The Climate Modeling Alliance Atmosphere Dynamical Core: Concepts, Numerics, and Scaling

Abstract. The non-hydrostatic (NH) compressible Euler equations for dry atmosphere were solved in a simplified two-dimensional (2-D) slice framework employing a spectral element method (SEM) for the horizontal discretization and a finite difference method (FDM) for the vertical discretization. By using horizontal SEM, which decomposes the physical domain into smaller pieces with a small communication stencil, a high level of scalability can be achieved. By using vertical FDM, an easy method for coupling the dynamics and existing physics packages can be provided. The SEM uses high-order nodal basis functions associated with Lagrange polynomials based on Gauss–Lobatto–Legendre (GLL) quadrature points. The FDM employs a third-order upwind-biased scheme for the vertical flux terms and a centered finite difference scheme for the vertical derivative and integral terms. For temporal integration, a time-split, third-order Runge–Kutta (RK3) integration technique was applied. The Euler equations that were used here are in flux form based on the hydrostatic pressure vertical coordinate. The equations are the same as those used in the Weather Research and Forecasting (WRF) model, but a hybrid sigma–pressure vertical coordinate was implemented in this model. We validated the model by conducting the widely used standard tests: linear hydrostatic mountain wave, tracer advection, and gravity wave over the Schär-type mountain, as well as density current, inertia–gravity wave, and rising thermal bubble. The results from these tests demonstrated that the model using the horizontal SEM and the vertical FDM is accurate and robust provided sufficient diffusion is applied. The results with various horizontal resolutions also showed convergence of second-order accuracy due to the accuracy of the time integration scheme and that of the vertical direction, although high-order basis functions were used in the horizontal. By using the 2-D slice model, we effectively showed that the combined spatial discretization method of the spectral element and finite difference methods in the horizontal and vertical directions, respectively, offers a viable method for development of an NH dynamical core.

<p>This study investigates the prediction skill of the sub-seasonal prediction model that can depend on the choice of dynamical cores: the finite volume (FV) dynamical core on a latitude-longitude grid system versus spectral element (SE) dynamical core on a cubed-sphere grid system. Recent researches showed that the SE dynamical core on a uniform grid system increases parallel scalability and removes the need for polar filters mitigating uncertainty in climate prediction, particularly for the Arctic region. However, it remains unclear whether the choice of dynamical cores can actually yield significant skill changes or not. To tackle this issue, we implemented a sub-seasonal prediction model based on the Community Atmospheric Model version 5 (CAM5) by incorporating the above two dynamical cores with virtually the same physics schemes. Sub-seasonal prediction skills of the SE dynamical core and FV dynamical core are verified with ERA-interim reanalysis during the early winter (November – December) and the late winter (January – February) from 2001/2002 to 2017/2018. The prediction skills of the two different dynamical cores were significantly different regardless of the virtually same physics schemes. In the ocean, the predictability of the SE dynamical core is similar to the FV dynamical core, mostly because of our simulation configuration imposing the same boundary and initial conditions at the surface. Notable differences in the one-month predictability between the two cores are found for the wintertime Arctic and mid-latitudes, particularly over North America and Eurasia continents. With the one-month lead, SE dynamical core exhibited higher predictability over North America in late winter, whereas the FV dynamical core showed relatively higher predictability in East Asia and Eurasia in early winter. One of the reasons for these differences may be the different manifestations of Arctic-midlatitudes linkage in the two dynamical cores; the SE dynamical core captures warmer Arctic and colder mid-latitudes relatively well than the FV dynamical core. Therefore, we conclude that the careful choice of dynamical cores of sub-seasonal prediction models is needed.</p>

