Sub-grid Processes Research Articles

Technical note: Emulation of a large-eddy simulator for stratocumulus clouds in a general circulation model

Abstract. Here we present for the first time a proof of concept for an emulation-based method that uses a large-eddy simulations (LESs) to present sub-grid cloud processes in a general circulation model (GCM). We focus on two key variables affecting the properties of shallow marine clouds: updraft velocity and precipitation formation. The LES is able to describe these processes with high resolution accounting for the realistic variability in cloud properties. We show that the selected emulation method is able to represent the LES outcome with relatively good accuracy and that the updraft velocity and precipitation emulators can be coupled with the GCM practically without increasing the computational costs. We also show that the emulators influence the climate simulated by the GCM but do not consistently improve or worsen the agreement with observations on cloud-related properties, although especially the updraft velocity at cloud base is better captured. A more quantitative evaluation of the emulator impacts against observations would, however, have required model re-tuning, which is a significant task and thus could not be included in this proof-of-concept study. All in all, the approach introduced here is a promising candidate for representing detailed cloud- and aerosol-related sub-grid processes in GCMs. Further development work together with increasing computing capacity can be expected to improve the accuracy and the applicability of the approach in climate simulations.

Spatio-Temporal Variation of Critical Relative Humidity Based on Multiple Datasets

Clouds remain an important source of uncertainty in climate simulations, in large part because subgrid processes are not well represented. Critical relative humidity (RHc) is an important metric for subgrid-scale variability in humidity in cloud parameterization. Based on CloudSat and CALIPSO satellite data, we explored the spatial and temporal distribution characteristics of RHc, assessed the ability of ERA-5 and MERRA-2 reanalysis and CMIP-6 climate models to characterise humidity subgrid variability and further explored the influence of meteorological factors and aerosols. The statistical results showed that there was significant variation in the spatial distribution of RHc, with large variations in both latitude and altitude, as well as more pronounced monthly variations, and that there were differences in monthly variations between regions. Both the reanalysis data and the climate models were able to reproduce similar spatial and temporal distribution patterns but differed significantly in their specific values. The temporal correlations with satellite observations were also relatively poor. In addition, aerosols and meteorological conditions affected the distribution of RHc by influencing the cloud fraction at a certain relative humidity level, indicating that their influence needs to be considered in future parameterization schemes.

Distributional added value analysis of daily CORDEX-CORE RegCM4-7 historical precipitation simulations over Africa

The present study proposes a Distributional Added Value (DAV) analysis of CORDEX-CORE RegCM4-7 precipitation simulations over Africa, with the Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks-Cloud Classification System-Climate Data Record (PERSIANN-CCS-CDR) data as reference, which is chosen based on a preliminary assessment with other available long-term satellite precipitation estimates. The results revealed that CORDEX-CORE RegCM4-7 does add value to its driving GCMs distributions. The results show a positive Added Value Coverage (AVC) of 36.45 to 59.9% for annual daily events, 25.69 to 56.97% for seasonal daily events, 40.97 to 51.74% for heavy annual daily events (above the 95th percentile), and 48.8 to 58.89% for very heavy annual daily events (above the 99th percentile) regardless of the driving GCMs at both high-resolution and low-resolution. The results also feature an overall non-significant AVC ranging from 17.21 to 52.99%, thus suggesting that CORDEX-CORE RegCM4-7 mostly does not alter distributional properties from the driving GCMs if no value could be added. Although differences in individual GCM-based RegCM4-7 outputs exist, the RCM's internal physics and sub-grid processes appeared to be more influential on the DAV results due to the similarity in high- and low-resolution results from the downscaled outputs. This hypothesis is supported by additional evidence from the consistency of the historical simulations results with the evaluation runs, and the strong spatial dependence of the results to inherent local forcing and processes governing regions where most of the Added Value (AV) is observed. Overall, the results indicate that, for its first attempt to produce high-resolution climate simulations at nearly 25 km resolution, CORDEX-CORE RegCM4-7 has achieved the promise of an AV, especially for extreme events over Africa, even if further improvements will still be needed.

