Michael Baldauf Deutscher Wetterdienst, Offenbach, Germany

26
19.09.200 6 Aktionsprogramm 2003 M. Baldauf, DWD 1 COSMO Priority Project: Further developments of the Runge- Kutta Time Integration Scheme COSMO General Meeting Bukarest, 18.-21.09.2006 Michael Baldauf Deutscher Wetterdienst, Offenbach, Germany

description

Michael Baldauf Deutscher Wetterdienst, Offenbach, Germany. COSMO Priority Project: Further developments of the Runge-Kutta Time Integration Scheme COSMO General Meeting Bukarest, 18.-21.09.2006. List of people contributing to the project: (alphabetical order) Michael Baldauf (DWD, D) - PowerPoint PPT Presentation

Transcript of Michael Baldauf Deutscher Wetterdienst, Offenbach, Germany

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 1

COSMO Priority Project:Further developments of the Runge-Kutta Time Integration Scheme

COSMO General Meeting

Bukarest, 18.-21.09.2006

Michael Baldauf Deutscher Wetterdienst, Offenbach, Germany

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 2

List of people contributing to the project: (alphabetical order)

• Michael Baldauf (DWD, D)• Gabriella Ceci (CIRA, I)• Guy deMorsier (MeteoCH, CH)• Jochen Förstner (DWD, D)• Almut Gassmann (Univ. Bonn, D) (FTE not counted)

• Paola Mercogliano (CIRA, I)• Thorsten Reinhardt (DWD, D)• Lucio Torrisi (CNMCA, I)• Pier Luigi Vitagliano (CIRA, I)• Klaus Stephan (DWD, D) (FTE not counted)

• Matthias Raschendorfer (DWD, D) (FTE not counted)

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 3

Task 1: Looking at pressure bias(Torrisi, Förstner)verifications of LM 7 km runs showed a higher positive pressure bias for the RK core than for the Leapfrog core, whereas other variables show comparable behaviour.

Work done:5-day verifications were done for several model

configurations

only little impact on PMSL by:• physics coupling• advection of qx (Bott, Semi-Lagrange)

most significant impact on PMSL:• new dynamical bottom boundary condition (DBBC)

( Task 6)• p‘T‘-dynamics in the RK-coreboth measurements reduce the pressure bias

verification area

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 4

Task 1: Looking at pressure bias

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 5

positive impact of p‘T‘-dynamics

Task 1: Looking at pressure bias

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 6

positive impact of DBBC

Task 1: Looking at pressure bias

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 7

Task 1: Looking at pressure bias

Status: verifications carried out

Work to do: longer verifications periods should be inspectedUpper air verificationsLateral boundary conditions pressure bias improved by another fast waves solver?? ( --> task 10)

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 8

Task 2: Continue RK case studies(Torrisi, deMorsier)extensive verification of the other tasks

Status: test case carried out; verifications were made

Work to do: test cases should be continued

Work done:case study 4.-8.Dez. 2004 (inversion in 800...1600m)was carried out: penetration of the stratus in Alpine valleys in 2km- sim. better performed by RK-core compared to LF-core

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 9

Status: • integral over a volume (arbitrary square-stone): ready

• Subr. init_integral_3D: define square-stone (in the transformed grid!), domain decomp. • Function integral_3D_total: calc. volume integral• Function integral_3D_cond: calc. vol. integral over individual processor

Work to do• flux integral over the surface

balance equation for scalar :

Task 3: Conservation(Baldauf)Tool for inspection of conservation properties will be developed. Integration area = arbitrarily chosen cubus (in the transformed grid, i.e. terrain-following)

temporal change

flux divergence

sources / sinks

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 10

Task 4: Advection of moisture quantities in conservation form(Förstner)implementation of a Courant-number independent advection algorithm for the moisture densities

Status: implemented schemes (Bott-2, Bott-4) behave well(Semi-Lagrange-scheme as a testing tool is also available)

task finished

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 11

Transport of TracerTransport of Tracerin a Real Case Flow Fieldin a Real Case Flow Field

initinit

Bott (2Bott (2ndnd))“Flux Form“Flux Form

- DIV”- DIV”+ Clipping+ Clipping

Bott (2Bott (2ndnd))“Conserv. “Conserv.

Form”Form”

semi-semi-LagrangeLagrange((tri-cubic)tri-cubic)+ Clipping+ Clipping

PP RK 2.1.4 + 2.1.3

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 12

Task 5: investigation of convergence(Ceci, Vitagliano, Baldauf)determination of the spatial and temporal order of convergence of the RK-scheme

in combination with advection schemes of higher order.

Planned test cases:• linear mountain flows (2D, 3D)• nonlinear mountain flows (dry case)• nonlinear mountain flows with precipitation

Status: implementation of LM and test environment. First tests with linear mountain flow.

Kinetic energy (v.l. 120,160)

8,700E-04

8,900E-04

9,100E-04

9,300E-04

9,500E-04

9,700E-04

0 2 4 6 8 10

Horizontal Resolution [km]

Kin

eti

c e

ne

rgy

[m

^2

/s^

2]

Work to do: determine L2, L – errors of KE, w, ..., dependent from x, t, ... for the tests cases

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 13

Task 6: deep valleys(Förstner, Torrisi, Reinhardt, deMorsier)detection of the reason for the unrealistic ‚cold pools‘ in Alpine valleys

Task 7: Different filter options for orography(Förstner)

Status: the orography filtering is now sufficiently weak for DWD-LMK applications (max. slopes 30% allowed)

The reason for the cold pools was identified: metric terms of the pressure gradient

Dynamical Bottom boundary condition (DBBC) (A. Gassmann (2004), COSMO-Newsl.) and a slope-dependent orography-filtering cures the problem to a certain extent.