Numerical models of the atmosphere combine a dynamical core, which approximates solutions to the adiabatic, frictionless governing equations for fluid dynamics, with tendencies arising from the parametrization of other physical processes. Since potential vorticity (PV) is conserved following fluid flow in adiabatic, frictionless circumstances, it is possible to isolate the effects of non‐conservative processes by accumulating PV changes in an air‐mass‐relative framework. This ‘PV tracer technique’ is used to accumulate separately the effects on PV of each of the different non‐conservative processes represented in a numerical model of the atmosphere. Dynamical cores are not exactly conservative because they introduce, explicitly or implicitly, some level of dissipation and adjustment of prognostic model variables which acts to modify PV. Here, the PV tracers technique is extended to diagnose the cumulative effect of the non‐conservation of PV by a dynamical core and its characteristics relative to the PV modification by parametrized physical processes.Quantification using the Met Office Unified Model reveals that the magnitude of the non‐conservation of PV by the dynamical core is comparable to those from physical processes. Moreover, the residual of the PV budget, when tracing the effects of the dynamical core and physical processes, is at least an order of magnitude smaller than the PV tracers associated with the most active physical processes. The implication of this work is that the non‐conservation of PV by a dynamical core can be assessed in case‐studies with a full suite of physics parametrizations and directly compared with the PV modification by parametrized physical processes. The non‐conservation of PV by the dynamical core is shown to move the position of the extratropical tropopause while the parametrized physical processes have a lesser effect at the tropopause level.

Water vapour plays a key role in the Earth's energy balance. Almost 50% of the absorbed solar radiation at the surface is used to cool the surface, through evaporation, and warm the atmosphere, through release of latent heat. Latent heat is the single largest factor in warming the atmosphere and in transporting heat from low to high latitudes. Water vapour is also the dominant greenhouse gas and contributes to a warming of the climate system by some 24°C (Kondratev 1972). However, water vapour is a passive component in the troposphere as it is uniquely determined by temperature and should therefore be seen as a part of the climate feedback system. In this short overview, we will first describe the water on planet Earth and the role of the hydrological cycle: the way water vapour is transported between oceans and continents and the return of water via rivers to the oceans. Generally water vapour is well observed and analysed; however, there are considerable obstacles to observing precipitation, in particular over the oceans. The response of the hydrological cycle to global warming is far reaching. Because different physical processes control the change in water vapour and evaporation/precipitation, this leads to a more extreme distribution of precipitation making, in general, wet areas wetter and dry areas dryer. Another consequence is a transition towards more intense precipitation. It is to be expected that the changes in the hydrological cycle as a consequence of climate warming may be more severe that the temperature changes.

The paper introduces a moist, deterministic test case of intermediate complexity for Atmospheric General Circulation Models (AGCMs). We suggest pairing an AGCM dynamical core with simple physical parameterizations to test the evolution of a single, idealized, initially weak vortex into a tropical cyclone. The initial conditions are based on an initial vortex seed that is in gradient‐wind and hydrostatic balance. The suggested “simple‐physics” package consists of parameterizations of bulk aerodynamic surface fluxes for moisture, sensible heat and momentum, boundary layer diffusion, and large‐scale condensation. Such a configuration includes the important driving mechanisms for tropical cyclones, and leads to a rapid intensification of the initial vortex over a forecast period of ten days. The simple‐physics test paradigm is not limited to tropical cyclones, and can be universally applied to other flow fields. The physical parameterizations are described in detail to foster model intercomparisons. The characteristics of the intermediate‐complexity test case are demonstrated with the help of four hydrostatic dynamical cores that are part of the Community Atmosphere Model version 5 (CAM 5) developed at the National Center for Atmospheric Research (NCAR). In particular, these are the Finite‐Volume, Spectral Element, and spectral transform Eulerian and semi‐Lagrangian dynamical cores that are coupled to the simple‐physics suite. The simulations show that despite the simplicity of the physics forcings the models develop the tropical cyclone at horizontal grid spacings of about 55 km and finer. The simple‐physics simulations reveal essential differences in the storm's structure and strength due to the choice of the dynamical core. Similar differences are also seen in complex full‐physics aqua‐planet experiments with CAM 5 which serve as a motivator for this work. The results suggest that differences in complex full‐physics simulations can be, at least partly, replicated in simplified model setups. The simplified experiments might therefore provide easier access to an improved physical understanding of how the dynamical core and moist physical parameterizations interact. It is concluded that the simple‐physics test case has the potential to close the gap between dry dynamical core assessments and full‐physics aqua‐planet experiments, and can shed light on the role of the dynamical core in the presence of moisture processes.

