HIRLAM mesoscale report

1
HIRLAM mesoscalereport

• Oslo Meeting
• 12-13 December 2005

2
CONTENTS

• 1) HIRLAM meso-scale experiences at model
  gridsizes below 10 km
• ? What is HIRLAM physics ?
• ? Operational or semi-operational verification
  results in
• HIRLAM countries
• ? Case study examples
• ? Conclusions on HIRLAM model performance on
  meso ?-scale
• 2) HIRLAM experiences with running IFS/ALADIN
  including status of
• HIRLAM physics implementation.
• ? Strategy
• ? Examples
• ? Conclusions

3
What is HIRLAM-physics ?

• HIRLAM physics has a long tradition since late
  1980s
• Many schemes for the key physical processes have
  been tested over the years.
• The origin of the present schemes used for
  turbulence and surface parameterisation goes back
  to Météo-France.
• The HIRLAM turbulence scheme is a special version
  of the socalled CBR-scheme including TKE. This
  scheme has been developed for several years in
  HIRLAM context.

4
What is HIRLAM physics (2)

• The HIRLAM surface scheme is a special version of
  ISBA (multi-tyle).
• The HIRLAM project has followed an independent
  line with regard to radiation parameterisation
  and schemes for condensation, convection and
  cloud cover (STRACO, KF-RK )
• The HIRLAM physics perform quite well and are
  very stable together with the HIRLAM
  semi-Lagrangian scheme in the HIRLAM hydrostatic
  model at resolutions down to about 5 km.

5
HIRLAM runs in grey zone

• It is becoming increasingly popular to run
• HIRLAM in the grey zone ( 3 km 8 km) both
  for
• tests and on a daily basis !
• HIRLAM 5 km daily runs at INM
• HIRLAM 5 km operational runs at DMI
• HIRLAM 5 km daily runs at SMHI

6
INM 5km daily runs

• Preliminary evaluation
• ?subjective
• 1) near surface winds improved compared to
  coarser mesh results.
• 2) Low level clouds improved.
• ?Objective
• Most parameters when verified with
  OBS-verification show similar results, but
  marginal improvements can be seen in some upper
  air fields , e.g. T(850 hPa), RH(500 hPa)

7
T(850 hPa) Aug-Oct. for 0.15 deg and 0.05 deg.
models at INM
8
T(500 hPa) Aug-Oct 2005 for 0.15 deg and 0.05
deg. resolution models at INM
9
RH(500hPa ) Aug-Oct. 2005 for 0.15 deg. and 0.05
deg. models at INM
10
DMI operational experiences

• 0.05 run compared to 0.15 run (40 levels) in
  OBS-verification
• Almost always significant improvement of 10m
  winds.
• ? 2m temperatures of similar or marginally
  better quality
• ? Precipitation scores are of similar quality
  both summer and winter (significant improvements
  in last few years)
• NOTE that
• OBS-verification normally favours the low
  resolution
• output !

11
DMI operational model areas T15, S05, Q05 in 2005
12
Diurnal cycle of T2m and V10m (August 2005,
Danish synop stations)
13
DMI operational rms/bias for T2m and V10m in
August 2005 (right) and November 2005 (left) for
Danish synoptic stations
14
Diurnal cycle of 2m relative humidity in August
2005 (left) and October (right) verified for
Danish synoptic stations.
15
DMI monthly precipitation scores for November
2005. Results of station sums are presented (T15
left, S05 right). Score based on daily
contingency tables in lower right corner (66,6 ,
67.0 respectively)
16
Greenland pieterak-simulation (T15 compared to
S05) Note that S05 only produce increased wind
over part of the domain, e.g. at the Pieterak
location. The strong winds were also observed
(not shown)
17
SMHI preliminary experience with 5 km version and
60 vertical levels using Kain-Fritsch
Rasch-Kristjansson cloud scheme

• Similar or marginally improved scores for 2 metre
  temperature
• Similar or slightly improved score for 10 metre
  wind
• Degraded cloud cover and also dew point (too few
  clouds)
• Increasing precipitation (somewhat excessive)
• Slightly degraded MSLP

18
Experimental 5km run at SMHI(preliminary nesting)
19
SMHI synop verification June-Sept. 2005
20
Meso-scale sensitivity studies of precipitation
with HIRLAM at FMI

• Summer cases of precipitation over Finland
  have been studied at grid resolutions of 11 km,
  5.5 km and 2.8 km using HIRLAM hydrostatic model
  and HIRLAM-Tartu non-hydrostatic model.
• A publication from 2005
• S. Niemela and C. Fortelius
• Applicability of Large-Scale Convection
  and Condensation Parameterization to meso-? scale
  in HIRLAM A Case Study of a Convective Event
• Mon Wea. Rev. 133, 2422-2435, Aug. 2005

21
Main findings of Niemelä and Fortelius (2005)
using different versions of the STRACO scheme
including sensitivity test on the use of
resolution dependent parameterization of
convection.