Confronting the Convective Gray Zone in the Global Configuration of the Met Office Unified Model

AbstractIn atmospheric models with kilometer‐scale grids the resolution approaches the scale of convection. As a consequence the most energetic eddies in the atmosphere are partially resolved and partially unresolved. The modeling challenge to represent convection partially explicitly and partially as a subgrid process is called the convective gray zone problem. The gray zone issue has previously been discussed in the context of regional models, but the evolution in regional models is constrained by the lateral boundary conditions. Here we explore the convective gray zone starting from a defined global configuration of the Met Office Unified Model using initialized forecasts and comparing different model formulations to observations. The focus is on convection and turbulence, but some aspects of the model dynamics are also considered. The global model is run at nominal 5 km resolution and thus contributions from both resolved and subgrid turbulent and convective fluxes are non‐negligible. The main conclusion is that in the present assessment, the configurations which include scale‐aware turbulence and a carefully reduced and simplified mass‐flux convection scheme outperform both the configuration with fully parameterized convection as well as a configuration in which the subgrid convection parameterization is switched off completely. The results are more conclusive with regard to convective organization and tropical variability than extratropical predictability. The present study thus endorses the strategy to further develop scale‐aware physics schemes and to pursue an operational implementation of the global 5 km‐resolution model to be used alongside other ensemble forecasts to allow researchers and forecasters to further assess these simulations.

Validation of an OpenFOAM®-based solver for the Euler equations with benchmarks for mesoscale atmospheric modeling

Within OpenFOAM, we develop a pressure-based solver for the Euler equations written in conservative form using density, momentum, and total energy as variables. Under simplifying assumptions, these equations are used to describe non-hydrostatic atmospheric flow. For the stabilization of the Euler equations and to capture sub-grid processes, we consider two Large Eddy Simulation models: the classical Smagorinsky model and the one equation eddy-viscosity model. To achieve high computational efficiency, our solver uses a splitting scheme that decouples the computation of each variable. The numerical results obtained with our solver are validated against numerical data available in the literature for two classical benchmarks: the rising thermal bubble and the density current. Through qualitative and quantitative comparisons, we show that our approach is accurate. This paper is meant to lay the foundation for a new open-source package specifically created for the quick assessment of new computational approaches for the simulation of atmospheric flows at the mesoscale level.

Combined physical and biogeochemical assessment of mesoscale eddy parameterisations in ocean models: Eddy induced advection at non-eddying resolutions

Ocean components of Earth System Models employed for climate projections do not routinely resolve mesoscale eddies for computational cost reasons, and the associated subgrid processes are still parameterised. While the performance of physics parameterisations in a numerical ocean model is normally assessed by examining the associated physical responses, biogeochemical responses are also important but often treated separately. Given recent advances in mesoscale eddy parameterisations, specifically for the eddy induced advection, this work systematically explores the joint consequences for physical as well as biogeochemical responses brought about by a more updated proposal for the eddy induced velocity coefficient, in the context of an idealised ocean relevant model. Relative to a high resolution mesoscale eddy resolving model, the more updated mesoscale eddy parameterisation is able to capture aspects of the model truth in the physical responses. The biogeochemical response is however rather more subtle, where a response with the conventional eddy parameterisation with a constant coefficient could arise from a physically inconsistent response. The present work highlights a need to assess both physical and biogeochemical aspects when judging the performance of eddy parameterisations, and additionally provides some important baseline model sensitivities that future assessments employing other parameterisations or in more complex settings could compare against.

A probabilistic, flux-conservative particle-based framework for transport in fractured porous media

Advection dominated transport processes in sub-surface formations are characterized by discontinuities in the fields of transported quantities, and realistic predictions are challenging for Eulerian transport schemes because they suffer from numerical diffusion. Henceforth, we have focused on developing a Lagrangian particle-tracking scheme for modeling advective solute transport in fractured media. To this end, we adopt an Embedded Discrete Fracture Model (EDFM) for fractured media with a permeable matrix. The flexibility to use non-conformal fracture–matrix discretizations makes EDFMs a compelling choice in field-scale flow problems. Unaffected by the numerical diffusion, Lagrangian transport schemes complement the potential of EDFMs by allowing the use of sufficiently large grid cell sizes for flow field computations.In an EDFM framework, the inter-continuum fluid mass exchange cannot be quantified by the particle trajectories/pathlines due to the unresolved fracture–matrix interfaces and different dimensionalities of the matrix and fracture discretizations. These constraints motivate the use of a stochastic particle-tracking scheme, and thus, we formulated a pathline-specific probability of inter-continuum particle transfer based on mass conservation of an elementary solute/fluid mass. The particle’s transfer probability is calculated for the maximum residence time period in its associated fracture/matrix control volume, thus making the scheme time-adaptive. In addition, a conditional residence time distribution was derived, which dictates the timestamp of the particle transfer.First, we showcase that the tracking scheme preserves the initially homogeneous solute concentration field, suggesting that the probabilistic rules are consistent with the inter-continuum fluxes. Additionally, we illustrate the estimation of an evolving solute plume and compare the results with those of an Eulerian counterpart. The presented stochastic approach enables straightforward formulations of Lagrangian models for dynamic and sub-grid processes, e.g., solute interactions with the solid phase, the mapping of their effects onto large scale transport and additionally, be included in random walk models for dispersion.

