convective strength across the mesoscale convective system. However, deep convection is produced too early, the OLR (outgoing longwave radiation) is underestimated and the areas with reflectivities > 30 dBZ are overestimated due to too much rain above the freezing level, stronger updraughts and larger particle sizes in the model. The inclusion of a heterogeneous rain-freezing parameterisation and the use of different ice size distributions show better agreement with the observed reflectivity distributions; however, this simulation still produces a broader profile with many high-reflectivity outliers demonstrating the greater occurrence of convective cells in the simulations. Examining the phase composition shows that the amount of liquid and ice in the modelled convective updraughts is controlled by the following: the size of the ice particles, with larger particles growing more efficiently through riming and producing larger IWC (ice water content); the efficiency of the warm rain process, with greater cloud water contents being available to support larger ice growth rates; and exclusion or limitation of graupel growth, with more mass contained in slower falling snow particles resulting in an increase of in-cloud residence times and more efficient removal of LWC (liquid water content). In this simulated case using a 1 km grid length model, horizontal mass divergence in the mixed-phase regions of convective updraughts is most sensitive to the turbulence formulation. Greater mixing of environmental air into cloudy updraughts in the region of −30 to 0 °C produces more mass divergence indicative of greater entrainment, which generates a larger stratiform rain area. Above these levels in the purely ice region of the simulated updraughts, the convective updraught buoyancy is controlled by the ice particle sizes, demonstrating the importance of the microphysical processes on the convective dynamics in this simulated case study using a single-moment microphysics scheme. The single-moment microphysics scheme in the model is unable to simulate the observed reduction of mean mass-weighted ice diameter as the ice water content increases. The inability of the model to represent the observed variability of the ice size distribution would be improved with the use of a double-moment microphysics scheme.

Convection-Permitting Regional Weather Modeling with COSMO-EULAG: Compressible and Anelastic Solutions for a Typical Westerly Flow over the Alps

A comparison between anelastic and compressible convection-permitting weather forecasts for the Alpine region is presented. This involves mesoscale simulation of a typical westerly flow accompanied by a passage of frontal systems as well as intense airmass convection and orographic convection. The limited-area model employing a 2.2-km horizontal grid length is driven by time-dependent boundary conditions from a coarse-resolution model. The results obtained with the anelastic and the compressible model versions show good agreement. Validations of the 10-m wind, 2-m temperature, 2-m dewpoint temperature, total cloud cover, and surface precipitation against observations for a seven-member forecast ensemble reveal only minor differences between the two configurations. The sensitivity study demonstrates only a small impact of realistic pressure perturbations (about a reference profile) on the solutions. Overall, anelastic approximation proves remarkably accurate in simulating moist mesoscale dynamics. The study has been conducted using a newly developed hybrid limited-area nonhydrostatic version of the Consortium for Small-Scale Modeling (COSMO) model. This version facilitates the use of two alternative dynamical cores: compressible (original) and anelastic (new). The new dynamical core, which is based on the Euler–Lagrangian (EULAG) solver, aims at integrating atmospheric flow equations at resolutions higher than O(1) km for steep orography. A coupler has been developed to merge the EULAG dynamical core with the COSMO numerical weather prediction framework.

Representation of the grey zone of turbulence in the atmospheric boundary layer

Abstract. Numerical weather prediction model forecasts at horizontal grid lengths in the range of 100 to 1 km are now possible. This range of scales is the "grey zone of turbulence". Previous studies, based on large-eddy simulation (LES) analysis from the MésoNH model, showed that some assumptions of some turbulence schemes on boundary-layer structures are not valid. Indeed, boundary-layer thermals are now partly resolved, and the subgrid remaining part of the thermals is possibly largely or completely absent from the model columns. First, some modifications of the equations of the shallow convection scheme have been tested in the MésoNH model and in an idealized version of the operational AROME model at resolutions coarser than 500 m. Secondly, although the turbulence is mainly vertical at mesoscale (> 2 km resolution), it is isotropic in LES (< 100 m resolution). It has been proved by LES analysis that, in convective boundary layers, the horizontal production of turbulence cannot be neglected at resolutions finer than half of the boundary-layer height. Thus, in the grey zone, fully unidirectional turbulence scheme should become tridirectional around 500 m resolution. At Météo-France, the dynamical turbulence is modelled by a K-gradient in LES as well as at mesoscale in both MésoNH and AROME, which needs mixing lengths in the formulation. Vertical and horizontal mixing lengths have been calculated from LES of neutral and convective cases at resolutions in the grey zone.