HIRLAM towards mesoscale and physics towards AROME

1
HIRLAM towards meso-scale and physics towards
AROME

• Per Undén
• HIRLAM-6

2
HIRLAM Vision for 2010

• (I) Best available 2.5 km meso-scale modelling
  system operational in most of the HIRLAM
  countries for parts or all of the territory
  and applicable for 1 km
• Non-hydrostatic (non approximated equations)
• Efficient dynamics (long time steps possible)
• Advanced physics for clouds and precipitation
  species and turbulence (convection mainly
  resolved)
• Advanced data assimilation that initialises
  particularly the moist processes (rain and
  clouds)
• Integrated system for estimating the
  probabilities
• Transparent boundary treatment

3

• (II) Synoptic scale 10 km model for coupling from
  ECMWF and for regional forecasting
• Of equal quality to global model and tightly
  coupled to ECMWF
• Consistent physics with meso-scale model
• To provide best forecasts of synoptic
  disturbances with short data cut off
• To provide coupling to meso-scale and other
  models and many applications

4
Why meso-scale model?

• Is stated in HAC strategy 2002 and in HIRLAM-6
  MoU and Management Group needed to take
  intitiative
• Is absolutely needed in mountainous countries
• Is needed for severe precipitation weather not
  resolved otherwise
• Needed for applications (wind energy, radar usage
  , environmental dispersion, aviation
• Because everybody else does it

5
How?

• Develop own meso-scale model? No
• Tartu NH outside Project work and not full Euler
  equations and not really used in member
  institutes
• Not enough dynamics staff
• Large effort years of development
• Duplication of work in Europe
• New developments in assimilation and physics also
• Still resources needed for synoptic model as well
• Collaborate - Yes
• Quickly re-gain lost time of development
• Sharing research resources with partner(s)
• Actively contributing in many key areas
• Relying on partner(s) for other things
• Enough critical mass in all areas

6
Why ALADIN?

• HIRLAM and ALADIN are closely related
  scientifically
• Spectral LAM and analysis with extension zones
• Both born out of ECMWF and ECMWF similarities
• ECMWF code commonality in ALADIN
• ALADIN non-hydrostatic model breakthrough
• Advanced meso-scale physics through Meso-NH -gt
  AROME
• Advance Data Assimilation developing
• Meteo-France participates in HIRLAM since 1992
• Good experience of MF collaboration in HIRLAM
• Informal offers in 2003 to use ALADIN NH dynamics
• Likely to be a true collaboration from both sides
  
• Many partners of different size
• Organisation more dispersed than HIRLAM

7
Choices made (MG Sept 2003, HAC Jan 2004, Council
June 2004, Dec 2004)

• Collaborate with MF/ALADIN at code collaboration
  level on meso-scale modelling gt
• Achieves all the advantages of ALADIN and AROME
  dynamics, physics and Data Assimilation
  developments AND ECMWF IFS System and Code
  standards
• Commitments to synchronise the code at regular
  intervals
• Coordination of research in the agreed Area and
  sharing of the work

8
Further necessary steps

• Agreement on collaboration in meso-scale
  modelling and associated areas
• Definition of cross steering mechanisms
• Formulation of rights and obligations and MoUs
  2006-
• Coordination of common research areas for 2006 -

9
Choices not made (yet?)

• Code collaboration in the rest (synoptic scale
  model)
• Or HIRLAM may continue with own separate code for
  synoptic scale
• Scientific and technical extra work to maintain
  and develop two code systems (the synoptic and
  meso-scale model share large parts of the codes
  and scripts)
• Allows for clear HIRLAM brand for synoptic
  scale model (10 km)
• Some inconsistencies in coupling of boundaries to
  meso-scale model (physics and also grid)
• HIRLAM physics in ALADIN synoptic model
• HIRLAM brand can still be maintained

10
HIRLAM status

• Hydrostatic 2 TL SL-SI LAM
• Non-hydrostatic p-anelastic kernel (Tartu)
• Reference at 22 km 40 levels
• 50 km 30 22 17 11 5 km (2.5)(1.4)
• Comprehensive physics for synoptic scales
• (meso-b 40-10 km)
• SW/LW simple radiation
• CBR TKE 1D turbulence advected TKE
• ISBA surface tiled scheme with assimilation/snow/s
  oil
• STRACO conv/cond (Kuo-Sundqvist type)
• Kain-Fritsch/Rash-Kristjansson alternative
• Meso-scale orography/Sub grid scale implementing