Abstract. Global climate models (GCMs) have advanced in many ways as computing power has allowed more complexity and finer resolutions. As GCMs reach storm-resolving scales, they need to be able to produce realistic precipitation intensity, duration, and frequency at fine scales with consideration of scale-aware parameterization. This study uses a state-of-the-art storm-resolving GCM with a nonhydrostatic dynamical core – the Model for Prediction Across Scales (MPAS), incorporated in the atmospheric component (Community Atmosphere Model, CAM) of the open-source Community Earth System Model (CESM), within the System for Integrated Modeling of the Atmosphere (SIMA) framework (referred to as SIMA-MPAS). At uniform coarse (here, at 120 km) grid resolution, the SIMA-MPAS configuration is comparable to the standard hydrostatic CESM (with a finite-volume (FV) dynamical core) with reasonable energy and mass conservation on climatological timescales. With the comparable energy and mass balance performance between CAM-FV (workhorse dynamical core) and SIMA-MPAS (newly developed dynamical core), it gives confidence in SIMA-MPAS's applications at a finer resolution. To evaluate this, we focus on how the SIMA-MPAS model performs when reaching a storm-resolving scale at 3 km. To do this efficiently, we compose a case study using a SIMA-MPAS variable-resolution configuration with a refined mesh of 3 km covering the western USA and 60 km over the rest of the globe. We evaluated the model performance using satellite and station-based gridded observations with comparison to a traditional regional climate model (WRF, the Weather Research and Forecasting model). Our results show realistic representations of precipitation over the refined complex terrains temporally and spatially. Along with much improved near-surface temperature, realistic topography, and land–air interactions, we also demonstrate significantly enhanced snowpack distributions. This work illustrates that the global SIMA-MPAS at storm-resolving resolution can produce much more realistic regional climate variability, fine-scale features, and extremes to advance both climate and weather studies. This next-generation storm-resolving model could ultimately bridge large-scale forcing constraints and better inform climate impacts and weather predictions across scales.

The use of mesh-based methods for modeling elongated composite structures with inhomogeneities leads to an increase in computational costs when discretizing the waveguide part of greater linear dimensions, while the semi-analytical numerical methods do not allow one to directly describe the structures with local inhomogeneities of an arbitrary shape. To compensate for the shortcomings of these two classes of numerical methods, we propose a hybrid scheme based on the spectral element method (SEM) and the semi-analytical finite element method (SAFEM) for studying the anti-plane vibrations of a composite structure in the frequency domain. Thus, for a waveguide, this scheme makes it possible to represent the solution via the SAFEM as a sum of modes or guided waves, and the adjacent regions are discretized using the SEM. The displacement and stress continuity conditions are imposed on the common boundary of two domains. To couple the solutions, we introduce an auxiliary function of displacement, which is approximated by applying the same basis functions as those used in SEM and SAFEM (Lagrange interpolation polynomials on the Gauss–Legendre–Lobatto nodal points). The unknown expansion coefficients of this function are determined by the Galerkin and collocation methods. It has been established that both methods provide the same accuracy. The results obtained by the hybrid approach employing the Galerkin and collocation methods are compared with the results calculated in the standard finite element software. It is shown that they are in good agreement. The presented hybrid approach can be straightforwardly extended to the case of in-plane motion, but it requires significant refinement for a three-dimensional case.