Non‐Local Parameterization of Atmospheric Subgrid Processes With Neural Networks

AbstractSubgrid processes in global climate models are represented by parameterizations which are a major source of uncertainties in simulations of climate. In recent years, it has been suggested that machine‐learning (ML) parameterizations based on high‐resolution model output data could be superior to traditional parameterizations. Currently, both traditional and ML parameterizations of subgrid processes in the atmosphere are based on a single‐column approach, which only use information from single atmospheric columns. However, single‐column parameterizations might not be ideal since certain atmospheric phenomena, such as organized convective systems, can cross multiple grid boxes and involve slantwise circulations that are not purely vertical. Here we train neural networks (NNs) using non‐local inputs spanning over 3 × 3 columns of inputs. We find that including the non‐local inputs improves the offline prediction of a range of subgrid processes. The improvement is especially notable for subgrid momentum transport and for atmospheric conditions associated with mid‐latitude fronts and convective instability. Using an interpretability method, we find that the NN improvements partly rely on using the horizontal wind divergence, and we further show that including the divergence or vertical velocity as a separate input substantially improves offline performance. However, non‐local winds continue to be useful inputs for parameterizating subgrid momentum transport even when the vertical velocity is included as an input. Overall, our results imply that the use of non‐local variables and the vertical velocity as inputs could improve the performance of ML parameterizations, and the use of these inputs should be tested in online simulations in future work.

Lagrangian Particle Dispersion Models in the Grey Zone of Turbulence: Adaptations to FLEXPART-COSMO for Simulations at 1 km Grid Resolution

Lagrangian particle dispersion models (LPDMs) are frequently used for regional-scale inversions of greenhouse gas emissions. However, the turbulence parameterizations used in these models were developed for coarse resolution grids, hence, when moving to the kilometre-scale the validity of these descriptions should be questioned. Here, we analyze the influence of the turbulence parameterization employed in the LPDM FLEXPART-COSMO model. Comparisons of the turbulence kinetic energy between the turbulence schemes of FLEXPART-COSMO and the underlying Eulerian model COSMO suggest that the dispersion in FLEXPART-COSMO suffers from a double-counting of turbulent elements when run at a high resolution of 1 times 1 ,hbox {km}^2. Such turbulent elements are represented in both COSMO, by the resolved grid-scale winds, and FLEXPART, by its stochastic parameterizations. Therefore, we developed a new parametrization for the variations of the winds and the Lagrangian time scales in FLEXPART in order to harmonize the amount of turbulence present in both models. In a case study for a power plant plume, the new scheme results in improved plume representation when compared with in situ flight observations and with a tracer transported in COSMO. Further in-depth validation of the LPDM against methane observations at a tall tower site in Switzerland shows that the model’s ability to predict the observed tracer variability and concentration at different heights above ground is considerably enhanced using the updated turbulence description. The high-resolution simulations result in a more realistic and pronounced diurnal cycle of the tracer concentration peaks and overall improved correlation with observations when compared to previously used coarser resolution simulations (at 7 km times 7 km). Our results indicate that the stochastic turbulence schemes of LPDMs, developed in the past for coarse resolution models, should be revisited to include a resolution dependency and resolve only the part of the turbulence spectrum that is a subgrid process at each different mesh size. Although our new scheme is specific to COSMO simulations at 1 times 1 ,hbox {km}^2 resolution, the methodology for deriving the scheme can easily be applied to different resolutions and other regional models.

