Development of WACCM With the Non‐Hydrostatic MPAS‐A Dynamical Core

The non‐hydrostatic Model for Prediction Across Scales‐Atmosphere (MPAS‐A) dynamical core has recently been adapted for the Specified Chemistry Whole Atmosphere Community Climate Model (SC‐WACCM). In this study, the mean zonal wind and temperature climatology from SC‐WACCM/MPAS‐A is compared with the results from SC‐WACCM using the finite volume and spectral element dynamical cores, as well as the zonal wind and temperature climatology of Upper Atmosphere Research Satellite mission and SABER. The simulations have been performed at horizontal resolutions of ∼100 km. Generally a good agreement is seen between the results from the three dynamical cores, which verifies that the new dynamical core is working with WACCM.


Introduction
The need for ground-to-exosphere General Circulation Models (GCMs) were specified for the first time by Roble (2000).These models, commonly known as the whole atmosphere models, are needed for a more accurate study of the relevant physical and chemical interactions, climate change, climate response to solar variability, space weather, and the interpretation of global observations (Akmaev, 2011;Liu et al., 2010;Roble, 2000).Two decades following the work of Roble, whole atmosphere modeling has developed into an active and fast growing area of research.
As discussed in Jackson et al. (2019) and Akmaev (2011) there is a strong motivation to accurately represent the influence and variability of lower atmosphere dynamics and weather in physics-based space weather forecasts and applications.For accurate simulation of the region between the Sun and the Earth, the impacts from the lower atmosphere need to be taken into account in addition to the atmospheric impacts from the solar output and magnetosphere (Jackson et al., 2019;Liu, 2016;Liu et al., 2010Liu et al., , 2018)).Such models will be utilized increasingly more in geospace applications.
Geospace applications require atmospheric models with tops in excess of 500 km, well into the upper thermosphere, where the processes have fast temporal scales and large vertical scales.In these regions, the accuracy of the hydrostatic and shallow atmosphere approximation degrade, especially on time scales comparable to or shorter than buoyancy period and vertical scales longer than the scale height (Deng & Ridley, 2014;Deng et al., 2008).However, whole atmosphere models are usually built from existing GCMs, which are generally based on the shallow-atmosphere and hydrostatic approximation.Examples of currently available whole atmosphere models based on these assumptions include: the Whole Atmosphere Model (Akmaev et al., 2008;Fuller-Rowell et al., 2008), the Whole Atmosphere Community Climate Model with thermosphere and ionosphere extension (WACCM-X) (Liu et al., 2010(Liu et al., , 2018)), and the Ground to topside model of the Atmosphere and Ionosphere for Aeronomy (GAIA) (Fujiwara & Miyoshi, 2010;Jin et al., 2012).
The shallow-atmosphere approximation allows us to neglect the terms related to the spherical curvature of the atmosphere and the variations of the gravitational field (Thuburn & White, 2013).In addition, the contribution of the horizontal component of the Earth's angular velocity to the Coriolis acceleration is neglected (Bochert et al., 2019).These neglected terms impact the accuracy of the models.For example, the results of a scale study by White and Bromley (1995) showed that for diabatically driven flows in the tropics and planetary-scale flows the neglected terms of the Coriolis acceleration might be in the order of 10% (White & Bromley, 1995).This is not a small error considering the need, for example, for the precise prediction of satellite orbits.
Using the hydrostatic approximation the vertical momentum equation reduces to a balance between Earth's gravitational pull and pressure gradient force.Vertical forces such as Coriolis, centrifugal acceleration, and ion drag are thus not considered (Ridley et al., 2006).The hydrostatic assumption becomes problematic as we move toward higher-resolution simulations.At enhanced resolutions the models are able to capture flow features with comparable scales of motion in the horizontal and vertical direction, such as fine scale gravity waves.To correctly calculate the propagation of these gravity waves, non-hydrostatic effects need to be taken into account (Satoh et al., 2014).Therefore, non-hydrostatic dynamical cores are of great importance for high-resolution applications.Several non-hydrostatic whole atmosphere models have recently been developed including the Icosahedral Nonhydrostatic model (ICON) at the Deutscher Wetterdienst and the Max Planck Institute for Meteorology (Zängl et al., 2014), and the Navy Environmental Prediction System Using the NUMA Core (NEPTUNE) developed at the Naval Research Laboratory and the Naval Postgraduate School.WACCM-X (Liu et al., 2010(Liu et al., , 2018)), developed at the NSF National Center for Atmospheric Research (NCAR), extends from the Earth surface to the exobase (∼600 km).However, the dynamical cores used in current WACCM-X configurations (both finite volume, FV, and spectral element, SE) are based on the hydrostatic assumption.In the current study, we have developed and tested the Specified Chemistry Whole Atmosphere Community Climate Model (SC-WACCM) (Smith et al., 2014) with the non-hydrostatic Model for Prediction Across Scales-Atmosphere (MPAS-A) dynamical core (Skamarock et al., 2012).This is the first step toward the final goal of using MPAS-A as a dynamical core for WACCM-X.In addition, this work is closely connected with the System for Integrated Modeling of the Atmosphere (SIMA) initiative at NSF NCAR.The SIMA project aims to provide a unified and integrated approach to atmospheric modeling.There is a desire and need for modeling systems to reduce complexity and provide capabilities for broader, more comprehensive applications.One of the goals of the SIMA project is to enable the use of the non-hydrostatic MPAS-A dynamical core within the Community Earth System Science Model (CESM) framework (NCAR, 2020).This paper is structured as follows.We begin Section 2 with a description of the numerical models used and modifications made to perform the simulations in this study.In Section 3, we present results from running SC-WACCM with MPAS-A for the first time.The mean zonal wind and temperature climatology from SC-WACCM/ MPAS-A is compared with the results from SC-WACCM using the finite volume and spectral elements dynamical cores, as well as available observation.To better understand the wind and temperature climatology, a preliminary wave forcing analysis is also performed in this section.Summary remarks are included in Section 4.