11
Model Physics in High Resolution NWP
Physics No Mans Land
Klemp (2002) Dublin WS
12
6 h WRF Forecast Reflectivity at 6/15/02 18z
4 km Lin
10 km Lin, KF
Radar Composite
10 km Lin
(Weisman, Wang, Klemp)
(dBZ)
13
HIRLAM components for Meso-scale

• Surface scheme ISBA, snow, soil models
• Still mainly horizontally uncoupled upper air
  flow couples but flux aggregation ?
• Tiled or untiled ? tiles still exist below 1 km
• Turbulence scheme CBR TKE - moist
• 1D or 3D ? 1D at 3 km and 3D at 1 km?
• Interactions with cloud scheme and convection
• Shallow convection ?
• Radiation scheme slopes considered
• 3D ? - More advanced more species
• Cloud scheme more advanced and more species
• MSO/SSO
• MSO relaxed but SSO needed at 1 km

14
AROME and ALADIN components

• Turbulence CBR HIRLAM 1d
• Externalised ISBA HIRLAM tiled
• (Kain-Fritsch synoptic scale - HIRLAM)
• Town model
• Advanced cloud physics
• Radiation scheme (Morcrette, ECMWF)
• Chemical modelling

15
HIRLAM strategy and plans

• Maintain and improve HIRLAM physics for synoptic
  scales
• Interface HIRLAM physics in ALADIN and compare in
  meso-scale and with AROME
• Integrate some as AROME option or mainly AROME in
  meso-scale
• Compare synoptic ALADIN with HIRLAM physics with
  HIRLAM
• Plan for transfer to ALADIN with HIRLAM switches

16
HIRLAM work with ALADIN in 2005

• Make experiments at 11 and 2.5 km
• Interface HI -ALADIN coupling-boundaries
• Interface some HI physics
• Implement climate generation software
• Introduce AROME system
• Experience from AROME from events -evaluate some
  HI physics
• Develop meso-scale diagnostics

17
Synoptic HIRLAM work 2005

• 3D-VAR improvements
• 4D-VAR to be made operational (almost)
• More satellite use
• Surface analysis SST , ice, SAF products
• Turbulence and moist version
• Surface scheme, fluxes, snow
• Meso-scale / sub-grid scale orography
• KF convection, statistical cloud scheme
• SL dynamics
• Transparent boundary conditions
• Verification and diagnostics
• CVS code maintenance and unification -HIRLAM

18
(No Transcript)
19
(No Transcript)
20
6.3.5 test
21
Challenges for meso-scale forecasting

• Meso-scale data and assimilation
• Modelling and predictability
• Probability / ensemble forecasting
• Presentation to users
• Verification at high resolution and of extreme
  events
• Computational resources

22
Summary

• Synoptic system state of the art
• Synoptic system needs to be maintained
• Meso-scale modelling through collaboration
• Synoptic scale should be collaborated on
• A number of HIRLAM physics modules valid in
  meso-scale or combine the best
• Many challenges in meso-scale not all in
  physics

23
Physics interfaces and options

• All building blocks of physics routines should be
  available without restrictions
• ARPEGE/ALADIN
• Meso-NH
• HIRLAM
• ALARO
• Start with low-level routines to adapt for
  interface and high level changes to use them
• Iterate and work and clarify the design

24
Reference system

• 6.2.2 Feb 2004 -gt Reference 6.3 RCR
• Climate files and corrections
• 6.2.3 22 Mar climate corrections
• 6.2.4 29 March
• Radiation changes
• Increased mixing in stable conditions
• Much increased roughness length
• parallelisation much improved
• 6.2.5, 6.3.2 minor corrections HIRVDA
• 6.3.3 28 June water vapour saturation mod
• 6.3.4 21 Sep many physics corr cleanup FAK
• Mod (reduced) z0 and drag and smoother mixing
• Ritchie-Tanguay T eq. SL extrap. techn.
  Changes, job subm.
• 6.3.5 Nov
• Rotated surface stress and reduced stable mixing
• Phys-Dyn coupling
• Melting of soil ice faster many technical
  climate files

25
Envisaged HIRLAM configurations
26

• High performance computer resources
• 4 x higher resolution gt 4x4x4, but..
• Smaller area(s) ,
• More efficient dynamics
• - more expensive physics
• Linux clusters very promising to reduce costs
• Moores law 2x / 1.5 years wait .
• But EPS .. More forecasts- lagged average?

27
(No Transcript)