Tropical Cirrus in Global Storm‐Resolving Models: 2. Cirrus Life Cycle and Top‐of‐Atmosphere Radiative Fluxes

AbstractCirrus clouds of various thicknesses and radiative characteristics extend over much of the tropics, especially around deep convection. They are difficult to observe due to their high altitude and sometimes small optical depths. They are also difficult to simulate in conventional global climate models, which have coarse grid spacings and simplified parameterizations of deep convection and cirrus formation. We investigate the representation of tropical cirrus in global storm‐resolving models (GSRMs), which have higher spatial resolution and explicit convection and could more accurately represent cirrus cloud processes. This study uses GSRMs from the DYnamics of the Atmospheric general circulation Modeled On Non‐hydrostatic Domains (DYAMOND) project. The aggregate life cycle of tropical cirrus is analyzed using joint albedo and outgoing longwave radiation (OLR) histograms to assess the fidelity of models in capturing the observed cirrus cloud populations over representative tropical ocean and land regions. The proportions of optically thick deep convection, anvils, and cirrus vary across models and are portrayed in the vertical distribution of cloud cover and top‐of‐atmosphere radiative fluxes. Model differences in cirrus populations, likely driven by subgrid processes such as ice microphysics, dominate over regional differences between convectively active tropical land and ocean locations.

Impact of sub-grid particle formation in sulfur-rich plumes on particle mass and number concentrations over China

Particle formation in point source plumes is a typical sub-grid process not explicitly resolved in chemical transport models, and is an important contributor to atmospheric aerosols. The sub-grid particle formation (SGPF) is closely associated with the concentration of hydroxyl radical and condensation sinks in the plume, thus it has diurnal, seasonal and spatial variations. The impacts of SGPF in point source plumes on both aerosol mass and number concentrations were first investigated in China by a newly developed global nested chemical transport model with a SGPF parameterization scheme. Although a mean oxidation fraction of 2.5% sulfur dioxide for parameterizing the SGPF is suggested, the oxidation fraction and the particles formed in point source plumes have clear spatiotemporal variations based on the results of the SGPF scheme. The SGPF can enhance the annual mean concentration of sulfate by more than 1 μg m−3 (10–25%) over high emission areas (HEAs) in central-eastern China (CEC). In terms of particle number concentration, the SGPF contributes much more to nucleation mode particles with increasing number concentration, by 25–50% in HEAs, 10–25% in CEC, and 1–5% in northwestern China (NWC) compared with lower values (within ±5%) for the accumulation mode. This study reveals the significant impacts of SGPF over China and its strong dependence on meteorological and environmental factors. Physically-based SGPF scheme should be incorporated when studying the regional effects of point source plumes on air quality and climate.

Evaluation and Comparison of Two Deep Convection Parameterization Schemes at Convection-Permitting Resolution

AbstractCoarse-grained results from a large-eddy simulation (LES) using the Weather Research and Forecasting (WRF) Model were compared in this study with the WRF simulations at a typical convection-permitting horizontal grid spacing of 3 km for an idealized case of deep moist convection. The purpose of this comparison is to identify major differences at the subgrid process level between two widely used deep convection parameterization schemes in the WRF Model. It is shown that there are considerable differences in subgrid process representations between the two schemes due to different parameterization formulations and underlying assumptions. The two schemes not only differ in trigger function, subgrid cloud model, and closure assumptions but also disagree with the coarse-grained LES results in terms of vertical mass flux profiles. Thus, it is difficult to discern which scheme is more advantageous over the other at the subgrid process level. The conclusions from this study highlight the importance of establishing benchmarks using observations and LES to develop and evaluate convection parameterization schemes suitable for models at convection-permitting resolution.