WACCM Dynamical Cores
WACCM (Gettelman et al., 2019;Marsh et al., 2013) is one of the atmosphere components of the NSF NCAR CESM.It has a vertical domain spanning the range of altitude from the Earth's surface to 5.9 × 10 6 hPa (∼140 km).WACCM can run either with interactive chemistry or with specified chemistry.The latter configuration (SC-WACCM), is used in this study.As described in Smith et al. (2014), radiatively active atmosphere constituents, including water vapor, CO 2 , CH 4 , ozone, N 2 O, CFC-11, CFC-12, NO, O, and O 2 , are specified in the model.This atmospheric model is coupled to land, data ocean, and sea-ice components.For a detailed description of this model please refer to Smith et al. (2014).
The FV dynamical core has been used in the earlier versions of WACCM, and it uses a regular latitude-longitude grid.One main disadvantage of this type of grid is the decreasing longitudinal grid spacing as the poles are approached (often referred to as polar singularity).Polar filtering must be used for these types of grids to ensure numerical stability.However, polar filtering degrades the model performance at higher spatial resolution.More recently, a new dynamical core that has been used with WACCM is the SE core (Lauritzen et al., 2018).This dynamical core works with a quasi-uniform cubed-sphere grid which eliminates the polar singularity problem, and as a result, the need for polar filters.This allows the SE core to be used successfully in high-resolution simulations (Liu et al., 2014).Both the SE and the FV models are hydrostatic and use a sigma-pressure vertical coordinate system.
The focus of this work has been on developing and adapting MPAS-A to operate as a dynamical core for WACCM.MPAS-A is a fully compressible non-hydrostatic model using the finite-volume method and is described in detail in Skamarock et al. (2012).The model uses a C-grid staggering of the prognostic variables, and a horizontally unstructured spherical centroidal Voronoi mesh discretizes the sphere.The unstructured Voronoi meshes allow the use of variable horizontal resolution.The vertical discretization uses a terrain-following heightbased coordinate, and it lessens the impact of the small-scale terrain structure by allowing for the progressive smoothing of the coordinate surfaces with height (Klemp, 2011).MPAS-A applies a third-order Runge-Kutta scheme and explicit time-splitting technique for temporal integration (Klemp et al., 2007;Wicker & Skamarock, 2002).The time-splitting technique integrates gravity waves and horizontally propagating acoustic waves employing smaller explicit sub-steps within the Runge-Kutta step along with implicit integration of the vertically propagating acoustic modes.Some key developments were required for the coupling of MPAS dynamical core to the Community Atmosphere Model (CAM) physics.The coupling between a z-based dynamical core (constant volume) with a p-based (constant pressure) physics package such as CAM is non-trivial in terms of ensuring thermodynamic and energetic consistency (Lauritzen et al., 2022).Please see Huang et al. (2022) for details on this.
The rigid-lid model top in the WACCM configuration of MPAS-A used in this study is at approximately 135.5 km Above Sea Level.To maintain numerical stability in the presence of large winds in the lower thermosphere, we have reduced the transport time step from 900 s (the default for the ∼120 km mesh) to 600 s.We also employed another modification which involves a generalization of the vertically semi-implicit acoustic and gravity wave integration in MPAS-A.The semi-implicit off-centering is given in equations ( 14) through ( 16) in Klemp et al. (2007), and the weighting function for the old and new time levels is given in ( 17) where, for a variable ϕ, these weights are: The coefficient β s is typically a constant in MPAS-A and its default value is 0.1.However, in this MPAS-A/ WACCM configuration we have modified the MPAS-A solver to allow β s to vary with height, where β s = 0.1 through the troposphere and stratosphere, and transitions to a value of 0.5 in the lower thermosphere.The transition occurs between 40 and 80 km and uses a sine function to obtain a smooth transition.This modification to the off-centering of the vertically implicit acoustic time steps allows for a better control of the acoustic energy in the thermosphere and will produce more dissipation in the acoustic modes at upper levels.Hydrostatic dynamical cores used in the WACCM configuration, for example, the FV or SE dynamical cores, typically use increased horizontal filters (in addition to time-step reductions) to maintain stability, but with the changes to β s we have not had to increase the horizontal spatial filters in MPAS-A for the WACCM application.

