A Look at High-Order Finite- Volume Schemes for Simulating Atmospheric Flows Paul Ullrich University of Michigan.

Slides:



Advertisements
Similar presentations
PARMA UNIVERSITY SIMULATIONS OF THE ISOLATED BUILDING TEST CASE F. AURELI, A. MARANZONI & P. MIGNOSA DICATeA, Parma University Parco Area delle Scienze.
Advertisements

A High-Order Finite-Volume Scheme for the Dynamical Core of Weather and Climate Models Christiane Jablonowski and Paul A. Ullrich, University of Michigan,
Geophysical Fluid Dynamics Laboratory Review June 30 - July 2, 2009 Geophysical Fluid Dynamics Laboratory Review June 30 - July 2, 2009.
Joint Mathematics Meetings Hynes Convention Center, Boston, MA
2ª aula Evolution Equation. The Finite Volume Method.
Finite Volume II Philip Mocz. Goals Construct a robust, 2nd order FV method for the Euler equation (Navier-Stokes without the viscous term, compressible)
(c) MSc Module MTMW14 : Numerical modelling of atmospheres and oceans Staggered schemes 3.1 Staggered time schemes.
RAMS/BRAMS Basic equations and some numerical issues.
Algorithm Development for the Full Two-Fluid Plasma System
Günther Zängl, DWD1 Improvements for idealized simulations with the COSMO model Günther Zängl Deutscher Wetterdienst, Offenbach, Germany.
ICONAM ICOsahedral Non-hydrostatic Atmospheric Model -
Coupling Continuum Model and Smoothed Particle Hydrodynamics Methods for Reactive Transport Yilin Fang, Timothy D Scheibe and Alexandre M Tartakovsky Pacific.
Adaptive mesh refinement for discontinuous Galerkin method on quadrilateral non-conforming grids Michal A. Kopera PDE’s on the Sphere 2012.
Conservative Cascade Remapping between Spherical grids (CaRS) Arunasalam Rahunanthan Department of Mathematics, University of Wyoming. NCAR, SIParCS Student.
1 Internal Seminar, November 14 th Effects of non conformal mesh on LES S. Rolfo The University of Manchester, M60 1QD, UK School of Mechanical,
High-Order Adaptive and Parallel Discontinuous Galerkin Methods for Hyperbolic Conservation Laws J. E. Flaherty, L. Krivodonova, J. F. Remacle, and M.
Shallow water equations in 1D: Method of characteristics
1cs533d-term Notes  list Even if you’re just auditing!
Modeling Fluid Phenomena -Vinay Bondhugula (25 th & 27 th April 2006)
Application of the CSLR on the “Yin-Yang” Grid in Spherical Geometry X. Peng (Earth Simulator Center) F. Xiao (Tokyo Institute of Technology) K. Takahashi.
Non-hydrostatic algorithm and dynamics in ROMS Yuliya Kanarska, Alexander Shchepetkin, Alexander Shchepetkin, James C. McWilliams, IGPP, UCLA.
A TWO-FLUID NUMERICAL MODEL OF THE LIMPET OWC CG Mingham, L Qian, DM Causon and DM Ingram Centre for Mathematical Modelling and Flow Analysis Manchester.
Numerical Schemes for Advection Reaction Equation Ramaz Botchorishvili Faculty of Exact and Natural Sciences Tbilisi State University GGSWBS,Tbilisi, July.
Discontinuous Galerkin Methods for Solving Euler Equations
P. Ackerer, IMFS, Barcelona About Discontinuous Galerkin Finite Elements P. Ackerer, A. Younès Institut de Mécanique des Fluides et des Solides,
Zhaorui Li and Farhad Jaberi Department of Mechanical Engineering Michigan State University East Lansing, Michigan Large-Scale Simulations of High Speed.
A COMPARISON OF SOME SHALLOW WATER TEST CASES IN HOMME USING NUMERICALLY AND ANALYTICALLY COMPUTED TRANSFORMATIONS Numerical Approximations of Coordinate.
The Finite Volume Method
Most physically significant large-scale atmospheric circulations have time scales on the order of Rossby waves but much larger than the time scales.
Solution of the St Venant Equations / Shallow-Water equations of open channel flow Dr Andrew Sleigh School of Civil Engineering University of Leeds, UK.