Proposal for future work:inspect the limitations of the terrain following coordinate for steeper slopes, e.g. • for application of aLMo 2 (MeteoCH) in Alpine region• for future LMK ~1 km horizontal resolution

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 14

starting point after 1 h after 1 h

modified version:pressure gradient on z-levels, if

|metric term| > |terrain follow. term|

cold pool – problem in narrow valleys

is essentially induced by pressure gradient term

T (°C)

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 15

Dynamic BottomBoundary...

... Condition...... for metric pressure gradient term in equation for u- and v-component.Gaßmann (COSMO Newsletter No. 4)

“(Positive) Pressure Bias Problem”blue: Old Bottom Boundary Cond.red: Dynamic Bottom Boundary Cond.(Figures by Torrisi, CNMCA Rom)

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 16

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 17

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 18

Improved vertical advection for the dynamic var. u, v, w, T (or T‘), p‘

motivation: resolved convection

• vertical advection has increased importance => use scheme of higher order (compare: horizontal adv. from 2. order to 5. order)

• => bigger w (~20 m/s) => Courant-crit. is violated =>implicit scheme or CNI-explicit scheme

up to now: implicit (Crank-Nicholson) advection 2. order (centered differences)

new: implicit (Crank-N.) advektion 3. order LES with 5-banddiagonal-matrix

but: implicit adv. 3. order in every RK-substep; needs ~ 30% of total computational time!

planned: use outside of RK-scheme (splitting-error?, stability with fast waves?)

Task 8: Higher order discretization in the vertical for RK-scheme(Baldauf)

Status: implicit scheme of 3. order implemented (5-banddiagonal solver, ...)

Work to do: best combination with time integration scheme?test suite; verification

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 19

Idealized 1D advection test

analytic sol.implicit 2. orderimplicit 3. orderimplicit 4. order

C=1.580 timesteps

C=2.548 timesteps

Task 8: Improved vertical advektion for dynamic var. u, v, w, T, p‘

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 20

case study ‚25.06.2005, 00 UTC‘

total precipitation sum after 18 hwith vertical advection 2. order

difference total precpitation sum after 18 h‚vertical advection 3. order – 2. order‘

Task 8: Improved vertical advektion for dynamic var. u, v, w, T, p‘

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 21

Task 9: Physics coupling scheme(Förstner, Stephan, Raschendorfer)original task: problems with reduced precipitation could be due to a nonadequate coupling between physics scheme and dynamics

Status: NPDC-scheme analogous to WRF was implemented.Problems occuring reduced variant is used now

Work to do:what are the reasons for the failure of the WRF-PD-scheme in LM? (turbulence scheme?)

test tool (Bryan-Fritsch-case) is developed in PP ‚QPF‘, task 4.1

Problems in new physics-dynamics coupling (NPDC):

• Negative feedback between NPDC and operational moist turbulence parameterization (not present in dry turbulence parameterization)

• 2-z - structures in the specific cloud water field (qc)

• 2-z - structures in the TKE field, unrealistic high values, where qc > 0

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 22

Physics (I)• Radiation• Shallow Convection• Coriolis force• Turbulence

DynamicsRunge-Kutta [ (phys) + (adv) fast waves ]

‚Physics (I)‘-Tendencies: n(phys I)

Physics (II)• Cloud Microphysics

Physics-Dynamics-CouplingPhysics-Dynamics-Couplingn = (u, v, w, pp, T, ...)n

n+1 = (u, v, w, pp, T, ...)n+1

* = (u, v, w, pp, T, ...)*

‚Physics (II)‘-Tendencies: n(phys II)

+ n-1(phys II)

- n-1(phys II)

Descr. of Advanced Research WRF Ver. 2 (2005)

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 23

Task 10: Testing of alternative fast wave scheme(Gassmann, Förstner, Baldauf)• p‘T‘-RK-scheme• ‚shortened-RK2‘-scheme (Gassmann)• this allows the use of the ‚radiative upper boundary condition‘ (RUBC)

Status:• p‘T‘-RK-scheme is already tested and is used now in LMK• ‚shortened RK2‘-scheme works• RUBC is tested in idealized and one real test case

Work to do:• implement ‚shortened RK2‘ version in official LM version• experiments especially with RUBC• tests both versions in CLM-application (dx=18 km)

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 24

contours: vertical velocity isolines: potential temperature

Runge-Kuttaold p*-T-dynamics

Runge-Kuttanew p*-T*-dynamics

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 25

Choose CN-parameters for buoyancy in p‘T‘-dynamics from stability analysis=0.5 (‚pure‘ Crank-Nic.) =0.6 =0.7

=0.8 =0.9 =1.0 (purely implicit)

choose =0.7 as the best value

19.09.2006

Aktionsprogramm 2003

M. Baldauf, DWD 26

• grid length: x = 2.8 km

• direct simulation of the coarser parts of deep convection

• interactions with fine scale topography

• timestep t=30 sec.

• 421 x 461 x 50 grid points~ 1200 * 1300 * 22 km³lowest layer in 10 m above ground

• forecast duration: 18 hstarted at 0, 3, 6, 9, 12, 15, 18, 21 UTC

• center of the domain 10° E, 50° N

• boundary values from LME(x = 7 km)

LMK (Lokal-Modell Kürzestfrist)

in pre-operational mode at DWD since 14.08.2006