Description of Simulations
In order to validate the WACCM/MPAS-A platform a 1 year simulation was conducted.The mean zonal wind and temperature climatology from WACCM/MPAS-A is compared with the results from WACCM using the FV and SE dynamical cores.In these simulations, the atmospheric models have a horizontal resolution of nominally 1°.
For MPAS this corresponds to the icosahedral mesh with 40,962 columns and approximately 120-km cell-center spacing; for FV this corresponds to the 0.9°× 1.25°latitude-longitude grid; for SE this is the ne30 cubed-sphere mesh.All the configurations used in this study have 70 vertical levels.At these resolutions gravity wave parameterization is needed.In the parameterization, wave sources from orography, deep convection, and frontogenesis are taken into consideration (Gettelman et al., 2019;Richter et al., 2010).Gravity wave forcing is a key driver of the wind and temperature structure in the middle atmosphere.Hence, gravity wave forcing from these simulations are also compared.A summary of the model configurations used in this study is presented in Table 1.For the FV and SE models, we have used the default settings for these dynamical cores as available in the released version of CESM2.1.It should be noted that during the current development of CESM3 (ongoing at the time of writing) several steps have been taken to increase the stability of the SE dynamical core so that longer time-steps than what is seen in Table 1 can be used.

Results
In this section, we will first present the zonal mean climatology from the WACCM/MPAS simulations and compare that with simulation results from WACCM/FV and WACCM/SE, as well as available observation.Next, we will present the preliminary analysis of the wave forcing.It should be noted that there are many differences between the three dynamical cores used in this study that could contribute to the differences in the simulation results presented in the following sections.Such as differences in the vertical coordinates, horizontal staggering of the variables, and dissipation configurations.In this paper we will not be presenting a model comparison taking these differences into account.Rather this study aims to report the steps taken to develop WACCM with MPAS and show that the preliminary simulations results agree well with available climatology, as well as results from other dynamical cores, when compared qualitatively.