• 1) The best results are obtained with
  non-hydrostatic dynamics, with a fully grid-size
  dependent convection scheme, (e.g. triggering
  mechanism etc. ) The results for a 5 km grid
  spacing produced the best results and a radar
  reflectivity distribution that resembles the
  observed one surprisingly well !
• 2) Experiments down to 2.8 km with no
  convection scheme (explicit condensation) cannot
  compete with experiments using full physics
  Precipitation release is too much delayed with a
  wrong model frequency distribution of
  precipitation intensity !
• 3) The scale dependent formulations of
  convection in the grey zone is benefitial for
  the quality of the precipitation prediction, but
  the presently used formulations might be improved
  to be more optimal !

22
11 km small-scale convective
showers on 25 May 2001 (Finland)
5.6 km
Precipitation intensity simulations (Niemelä and
Fortelius 2005)
2.8 km
23
NH
HYD
5.6 km Dynamically forced convection case 27-29
July 2004
OBS
24
HIRLAM physics in IFS/ALADIN Implementation
strategy

• Adapt to existing coding practice in IFS/ALADIN
  whenever possible.
• Adapt to existing code structures when
  interfacing HIRLAM physics.
• Adapt to the reference system of equations of the
  new AROME forecasting system.
• An initial implementation of HIRLAM physics
  should involve only a limited part of the physics
  in case that alternative schemes are available,
  in order to limit the efforts of first
  validations

25
HIRALD double nested system setupsGrid sizes
10km (outer), 2.5 km (inner)
26
Experiences with default ALADIN physics (Cycle
29t2, no cloud condensate variables)

• ALADIN non-hydrostatic model with ALADIN physics
  runs stably on a daily basis for grid sizes at
  2.5 km and 40 levels in the vertical ( stable for
  60 levels ?)
• Precipitation forecasts are sometimes noisy
  with a tendency to produce too much precipitation
  in some precipitation bands.
• Precipitation has a problem to go across
  mountains, e.g. the Norwegian mountains.

27
Precipitation forecast (2.5 km) withstandard
ALADINphysic on 25 October 2005 from 6 UTC-18
UTC.
28
HIRLAM operational model accumulated
precipitation at a grid size of 5.5 km on 25
October 2005 from 6 UTC -18 UTC.
29
(No Transcript)
30
Status of HIRLAM-physics in IFS/ALADIN

• 1) The processes of radiation, turbulence
  and convection/condensation have all been coded,
  and have been tested to some extent. Technically,
  the convection/condensation processes have been
  most difficult because of a more complicated
  computation of pseudo-fluxes. This involves some
  overhead because the flux terms were not
  previously computed. The tendency computations of
  the physics are not eliminated since the code
  should still work for the synoptic scale HIRLAM
  model system.
• 2) Some numerical stability problems were
  encountered during the test period after initial
  coding. Some of these problems have been caused
  by coding errors, but other problems might be
  related to the combination of different physics
  packages (ALADIN turbulence HIRLAM condensation
  and convection). Similar type of problems have
  been encountered at Meteo-France when testing the
  Lopez condensation scheme.
• 3) Daily numerically stable runs have
  started in late november 2005 using a HIRLAM
  scheme for condensation, convection and clouds.
  The remaining physics will be included early in
  2006.

31
6h-18h acc. precipitation from standard ALADIN
phys. run at DMI , resolution 10 km, valid at 6.
Dec 2005
32
6h-18h acc. precipitation from HIRLAM phys
implementation in ALADIN , valid at 6. Dec 2005
33
Experimental test setup (10 km hydrostatic)
,6h-12h accumulated precipitation on 1 July 2003.
(ALADIN phys)
34
Experimental test setup result with HIRLAM
condensation/convection (STRACO), 6h-12h
accumulated precipitation, 1 July 2003.
35
Conclusions (1)

• Experience with daily runs of HIRLAM in grey
  zone shows improved results for 10m winds and
  neutral or marginally improved results for 2m
  temperature compared to coarser mesh results.
• The results for cloud and precipitation are
  mixed, depending on the choice for the physics
  used.
• The results of scale depending convection is
  encouraging and indicate that planned new ALARO
  physics developments have good chances to produce
  improved results for the grey zone

36
Conclusions (2)

• Implementation and test of HIRLAM physics in
  IFS/ALADIN is in good progress.
• Daily runs with HIRLAM condensation/convection
  scheme will be extended to include the HIRLAM TKE
  scheme and the radiation scheme.
• Phasing of HIRLAM physics into IFS will be done
  at a suitable time !