Non-Amazonian South America has one of the highest rates of conversion of native ecosystems globally. Most of the studies investigating the climate impacts of these changes focus on the Amazon while the possible influences that these changes may have on climate of non-Amazonian regions have received less attention. The aim of this thesis was to evaluate the impacts of land use and land cover on the mean and extreme climate of non-Amazonian South America by conducting modelling experiments for pre-clearing (before year 1500) and present (year 2005) land covers. It develops new data sets of changes in land surface characteristics for this period and applies a high resolution regional climate model to simulate the potential impacts of changes in natural vegetation cover. The thesis begins by providing a theoretical framework of surface-atmosphere interactions. It then reviews the process of land use and land cover change and subsequent climatic consequences in non-Amazonian South America and identifies those ecosystems most affected and least studied. The review highlights that non-Amazonian regions have lost more than 3.4 million km2 of natural vegetation since European colonization. Despite the magnitude of ecosystem loss, non-Amazonian South America accounts for significantly fewer studies addressing land surface-atmospheric processes compared to the Amazon region, highlighting the knowledge gap in relation to the main topic of this thesis. Based on these results, the following chapters address this knowledge gap by focusing on the climatic impacts of land use and land cover change in four main ecosystems: the Atlantic Forest (Brazil, Argentina and Paraguay), Cerrado (Brazil), Dry Chaco (Argentina, Bolivia and Paraguay) and the Chilean Matorral (Chile). I then apply the variable resolution CSIRO Conformal Cubic Atmospheric Model (CCAM) global climate model at a 25 km spatial resolution over the South American continent to quantify the seasonal climate impacts of each of historic land cover change. The results of computed modelling experiments show significant changes in surface fluxes, temperature, precipitation and moisture in all ecosystems. For instance, simulated temperature changes were stronger in the Cerrado and the Chilean Matorral with an increase of between 0.7 and 1.4 °C. Changes in the hydrological cycle revealed high regional variability. Results also show that the loss of natural vegetation has significantly affected temperature extremes as a decrease in the number of warm days and an increase in the number of warm nights. Importantly, there is a strong dependence on both seasonality and the vegetation contrast inflicted by land use/cover change, with large roughness changes resulting in increasing wind speed and advection, while smaller roughness changes result in feedbacks more reliant on the distinction between sensible and latent heat fluxes. This explains the dry season response in both temperature extremes and the increase in aridity according to land use/cover change, whereby regions with increased wind speed reduce warm day temperature extremes, despite increasing mean temperature trend, and have a greater impact on atmospheric water demand than those regions that mainly increase sensible heat fluxes. These results can explain the observed trends in temperature extremes in non-Amazonian South America and highlights the need to embed land use/cover change as a forcing within future climate change scenarios. The main conclusions of the thesis are: a) non-Amazonian South America is one of the most impacted and least studied regions worldwide in terms of climatic impacts of land use and land cover change, b) modelled impacts of this process are expressed as significant changes in surface temperature and the hydrological cycle through changes in soil and atmospheric moisture, and have increased the aridity in all the examined ecosystems. The thesis highlights the importance of considering the influence of land use and land cover on the mean and extreme climate through changes in biophysical properties that significantly impact the surface-atmospheric coupling and therefore the hydrological cycle. This is critical consideration for national natural ecosystem management strategies as conserving and restoring natural vegetation cover may help mitigate the negative consequences of climate change and therefore have a direct influence on the welfare of the region’s 200 million inhabitants.