Zonal Mean Wind and Temperature Climatology
Here, we present the WACCM/MPAS zonal mean climatology of zonal winds and temperature.First we will present the zonal mean zonal wind from the WACCM/MPAS simulations and compare them with the other two dynamical cores, as well as with climatology and recent radar observations.The data set produced and presented in Swinbank and Ortland (2003) describes the monthly zonal mean zonal winds from the surface to the upper mesosphere.Wind measurements are a combination of measurements from the High Resolution Doppler Imager (HRDI) and the Met Office stratospheric data assimilation system results.In order to bridge the gap between the stratospheric winds and HRDI mesospheric winds, the Upper Atmosphere Research Satellite Reference Atmosphere Project (URAP) temperature data was used to derive balanced winds.Figure 1 shows the cross sections of the merged wind data for the months of March, June, September, and December in the URAP baseline year, from April 1992 to March 1993 (Swinbank & Ortland, 2003).The areas between the dashed lines in 1 are locations where URAP data are sparse and values are either interpolated or extrapolated from other regions.
The URAP zonal wind has several notable futures in the mesosphere: zonal mean zonal wind reverses from eastward to westward in the winter hemisphere at a higher altitude (between 0.001 and 0.0001 hPa, approximately 100 km), and from westward to eastward in the summer hemisphere at a lower altitude (between 0.01 and 0.001 hPa, approximately 85 km).We also see the asymmetry in the strength of the reversal.The strength of the reversal is stronger in the summer hemisphere in both June and December.This wind climatology in the Mesosphere and Lower Thermosphere (MLT) region is known but not well reproduced in parameterized models such as WACCM.Using parameterization we get very symmetric wind reversal both in height and strength.It should also be mentioned that the URAP zonal wind climatology may contain a daytime aliasing of the diurnal tide signal, as discussed in Swinbank and Ortland (2003).This feature is a limitation of the URAP climatology.
The mean zonal wind results from the WACCM simulations for the solstices and equinoxes are presented in Figures 2 and 3, respectively.In these figures, both the color and line contour correspond to the mean zonal wind.
The features of the zonal mean wind in WACCM/MPAS, such as stratospheric jets and mesospheric wind reversals, are generally similar to WACCM/FV and WACCM/SE.The same parameterization configuration is used in all these cases.One must note that the 70 level WACCM used in this study is only scientifically supported with the FV dynamical core (Gettelman et al., 2019).The parameter tuning for the SE and MPAS dynamical core is beyond the scope of this manuscript.
A noticeable difference between the zonal winds in the WACCM/MPAS simulations compared to observation is the extension of the tropospheric westerly jets in the Southern Hemisphere into the lower stratosphere in December.That is seen somewhat in the WACCM/SE simulation, but less in the WACCM/FV simulations.At the solstices (June and December), as seen in Figure 2, there are strong jets through-out the stratosphere in the winter hemisphere.The strength of the winter stratospheric jets in December in the WACCM simulations is underestimated for all three dynamical cores, by about 30 ms 1 in FV and SE, and 20 ms 1 in MPAS in comparison to the URAP climatology.On the other hand, the strength of the winter stratospheric jets in June is overestimated in comparison to the observation for FV and SE (by approximately 10 and 20 ms 1 , respectively), but is comparable to URAP climatology for MPAS.The strength of the summer stratospheric wind at subtropical latitudes in June is underestimated for all three dynamical cores in comparison to the URAP climatology (by approximately 10 ms 1 for MPAS).The strength of the summer stratospheric jets at subtropical latitudes in December is overestimated for SE (by 10 ms 1 ) and underestimated for the FV and MPAS dynamical cores (by approximately 10 and 20 ms 1 , respectively).In addition, in December the summer jet shows a greater split between subtropical and high latitudes than seen in observations, and the strongest part of the jet is located in the Tropics (0-30°S), rather than near 60°S.
A wind reversal at the mesopause is observed in the climatology.All three models were able to reproduce the wind reversals seen in the observations at both solstices in the summer hemisphere.Although the wind reversal in the winter hemisphere is captured in June, we only see a slow down of the jets in the winter hemisphere in December.A stronger eastward reversal in the summer hemisphere is seen in all three models.The wind reversal occurs at roughly the same height for all the models (between 0.1 and 0.01 hPa, approximately 70 km).In both June and December, the wind structures are more symmetric in the model results than what is seen in the climatology.The lower thermospheric winds in the winter hemisphere in the WACCM simulations at the solstices are in reasonable agreement with observations, there is a greater split between the tropical and high latitudes wind speeds in the June simulation results in comparison to the observation.In all three models the strongest part of the jet in the winter hemisphere in June, is seen near 60°S rather than the Tropics.
At the equinoxes (March and September), as seen in Figure 3, the zonal mean winds are generally weaker than the solstices.In September the strength of the westerly jet in the Southern Hemisphere is underestimated by the FV dynamical core by approximately 20 m/s and overestimated by the SE and MPAS dynamical cores by approximately 10 and 20 m/s, respectively.In March, the strength of the jet in the Southern Hemisphere is closer to climatology for all three dynamical cores.At the equinoxes, MPAS has the strongest stratospheric jet in the Southern Hemispheres, in comparison to the other two dynamical cores.In the Northern Hemisphere, all three dynamical cores overestimate the strength of the jets with respect to observations.The zonal mean temperature at the solstices and equinoxes from the three dynamical cores is presented in Figures 4 and 5, respectively.The color and line contour both correspond to the zonal mean temperature values in Kelvin.In general we see a good agreement from the WACCM/MPAS results and the simulation results from the FV and SE dynamical cores.All three dynamical cores were able to reproduce the cold summer mesopause.
The similarities and differences between the simulation results from the different dynamical cores in the zonal mean wind field also manifest themselves in the temperature field.At the equinoxes, the stratospheric temperatures in the Southern Hemisphere in MPAS reflects the differences in the zonal wind field discussed previously.
The lower stratospheric MPAS temperatures in the Southern Hemisphere are associated with the stronger jets seen in the simulation in MPAS in comparison to the other two dynamical cores.At the solstices, the mesospheric temperature in the summer polar region is much warmer in MPAS.The mesospheric temperature is approximately 20 K warmer in this region in the MPAS simulation results in June and approximately 10 K warmer in December.The warmer mesospheric temperatures in MPAS manifest themselves as a weaker jet in the summer polar region.On the other hand the stratospheric temperature in the summer polar region is cooler in MPAS in June and December in comparison to the other two dynamical cores.The stratosphere temperature difference is most evident in June, where we see an approximate difference of 10 K.In addition, the tropopause in MPAS appears to be warmer than the other two dynamical cores in the summer polar region.The colder lower stratospheric temperatures in MPAS are associated with the stronger jet seen in the simulation in the Southern Hemisphere in December and the further extension of the tropospheric jets into the stratosphere, as discussed previously.
From temperature observations by the Sounding of the Atmosphere using the Broadband Emission Radiometry (SABER), the minimum summer mesopause in the polar region is warmer in December than in June, as can be seen in Figure 6 ( Xu et al., 2007).In addition, we expect the lowest mesopause altitude to be higher in December in comparison to June.The summer mesopause height in the polar region from the WACCM/MPAS simulation is higher in December.However, the summer polar mesopause temperatures from the WACCM/MPAS simulations in December (∼150 K) is lower than in June (∼160 K).Both the Southern and Northern Hemisphere's polar temperature at the summer mesopause are ∼10-20 K warmer in the MPAS simulations in comparison to the SABER mesopause temperatures presented by Xu et al. (2007).A similar trend is also seen in the SE and FV simulation results, although the temperatures are even warmer in the summer polar region in the MPAS results, as discussed earlier.It should be noted that these results were obtained using the default settings of the gravity wave parameterization, and no tuning has been attempted at this point.

