+ All Categories
Home > Documents > ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and...

ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and...

Date post: 17-Dec-2015
Category:
Upload: olivia-freeman
View: 225 times
Download: 0 times
Share this document with a friend
20
ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1 , Günther Zängl 1 , and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute for Meteorology 13 th EMS Annual Meeting 09 – 13 September 2013, Reading, United Kingdom
Transcript
Page 1: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

ICON The new global nonhydrostatic model of DWD and

MPI-M

Daniel Reinert1, Günther Zängl1, and the ICON-team1,2

1Deutscher Wetterdienst / 2Max-Planck-Institute for Meteorology

13th EMS Annual Meeting

09 – 13 September 2013, Reading, United Kingdom

Page 2: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

ICON – ICOsahedral Nonhydrostatic Model

Daniel Reinert – 12.09.2013 2

Joint development project of DWD and Max-Planck-Institute for Meteorology for building a next-generation global NWP and climate modelling system

Atmosphere and ocean model

Outline

I. Project goals

II. Horizontal grid structure and accompanying problems

III. ICON NWP physics suite

IV. Selected results

V. Roadmap and Summary

DWDMPI

Page 3: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

I. Primary development goals

Daniel Reinert – 12.09.2013 3

At DWD: • Replace current global model GME • Replace regional model COSMO-EU by a high-

resolution window over Europe.

At MPI-M: • Use ICON as dynamical core of an Earth System

Model (MPI-ESM2) Horizontal grid with nest over Europe

Improved conservation properties (at least mass) and consistent tracer transport (tracer air-mass consistency)

Applicability on a wide range of scales from 100 km to 1 km

Scalability and efficiency on massively parallel computer architectures with O(104 +) cores

Local refinement/nesting capability

Page 4: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

II. ICON’s unstructured grid

Primal cells: triangles

uses icosahedron for macro triangulation

C-type staggering:

local subdomains (“nests”)

4Daniel Reinert – 12.09.2013

local domain(s)

global domain

velocity at edge midpoints mass at cell circumcenter

Triangular C-Grid

Page 5: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

Equations (dry adiabatic) and solver

Fully compressible nonhydrostatic vector invariant form, shallow atm.

5Daniel Reinert – 12.09.2013

Solver: Finite volume/finite difference discretization (mostly 2nd order)

Two-time level predictor-corrector time integration

Vertically implicit (vertical sound-wave propagation)

Fully explicit time integration in the horizontal (at sound wave time step; not split explicit!)

Mass conserving

Page 6: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

Checkerboard noise on triangular C-Grid

Main problem with triangular C-grid: suffers from spurious computational mode (e.g. Danilov (2010)), triggered by the discretized divergence operator (Wan (2013))

6Daniel Reinert – 12.09.2013

Divergence operator: applies the Gauss theorem

Truncation error (Wan (2013)):

Only 1st order accurate on triangular C-grid

Error changes sign from upward- to downward pointing triangle checkerboard

Example for synthetic velocity field (Wan, 2013)

Page 7: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

Controlling the checkerboard noise

Daniel Reinert – 12.09.2013 7

Goal: Eliminate 1st order error

Basic idea: Divergence averaging

I: Compute standard 1st order divII: Compute divergence estimate

based on immediate neighbors (2nd order bilinear interpolation)

III Averaging:

2nd order accurate for isosceles triangles

Page 8: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

Example: Baroclinic wave

8Daniel Reinert – 12.09.2013

Jablonowski-Williamson (2006) baroclinic wave test case

PS T

Page 9: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

Example: Baroclinic wave

9Daniel Reinert – 12.09.2013

Standard divergence operator Divergence averaging

Jablonowski-Williamson (2006) baroclinic wave test case

divdiv

“checkerboard” noise

PS T

Page 10: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

Daniel Reinert – 12.09.2013 10

III. ICON NWP-physics

Process Author Scheme Origin

Radiation Mlawer et al. (1997)Barker et al. (2002)

