2:00 PM - 2:15 PM
[MGI35-02] Development of a global non-hydrostatic Martian atmospheric model and its high-resolution simulation
Keywords:Mars, atmosphere, global, non-hydrostatic, high-resolution simulation, model development
We are developing a non-hydrostatic global Martian atmospheric model (Martian SCALE-GM) which is suitable for large-scale parallel computation, targeting to perform high-resolution simulations described above on the post-K computer. SCALE-GM (http://scale.aics.riken.jp/) is being developed by using the dynamical core of NICAM (Tomita and Satoh, 2005; Satoh et al., 2008; Satoh et al., 2014), a non-hydrostatic model using a finite volume method in the icosahedral grid systems (Tomita et al., 2001, 2002), that has been used for simulations of Earth atmosphere and climate, and by aiming at sharing of physical process modules with the regional model (SCALE-RM) and application to other planetary atmospheres. We are developing Martian SCALE-GM by incorporating constants and physical process modules of the Martian atmosphere. The Martian physical modules are taken from DCPAM (https://www.gfd-dennou.org/library/dcpam/), an existing pan-planetary atmospheric general circulation model (GCM). DCPAM is a traditional, hydrostatic GCM using a spectral method for horizontal discretization.
In this study, we ported a Martian atmospheric radiation model (Forget et al., 1999) and a soil model from DCPAM to SCALE-GM and performed a test of temperature evolution in a vertical 1D-atmosphere and soil. The atmosphere model has 100 layers (Δz = 1 km), and distribution of dust is fixed with 0.2 optical depth. The soil model has 18 layers. Soil's heat capacity and thermal conductivity are set to 9.7×105 [J K−1 kg−1] and 0.076 [W m–1 K–1], respectively. Surface albedo is fixed at 0.2. For the calculation of surface fluxes and vertical diffusion, BH91B95 (Beljaars & Holstang, 1991; Beljaars, 1995) and MY2.5 (Mellor & Yamada, 1982) are used, respectively. The initial condition is the 200 K constant temperature atmosphere and soil. We have confirmed that both models (SCALE-GM and DCPAM) show almost the same temperature evolution.
Next, we performed 3D calculations under a Mars-like condition in 7 different horizontal resolutions and checked the dependence of the numerical solutions on the horizontal resolution. Horizontal grid intervals are 240×(1/2)n [km] (n = 0, 1, 2, 3, 4, 5, 6); i.e., 3.75 km at minimum. The horizontal diffusion representing turbulent mixing in sub-grid scales is given in the 3rd order Laplacian with the relaxation time for the shortest scale given by 100×(1/2)n [s]. The atmosphere is divided into 36 layers, which are concentrated near the surface. To avoid wave reflections at the model top, a sponge layer represented by 1st order Laplacian is set above 40 km altitude; the relaxation time for the shortest scale is given by 3600×(1/4)n [s] except for the case of Δx = 3.75 km, in which 3 s is used. The 4th order Runge-Kutta method is used for time-integration with a time step of 360×(1/2)n [s]. The surface albedo is set to 0.5. Neither topography nor condensation processes are included.
Numerical solutions showed that vertical convections due to diurnal solar heating did not appear clearly in the cases with Δx larger than 15 km, whereas they appeared in higher resolution cases. Furthermore, it was confirmed that the horizontal scale and occurrence local-time of the vertical convection are highly dependent on the horizontal resolution at least in this range.
(Figure: Horizontal resolution dependence of vertical wind. Snapshots at 2 km altitude after 30 days from the spring equinox of the Northern hemisphere.)