The Amazon forest has recently experienced substantial human-induced loss of forest cover. However, the extent to which such historical deforestation has altered regional observed precipitation through inter-regional atmospheric moisture transport remains unclear. Here, we combine satellite observations and an atmospheric moisture tracking model to quantify these feedbacks over the past four decades (1980-2019). We identify a contrasting northern increase and southern decrease dipole trend in observed precipitation across the Amazon basin. The pronounced reduction in precipitation for the southern Amazon basin reaches up to 3.9-5.4 mm yr-1 per year, resulting in an 8-11% decline in annual precipitation across the observation period. We discover that this reduction in precipitation is primarily (52-72%) related to widespread deforestation in the southern basin and upwind regions over South America. Deforestation substantially suppresses forest-sourced moisture, increases atmospheric stability and moisture outflow, leading to precipitation reduction. We also find that climate models substantially underestimate the sensitivity of precipitation to deforestation, implying that the Amazon forest is at risk of major loss much sooner than previously projected.

Advection of trace species (tracers), also called tracer transport, in models of the atmosphere and other physical domains is an important and potentially computationally expensive part of a model's dynamical core (dycore). Semi-Lagrangian (SL) advection methods are efficient because they permit a time step much larger than the advective stability limit for explicit Eulerian methods. Thus, to reduce the computational expense of tracer transport, dycores often use SL methods to advect passive tracers. The class of interpolation semi-Lagrangian (ISL) methods contains potentially extremely efficient SL methods. We describe a set of ISL bases for element-based transport, such as for use with atmosphere models discretized using the spectral element (SE) method. An ISL method that uses the natural polynomial interpolant on Gauss-Legendre-Lobatto (GLL) SE nodes of degree at least three is unstable on the test problem of periodic translational flow on a uniform element grid. We derive new alternative bases of up to order of accuracy nine that are stable on this test problem; we call these the Islet bases. Then we describe an atmosphere tracer transport method, the Islet method, that uses three grids that share an element grid: a dynamics grid supporting, for example, the GLL basis of degree three; a physics grid with a configurable number of finite-volume subcells per element; and a tracer grid supporting use of our Islet bases, with particular basis again configurable. This method provides extremely accurate tracer transport and excellent diagnostic values in a number of validation problems. We conclude with performance results that use up to 27,600 NVIDIA V100 GPUs on the Summit supercomputer.

The precipitable water (PW) content was estimated over southern Greece for three atmospheric layers, using 28-yr twice-daily radiosonde measurements of temperature, humidity, and atmospheric pressure. Precipitable water demonstrates considerable variation at the monthly and interannual time scales as well as statistically significant upward trends at the annual level and for June–August (JJA) and September–November (SON). The relationship between PW and surface climate was also found to vary by season and surface climate variable. Disaggregation of seasonal PW trends by synoptic type revealed that seasonal trends might be attributed to increases in moisture in a limited number of airmass types such that maritime air masses appear to be becoming moister while their continental counterparts are drying. Despite fundamental changes to the intrinsic PW properties of air masses for some synoptic categories, the significance of these for determining the overall trend in seasonal PW levels is related to their weighted contribution as determined by changing airmass frequency. Consequently, the climatological effects of a moistening or drying air mass may be offset by a declining frequency. It is also shown that increasing PW trends cannot be attributed unequivocally to airmass warming as significant increases in PW have occurred in conjunction with falling temperatures in some seasons and airmass types. Accordingly, factors other than the commonly assumed global-warming-related increase in atmospheric moisture via the link between temperature and saturation vapor pressure must be considered if the mechanisms responsible for PW trends over southern Greece are to be fully understood.

The rapid development of science and technology has enabled finer and finer resolutions in atmospheric general circulation models (AGCMs). Parallelization becomes progressively more critical as the resolution of AGCMs increases. This paper presents a new parallel version of the finite-difference Gridpoint Atmospheric Model of the Institute of Atmospheric Physics (IAP)–State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics (LASG; GAMIL) with various parallel optimization strategies, including two-dimensional hybrid parallel decomposition; hybrid parallel programming; parallel communications for coupling the physical packages, land surface, and dynamical core; and a cascading solution to the tridiagonal equations used in the dynamical core. The new parallel version under two different horizontal resolutions (1° and 0.25°) is evaluated. The new parallel version enables GAMIL to achieve higher parallel efficiency and utilize a greater number of CPU cores. GAMIL1° achieves 37.8% parallel efficiency using 960 CPU cores, while GAMIL0.25° achieves 57.5% parallel efficiency.