Modulation of the Bifurcation in Radiative‐Convective Equilibrium by Gray‐Zone Cloud and Turbulence Parameterizations

AbstractThis study investigates the mechanisms by which small‐scale turbulence and cloud physics determine the organization of large‐scale convection in radiative‐convective equilibrium (RCE), an idealization of the tropical atmosphere. Under uniform forcings similar to typical tropical conditions, the atmosphere in RCE might spontaneously separate into dry and moist regions on scales of 100–1,000 km, with convective clouds aggregating into a cluster in the latter. This phenomenon is known as convective self‐aggregation. Herein, we demonstrate that subtle changes in assumptions related to cloud physics and turbulence on scales of 1 km can dictate the emergence or suppression of convective self‐aggregation, resulting from a bifurcation of the dynamical system. The bifurcation occurs when a small dry patch forms in the domain and is sustained because it contributes to negative effective diffusivity of the circulation. Cloud‐radiation feedbacks and turbulence circulation interactions govern the formation of such dry patches, thereby modulating the bifurcation. This sensitive dependence on subgrid process models might be a fundamental barrier to climate predictability in light of inherent uncertainties in microscale processes. Because without the capability to include exact representations of those processes in climate models, slight differences in the different approximations used by modelers can lead to qualitative changes in climate predictions, at least for some processes.

The sequence of heavy precipitation and flash flooding of 12 and 13 September 2019 in eastern Spain. Part II: A hydro-meteorological predictability analysis based on convection-permitting ensemble strategies

AbstractOn 12 and 13 September 2019, widespread flash flooding caused devastating effects across eastern Spain. Within the framework of the HyMeX program, this study examines predictability of the long-lasting heavy precipitation episode (HPE) conducive to flash flooding. A set of short-range, convection-permitting ensemble prediction systems (EPSs) is built to cope with different sources of meteorological uncertainty. Specifically, the performance of an Ensemble Kalman Filter, tailored bred vectors and stochastic model parameterizations is compared to more standard ensemble generation techniques such as dynamical downscaling and multiple physics. Results indicate EPS focusing on sampling model uncertainties related to parameterization of subgrid process are skillful, especially when deep convection and its interaction with complex orography are important. Furthermore, representation of small-scale thermodynamical aspects is improved through data assimilation, leading to an enhanced forecasting skill as well. Nevertheless, predictability remains relatively low at the catchment scale in terms of exceedance probabilities in cumulative precipitation and peak discharge. The analysis presented herein could serve as a basis for the future implementation of real-time flash flood warning systems based on skillful meteorological EPSs over small-to-medium, semi-arid watersheds in eastern Spain and, by extension, over the flood-prone Western Mediterranean region.

Data assimilation empowered neural network parametrizations for subgrid processes in geophysical flows

Modeling the effect of subgrid-scale processes is one of the main obstacles in the accurate prediction of multiscale systems. An investigation considers how machine learning methods can be applied to model subgrid-scale processes and integrated within sequential data assimilation methods. It is found that the use of machine-learning-based closure modeling in conjunction with data assimilation improves the prediction of multiscale systems and can be considered a promising approach to numerical weather prediction tasks in the age of data.

Sensitivity of Simulations of Zambian Heavy Rainfall Events to the Atmospheric Boundary Layer Schemes

Weather forecasting relies on the use of numerical weather prediction (NWP) models, whose resolution is informed by the available computational resources. The models resolve large scale processes, while subgrid processes are parametrized. One of the processes that is parametrized is turbulence which is represented in planetary boundary layer (PBL) schemes. In this study, we evaluate the sensitivity of heavy rainfall events over Zambia to four different PBL schemes in the Weather Research and Forecasting (WRF) model using a parent domain with a 9 km grid length and a 3 km grid spacing child domain. The four PBL schemes are the Yonsei University (YSU), nonlocal first-order medium-range forecasting (MRF), University of Washington (UW) and Mellor–Yamada–Nakanishi–Niino (MYNN) schemes. Simulations were done for three case studies of extreme rainfall on 17 December 2016, 21 January 2017 and 17 April 2019. The use of YSU produced the highest rainfall peaks across all three cases; however, it produced performance statistics similar to UW that are higher than those of the two other schemes. These statistics are not maintained when adjusted for random hits, indicating that the extra events are mainly random rather than being skillfully placed. UW simulated the lowest PBL height, while MRF produced the highest PBL height, but this was not matched by the temperature simulation. The YSU and MYNN PBL heights were intermediate at the time of the peak; however, MYNN is associated with a slower decay and higher PBL heights at night. WRF underestimated the maximum temperature during all cases and for all PBL schemes, with a larger bias in the MYNN scheme. We support further use of the YSU scheme, which is the scheme selected for the tropical suite in WRF. The different simulations were in some respects more similar to one another than to the available observations. Satellite rainfall estimates and the ERA5 reanalysis showed different rainfall distributions, which indicates a need for more ground observations to assist with studies like this one.