RRTM ECHAM6

Non-orographic gravity wave drag

Scinocca (2003)Orr, Bechthold et al. (2010)

wave dissipation at critical level IFS

Cloud cover Köhler et al. (new development) diagnostic (later prognostic) PDF ICON

Microphysics Doms and Schättler (2004)Seifert (2010)

prognostic: water vapour, cloud water, cloud ice, rain, snow

COSMO

Saturation adjustment

Blahak (2010) isochoric adjustment COSMO

Convection Tiedtke (1989)Bechthold et al. (2008)

mass-flux shallow and deep IFS

Sub-grid scale orographic drag

Lott and Miller (1997) blocking, GWD IFS

Turbulent transfer / diffusion

Raschendorfer (2001) prognostic TKE COSMO

Soil/surfaceHeise and Schrodin (2002)Mironov and Ritter (2004)Mironov (2008)

TERRA (tiled + multi-layer snow)SEAICE FLAKE(fresh water lake scheme)

GME/COSMO

Page 11: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

every 30min

Reduced grid for radiation

Daniel Reinert – 12.09.2013 11

upscaling

downscaling

Rad

iati

ve t

ran

sfer

co

mp

uta

tio

ns

Hierarchical structure of the triangular mesh is very attractive for calculating physical processes (e.g. radiative transfer) with different spatial resolution compared to dynamics.

Radiation step

Em

pir

ical

co

rrec

tio

ns

Page 12: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

Proof of concept

Daniel Reinert – 12.09.2013 12

net surface shortwave flux (reduced – full grid) average over 30 x 48h forecast runs in June 2012

Reduced radiation grid currently generates positive bias in

Avg: 1.57

Page 13: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

Flat-MPI performance

13Daniel Reinert – 12.09.2013

Test setup: ICON RAPS 2.0, IBM Power 720/10/5 km, 8h forecast, reduced radiation grid

(S. Körner, DWD, 03/2013)

Recall goal: scalability up to O(104+) cores

40961024 1024 4096

tim

e (s

)

MPI tasks

Page 14: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

IV. Selected results of NWP test suite

Real-case 7-day forecasts with interpolated IFS analysis data

WMO standard verification against IFS analysis on 1.5° lat/lon grid.

Comparison against GME reference experiment with interpolated IFS analysis data.

Daniel Reinert – 12.09.2013 14

Basic requirement for operational use of ICON

ICON must outperform GME in terms of forecast quality/scores

ICON40L90 GME40L60

hor. resolution 40 km 40 km

vertical levels 90 60

top height 75 km 36 km

analysis data IFS IFS

Page 15: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

Verification: Surface Pressure, January 2012

Daniel Reinert – 12.09.2013 15

ICONGME

against IFS

Region: Northern hemisphere (NH)

SH: 21%

Verification: G. Zängl, U. Damrath, 08/2013 (DWD)

Page 16: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

Verification: Geopot 500 hPa, January 2012

Daniel Reinert – 12.09.2013 16

ICONGME

against IFS

Region: Northern hemisphere (NH)

SH: 9.4%

Verification: G. Zängl, U. Damrath, 08/2013 (DWD)

Page 17: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

Verification: Rh 700 hPa, January 2012

Daniel Reinert – 12.09.2013 17

ICONGME

against IFS

Region: Tropics (Tr)

Verification: G. Zängl, U. Damrath, 08/2013 (DWD)

ICON shows strong positive moisture bias in the tropics

Page 18: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

V. Roadmap towards operational application

18Daniel Reinert – 12.09.2013

Page 19: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

Summary

Verification results are mostly exceeding those of GME, but there are still some weaknesses/biases e.g. moisture field

Technical parts scale on massively parallel systems (I/O still needs performance improvements)

Optimization of forecast quality still ongoing

Tests with own 3D-Var data assimilation have started recently.

19Daniel Reinert – 12.09.2013

ICON is entering the home stretch for becoming operational

Page 20: ICON The new global nonhydrostatic model of DWD and MPI-M Daniel Reinert 1, Günther Zängl 1, and the ICON-team 1,2 1 Deutscher Wetterdienst / 2 Max-Planck-Institute.

Thank you for your attention !!


Recommended