Wave Forcing
The wind and temperature climatology in the middle and upper atmosphere are affected by both the parameterized and resolved wave forcing.Therefore, to obtain further insight into the dynamics of the middle and upper atmosphere in WACCM/MPAS simulation, we have performed an analysis of the parameterized and resolved wave forcing.The parameterized and resolved wave forcing from the WACCM/MPAS simulations are compared with the from the FV and SE dynamical cores.The parameterized gravity wave forcing is available directly from the simulations.The Transformed Eulerian Mean (TEM) method (Andrews et 1987) is used for calculating the momentum flux of the resolved waves and their divergence.
The resolved wave forcing at the solstices and equinoxes from the three dynamical cores are shown in Figures 7 and 8, respectively.The parameterized wave forcing at the solstices and equinoxes from the three dynamical cores are shown in Figures 9 and 10, respectively.In these figures the line contour correspond to the mean zonal wind, while the color contour correspond to the wave forcing.The features of wave distribution are generally similar in all three models in the middle and upper atmosphere.In the following we will discuss how these forcings are related to the wind and temperature structure presented in the previous section.
As can be seen in Figures 7-10, the overall wave activity continues to increase with altitude in the MLT region.
In December, the largest values of the resolved wave forcing is found at mid-high latitudes.The resolved forcing by planetary waves plays a dominant role in the winter stratosphere in driving the Brewer-Dobson circulation.
The weaker planetary wave forcing in the Southern winter hemisphere is evident, due primarily to the less land mass in the Southern Hemisphere.The forcing in the lower thermosphere is strong in both hemispheres, and resolved gravity waves likely contribute significantly.
The gravity wave forcing which is parameterized at this resolution is most important in the MLT region.For all three models, the wind reversals occurs at a similar altitude of approximately 0.01 hpa or 70 km.This wind reversal altitude corresponds well with the wave forcing for all three models.At the solstices, in the summer polar region of the mesosphere, we see a weaker parameterized wave forcing in MPAS compared to the FV and SE.This corresponds well with the warmer temperature in the summer polar region in the MPAS simulations in Figure 4.In December, the winter stratosphere/mesosphere jet is stronger in MPAS, despite the stronger parameterized wave forcing.This is because the resolved wave forcing in the winter mesosphere is weaker.