Hydrogen–helium chemical and nuclear galaxy collision: Hydrodynamic simulations on AVX-512 supercomputers

A computational hydrodynamic model of interacting galaxies is presented. The interstellar medium (ISM) is described by a model of gravitational multicomponent single-velocity hydrodynamics. A model based on first moments of the collisionless Boltzmann equation is used to describe the stellar component and dark matter. Subgrid processes of star formation and supernovae feedback, as well as cooling and heating functions are added to the hydrodynamic model. The computational model includes chemical and nuclear reactions of the basic forms of hydrogen and helium before the formation of the helium hydride ion. The hydrodynamic equations have a suitable form for solving by a unified computational method based on the “Harten – Lax – van Leer” (HLL) method. The equations and the numerical method are written in vector form to use advanced vector extensions AVX-512 to speed up the calculations. A parallel implementation of the model and some computational simulation experiments on the chemical dynamics of interacting galaxies are presented.

Study of the Vertical Structure of the Coastal Boundary Layer Integrating Surface Measurements and Ground-Based Remote Sensing.

The understanding of the atmospheric processes in coastal areas requires the availability of quality datasets describing the vertical and horizontal spatial structure of the Atmospheric Boundary Layer (ABL) on either side of the coastline. High-resolution Numerical Weather Prediction (NWP) models can provide this information and the main ingredients for good simulations are: an accurate description of the coastline and a correct subgrid process parametrization permitting coastline discontinuities to be caught. To provide an as comprehensive as possible dataset on Mediterranean coastal area, an intensive experimental campaign was realized at a near-shore Italian site, using optical and acoustic ground-based remote sensing and surface instruments, under different weather characteristic and stability conditions; the campaign is also fully simulated by a NWP model. Integrating information from instruments responding to different atmospheric properties allowed for an explanation of the development of various patterns in the vertical structure of the atmosphere. Wind LiDAR measurements provided information of the internal boundary layer from the value of maximum height reached by the wind profile; a height between 80 and 130 m is often detected as an interface between two different layers. The NWP model was able to simulate the vertical wind profiles and the eight of the ABL.

Multi-fluid modelling of bubbly channel flows with an adaptive multi-group population balance method

Mass, momentum, and energy transfer in bubbly flows strongly depends on the bubble’s size distribution, which determines the contact area between the interacting phases. Characterization of bubble sizes in polydisperse flows requires empirical modelling of sub-grid physical mechanisms such as break-up and coalescence. In the present work an adaptive multiple size-group (A-MuSiG) method is incorporated into the Eulerian multiphase solver available in Simcenter STAR-CCM+ in order to model polydisperse bubbly flows in horizontal and vertical channels. The disperse phasespace is discretized into multiple size-groups each represented by its own size, number-density, and velocity field. The diameter of the bubbles in each of the size-groups varies in time and space, dynamically adapting to the local flow conditions. The interphase momentum transfer between the continuous phase and polydisperse bubbles is modelled through drag, virtual mass, turbulent dispersion, and lift forces. For modelling sub-grid bubble break-up and coalescence processes, different phenomenological kernels are evaluated. The empirical parameters of the adopted kernels are calibrated in two steps. The initial stage of the analysis considers experimental channel flows at low Reynolds number and zero-gravity conditions, under which the bubble size distribution is solely dependent on coalescence. As part of the second phase of the evaluation, additional parametric simulations in turbulent channel flows are performed in order to calibrate the break-up models, assuming the coalescence scaling constants derived in the previous step. The obtained results demonstrate that in flows with high turbulent mixing the ensuing bubble dynamics are strongly coupled to the internal properties of the population, which in turn influence the developing multiphase interactions in a transient manner.

The parallel & distributed code for numerical simulation of relativistic magnetohydrodynamics simulation

A new code for modeling relativistic magnetohydrodynamic flows is described in the paper. The developed program code is based on a combination of adaptive nested grids. Magnetic hydrodynamics of the process is simulated using nested grids. Subgrid processes are simulated using regular grids. In the paper, we will outline the main components of program code building. These steps are common to other program codes for magnetized flows simulation.