[A-AS05] 大規模な水蒸気場と組織化した雲システム

[AAS05-01] Coupling Momentum and Variance in a MMF with a 2D CRM

★Invited Papers

*Walter Michael Hannah¹、Kyle Pressel²、Mikhail Ovchinnikov²、Qiu Yang²、Ruby Leung²、Gregory Elsaesser³、Matthew Norman⁴、Ben Hillman⁵ (1.Lawrence Livermore National Laboratory、2.Pacific Northwest National Laboratory、3.NASA、4.Oak Ridge Leadership Computing Facility、5.SNL)

キーワード：atmospheric modelling, multiscale modelling framework, convective momentum transport, GPU acceleration, tropical variability

An atmospheric multiscale modeling framework (MMF) allows an explicit representation of convection by coupling a cloud resolving model (CRM) with a general circulation model (GCM). This avoids the need for conventional parameterizations of clouds and turbulence. Traditional MMF implementations often use a 2D CRM with periodic boundary conditions to constrain the computational cost, but some consequences of this approach are that convective momentum transport cannot be explicitly represented and small-scale fluctuations within the CRM cannot be transported by the large-scale flow. These limitations can be problematic because they force the model to ignore key atmospheric processes. Two enhancements have been implemented in the E3SM-MMF to address these deficiencies. The first is the momentum transport scheme of Tulich (2015) in which the large-scale momentum field is represented in the CRM by two scalars, which allows the 3D momentum transport to be approximated by a 2D CRM. The second feature is a scheme that allows small scale fluctuations within the CRM to be transported between CRM instances on the global grid.