Recent machine learning (ML) models have shown great success for weather prediction tasks, suggesting that atmospheric dynamics can in principle be learned from data. However, such approaches suffer from instability or drift in long integrations and are hence usually not suited for climate simulations. Incorporating physical knowledge promises to alleviate such shortcomings. Here, we present PseudospectralNet (PSN), an architecture for a hybrid atmospheric model that combines a quasi‐geostrophic physics‐based dynamical core with a data‐driven core based on an UNet. Our architecture transforms between grid and spectral space at every time step and therefore mimics the pseudospectral solution approach many intermediate‐complexity atmospheric models follow. Neural networks are separately defined in the spectral and grid space and are combined with physics‐based dynamical cores in each of these spaces. We train PSN separately on data from quasi‐geostrophic models, primitive equation models, and reanalysis data. We use this model to study the effect of adding physics‐based cores to ML models on forecast error and numerical stability. We find in both regards that adding the physics‐based dynamical core to our model helps short‐term predictability and long‐term numerical stability of the hybrid model.

The aim of this article is to develop a dry‐mass conserving hydrostatic global spectral dynamical core in a general moist atmosphere, which can be regarded as an alternative, improved version of that used in the current Integrated Forecast System (IFS) of the European Centre for Medium‐Range Weather Forecasts (ECMWF). In contrast to the original IFS‐like core, the dry‐mass vertical coordinate is employed and the mass continuity equation is expressed in terms of the dry air density, which ensures the inherent conservation of dry air mass. Meanwhile, the thermodynamic equation is reformulated with a modified temperature variable and the formula used to compute the full pressure vertical velocity is derived rigorously. To assess the performance of this new core, an idealized tropical cyclone (TC) test is conducted. Simulation results from both the new core and the original IFS‐like dynamical core are presented and compared. The results show that the TC‐like storm produced by the new dynamical core is more intense, more compact and more concentric, and is thus much more in line with previous results from other global models. In this new dynamical core, the diagnosed full pressure vertical velocity is decomposed into four components, of which the first component, the dry hydrostatic pressure vertical velocity, dominates. Sensitivity experiments imply that despite their small numerical value the other three components should not be neglected, especially for medium‐range forecasts.

This study investigates the moisture budgets and resolution dependency of precipitation extremes in an aquaplanet framework based on the Community Atmosphere Model, version 4 (CAM4). Moisture budgets from simulations using two different dynamical cores, the Model for Prediction Across Scales-Atmosphere (MPAS-A) and High Order Method Modeling Environment (HOMME), but the same physics parameterizations suggest that during precipitation extremes the intensity of precipitation is approximately balanced by the vertical advective moisture transport. The resolution dependency in extremes from simulations at their native grid resolution originates from that of vertical moisture transport, which is mainly explained by changes in dynamics (related to vertical velocity ω) with resolution. When assessed at the same grid scale by area-weighted averaging the fine-resolution simulations to the coarse grids, simulations with either dynamical core still demonstrate resolution dependency in extreme precipitation with no convergence over the tropics, but convergence occurs at a wide range of latitudes over the extratropics. The use of lower temporal frequency data (i.e., daily vs 6 hourly) reduces the resolution dependency. Although thermodynamic (moisture) changes become significant in offsetting the effect of dynamics when assessed at the same grid scale, especially over the extratropics, changes in dynamics with resolution are still large and explain most of the resolution dependency during extremes. This suggests that the effects of subgrid-scale variability of ω and vertical moisture transport during extremes are not adequately parameterized by the model at coarse resolution. The aquaplanet framework and analysis described in this study provide an important metric for assessing sensitivities of cloud parameterizations to spatial resolution and dynamical cores under extreme conditions.