Summary and Conclusions
In this work we described the development of SC-WACCM with the non-hydrostatic MPAS-A dynamical core.An analysis of the mean zonal climatology was conducted to test this new capability.The wind and temperature climatology was compared with available observations.The gross features of the climatology, such as the structure of the stratospheric jets and wind reversals near the mesopause are represented reasonably well with the new model.However, there were several differences in the WACCM/MPAS simulations in comparison to the observations.In December, the tropospheric westerly jet in the Southern Hemisphere extends further into the lower stratosphere in the WACCM/MPAS simulations in comparison to observation.At the solstices, the strength of the winter stratospheric jet is underestimated in December but overestimated in June in the WACCM/MPAS simulations in comparison to observations.MPAS is not able to capture the asymmetry seen in the observation in the wind reversal in the mesosphere.In addition, MPAS is not able to reproduce the wind reversal in the winter hemisphere in December.The mean zonal wind and temperature climatology were also compared with results from SC-WACCM using the FV and SE dynamical cores.The location of the wind reversal are almost at the same altitude for all three models.Similar to MPAS, the FV and SE models are not able to capture the asymmetry in the mesospheric wind reversal.
The features in the zonal mean wind field manifest themselves in the temperature field as well: at solstices, (a) the lower stratosphere is cooler in MPAS near the summer polar region; (b) the mesosphere is warmer in MPAS near the summer polar region; and (c) the June summer mesopause is warmer.It is noted that no gravity wave tuning has been done for WACCM/MPAS to adjust the wind and temperature climatology.As mentioned previously, it is beyond the scope of this manuscript to explain simulation differences in terms of the discretizations used in the different dynamical cores (and their conservation properties) and no attempt has been made to tune the simulations for increased fidelity However, we have performed an analysis of the parametrized and resolved wave forcing in an attempt to quantify the impact of the resolved and sub-grid waves into the formation of the zonal mean climatology simulated by WACCM in the middle and upper atmosphere.Parameterized and resolved wave forcing from the WACCM/MPAS simulations are compared with the WACCM/FV and WACCM/SE results.The wind reversal seen in MPAS corresponds well with the wave forcing.The features of wave distribution are generally similar in all three models with minor differences.At the solstices, in the summer polar region of the mesosphere, we see a weaker parameterized wave forcing in MPAS which corresponds well with the warmer temperature in the summer polar region in the MPAS simulations, mentioned earlier.
The non-hydrostatic capability will be important for future high resolution simulations at convective scales.A global non-hydrostatic model will be a powerful tool for studying atmospheric gravity waves and their global effects at high resolutions.In addition, comparison of non-hydrostatic and hydrostatic dynamical cores will allow the assessment of the impact of non-hydrostatic effects on various spatial and temporal scales.Currently we are employing a shallow-atmosphere version of MPAS-A in these WACCM/MPAS-A simulations.For longer term geospace development we intend to employ the deep atmosphere version of MPAS-A (Klemp &  Skamarock, 2021Skamarock, , 2022;;Skamarock et al., 2021) with WACCM-X.The work presented here is the first step toward the final goal of using the non-hydrostatic MPAS-A with WACCM-X for geospace applications.