A Novel Wave-Propagation Approach For Fully Conservative Eulerian Multi-Material Simulation K. Nordin-Bates Lab. for Scientific Computing, Cavendish Lab.,
Applied NWP [1.2] “Once the initialization problem was resolved in the 1960s, models based on the primitive equations gradually supplanted those based.
1 Discretization of Fluid Models (Navier Stokes) Dr. Farzad Ismail School of Aerospace and Mechanical Engineering Universiti Sains Malaysia Nibong Tebal.
A conservative FE-discretisation of the Navier-Stokes equation JASS 2005, St. Petersburg Thomas Satzger.
A cell-integrated semi-Lagrangian dynamical scheme based on a step-function representation Eigil Kaas, Bennert Machenhauer and Peter Hjort Lauritzen Danish.
Discontinuous Galerkin Methods Li, Yang FerienAkademie 2008.
Discontinuous Galerkin Methods for Solving Euler Equations Andrey Andreyev Advisor: James Baeder Mid.
3.3.3: Semi-Lagrangian schemes AOSC614 class Hong Li.
7. Introduction to the numerical integration of PDE. As an example, we consider the following PDE with one variable; Finite difference method is one of.
© Fluent Inc. 11/24/2015J1 Fluids Review TRN Overview of CFD Solution Methodologies.
Recent Developments in the NRL Spectral Element Atmospheric Model (NSEAM)* Francis X. Giraldo *Funded.
Introducing Flow-er: a Hydrodynamics Code for Relativistic and Newtonian Flows Patrick M. Motl Joel E. Tohline, & Luis Lehner (Louisiana.
1 The Nonhydrostatic Icosahedral (NIM) Model: Description and Potential Use in Climate Prediction Alexander E. MacDonald Earth System Research Lab Climate.
FALL 2015 Esra Sorgüven Öner
AMS 691 Special Topics in Applied Mathematics Lecture 8
Discretization Methods Chapter 2. Training Manual May 15, 2001 Inventory # Discretization Methods Topics Equations and The Goal Brief overview.
CHANGSHENG CHEN, HEDONG LIU, And ROBERT C. BEARDSLEY
Governing Equations II
1 Application of Weighted Essentially Non-Oscillatory Limiting to Compact Interpolation Schemes Debojyoti Ghosh Graduate Research Assistant Alfred Gessow.
Numerical Algorithm Development and Testing in HYCOM.
Development of an Atmospheric Climate Model with Self-Adapting Grid and Physics Joyce E. Penner 1, Michael Herzog 2, Christiane Jablonowski 3, Bram van.
Deutscher Wetterdienst Flux form semi-Lagrangian transport in ICON: construction and results of idealised test cases Daniel Reinert Deutscher Wetterdienst.
Global variable-resolution semi-Lagrangian model SL-AV: current status and further developments Mikhail Tolstykh Institute of Numerical Mathematics, Russian.
Deutscher Wetterdienst 1FE 13 – Working group 2: Dynamics and Numerics report ‘Oct – Sept. 2008’ COSMO General Meeting, Krakau
Efficient Simulation of Large Bodies of Water by Coupling Two and Three Dimensional Techniques Geoffrey Irving Stanford University Pixar Animation Studios.
Representing Effects of Complex Terrain on Mountain Meteorology and Hydrology Steve Ghan, Ruby Leung, Teklu Tesfa, PNNL Steve Goldhaber, NCAR.
Lecture Objectives: - Numerics. Finite Volume Method - Conservation of  for the finite volume w e w e l h n s P E W xx xx xx - Finite volume.
1 On forward in time differencing: an unstructured mesh framework Joanna Szmelter Piotr Smolarkiewicz Loughborough University NCAR UK Colorado USA.
Reporter: Prudence Chien
Lecture 4: Numerical Stability
A TWO-FLUID NUMERICAL MODEL OF THE LIMPET OWC
Convergence in Computational Science
Objective Numerical methods.
Objective Numerical methods Finite volume.
Finite Volume Method Philip Mocz.
High Accuracy Schemes for Inviscid Traffic Models
topic11_shocktube_problem
Comparison of CFEM and DG methods
Low Order Methods for Simulation of Turbulence in Complex Geometries
Conservative Dynamical Core (CDC)
Presentation transcript:

A Look at High-Order Finite- Volume Schemes for Simulating Atmospheric Flows Paul Ullrich University of Michigan

Next Generation Climate Models High-order accurate Move away from latitude-longitude grids Utilize modern hardware (GPUs, Petascale computing) Adaptive mesh refinement?

The cubed sphere grid is obtained by placing a cube inside the sphere and “inflating” it to occupy the total volume of the sphere. Pros: –Removes polar singularities –Grid faces are individually regular Cons –Some difficulty handling edges –Multiple coordinate systems Many atmospheric models now utilize this grid. The Cubed Sphere Grid

Finite volume methods have several advantages over finite difference and spectral methods: –They can be used to conserve invariant quantities, such as mass, energy, potential vorticity or potential enstrophy. –Finite volume methods can be easily made to satisfy monotonicity and positivity constraints (i.e. to avoid negative tracer densities). –Lots of research has been done on finite volume methods in aerospace and other CFD fields. Why Finite Volumes?

Many atmospheric models make use of staggered grids (ie. Arakawa B,C,D-grids), where velocity components and mass- variables are located at different grid points. Staggered grids have certain advantages, such as better treatment of high-wavenumber wave modes. However, staggered grids have stricter timestep constraints. Unstaggered grids allow us to easily perform horizontal-vertical dimension splitting. Staggered grids also suffer from unphysical wave reflection at abrupt grid resolution discontinuities (on adaptive grids)… Unstaggered vs. Staggered Grids

The high-order upwind finite volume model consists of several components, a few of which will be covered here: The sub-grid-scale reconstruction The Riemann solver The implicit-explicit dimension-split integrator Finite Volume Formulation 1 2 3

1 Our sub-grid scale reconstruction can use only information on the cell- averaged values within each element. Cell 1Cell 2Cell 3Cell 4 Sub-Grid Scale Reconstruction

The least accurate and least computation-intensive method for building a sub-grid scale reconstruction assumes that all points within a source grid element share the same value. Sub-Grid Scale Reconstruction Piecewise Constant Method (PCoM) 1 Cell 1Cell 2Cell 3Cell 4

Increasing the accuracy of the method with respect to the reconstruction simply requires using increasingly high order polynomials for the sub-grid scale reconstruction. Sub-Grid Scale Reconstruction Piecewise Cubic Method (PCM) 1 Cell 1Cell 2Cell 3Cell 4 A cubic reconstruction will lead to a 4th order accurate scheme, if paired with a sufficiently accurate timestep scheme.

Since the reconstruction is inherently discontinuous at cell interfaces, we must solve a Riemann problem to obtain the flux of all conserved variables. The Riemann Solver 2 Cell 1Cell 2 ULUL URUR

A crude choice of Riemann solver can result in excess diffusion, which can severely contaminate the solution. The Riemann Solver 2 Rusanov Riemann solverAUSM + -up Riemann solver

Results: Shallow Water Model Williamson et al. (1992) Test Case 2 - Steady State Geostrophic Flow (  =45  ) Fluid Depth (h)

Results: Shallow Water Model

Williamson et al. (1992) Test Case 5 - Flow over Topography Total Fluid Depth (H)

Vertically propagating sound waves are a major issue for nonhydrostatic models. This suggests special treatment is required of the vertical coordinate. Vertical Discretization 3 Idea: Since we are using an unstaggered grid, its easy to split the horizontal and vertical integration and treat the vertical integration implicitly, even in the presence of topography. Since vertical columns are disjoint, each column only requires a single implicit solve; total matrix size = 5 x. In order to achieve high-order accuracy we use Implicit-Explicit Runge- Kutta-Rosenbrock (IMEX-RKR) schemes. The resulting method is valid on all scales, uses the horizontal timestep constraint, is high-order accurate and is only modestly slower than a hydrostatic model.

Care must be taken to choose a high-order-accurate timestepping scheme. Poor choices can lead to severely degraded model results. Vertical Discretization 3 1,2,3. Explicit steps 4. Implicit step 1,3,5. Explicit steps 2,4. Implicit steps

Temperature at 500m Results: 3D Nonhydrostatic Model Jablonowski (2011) Baroclinic Instability in a Channel

Summary Next generation atmospheric models will likely rely on high-order numerical methods to achieve accuracy at a reduced computational cost. We have successfully demonstrated a high-order finite volume method for the shallow-water equations on the sphere and for nonhydrostatic 2D and 3D modeling. Implicit-explicit Runge-Kutta-Rosenbrock (IMEX-RKR) methods are very good candidates for time integrators, and can likely be adapted to any unstaggered grid model (high-order FV, DG, SV).

Questions?

The Riemann solver introduces a natural source of damping, which can act to suppress oscillations in the divergence. The Riemann Solver 2 Advective Term (proportional to dm/dx) Diffusive Term (proportional to c dh 4 /dx 4 ) Example: Third-order reconstruction (parabolic sub-grid-scale) applied to the linear shallow-water equations plus Riemann solver.

Next Generation Climate Models Finite Volume High-order upwind High-order symmetric Compact Stencil Discontinuous Galerkin Spectral element / CG Spectral volume Semi-Lagrangian Advection Nonhydro- static Shallow Water Hydro- static