Data Availability Statement
For the simulations in this work we have used a development branch of NSF NCAR CESM/WACCM which is available at https://zenodo.org/records/10927619(Kamali et al., 2024).The model outputs used for this study are available through GLOBUS at https://tinyurl.com/tdw6kkba.

Figure 2 .
Figure 2. Zonal mean zonal wind (m/s) at the solstices (June and December) from the Whole Atmosphere Community Climate Model (WACCM)/finite volume, WACCM/spectral element, and WACCM/MPAS simulations.

Figure 3 .
Figure 3. Zonal mean zonal wind (m/s) at the equinoxes (March and September) from the Whole Atmosphere Community Climate Model (WACCM)/finite volume, WACCM/spectral element, and WACCM/MPAS simulations.

Figure 4 .
Figure 4. Zonal mean temperature (K) at the solstices (June and December) from the Whole Atmosphere Community Climate Model (WACCM)/finite volume WACCM/spectral element, and WACCM/MPAS simulations.The contour intervals increase from 10 to 100 K above 270 K.

Figure 5 .
Figure 5. Zonal mean temperature (K) at the equinoxes (March and September) from the Whole Atmosphere Community Climate Model (WACCM)/finite volume WACCM/spectral element, and WACCM/MPAS simulations.The contour intervals increase from 10 to 100 K above 270 K.

Figure 6 .
Figure 6.Profiles of diurnally averaged temperature from SABER at 80°for both hemispheres for December and June (solid line for Southern Hemisphere and dashed line for Northern Hemisphere).Reproduced from Xu et al. (2007).

Figure 7 .
Figure 7. Resolved wave forcing (m/s/d) at the solstices (June and December) from the Whole Atmosphere Community Climate Model (WACCM)/finite volume, WACCM/spectral element, and WACCM/MPAS simulations.Line contour correspond to the mean zonal wind, while the color contour correspond to the wave forcing.

Figure 8 .
Figure 8. Resolved wave forcing (m/s/d) at the equinoxes (March and September) from the Whole Atmosphere Community Climate Model (WACCM)/finite volume, WACCM/spectral element, and WACCM/MPAS simulations.Line contour correspond to the mean zonal wind, while the color contour correspond to the wave forcing.

Figure 9 .
Figure 9. Parameterized wave forcing (m/s/d) at the solstices (June and December) from the Whole Atmosphere Community Climate Model (WACCM)/finite volume, WACCM/spectral element, and WACCM/MPAS simulations.Line contour correspond to the mean zonal wind, while the color contour correspond to the wave forcing.

Figure 10 .
Figure 10.Parameterized wave forcing (m/s/d) at the equinoxes (March and September) from the Whole Atmosphere Community Climate Model (WACCM)/finite volume, WACCM/spectral element, and WACCM/MPAS simulations.Line contour correspond to the mean zonal wind, while the color contour correspond to the wave forcing.

Table 1
Summary of Model Configurations Used in This Study