Kain-Fritsch/Rasch-Kristjanson in Hirlam

1
Kain-Fritsch/Rasch-Kristjanson in Hirlam

• Javier Calvo

With contributions from
Karl-Ivar Ivarsson Colin Jones
2
Outline of the talk

• Moist physics schemes in HIRLAM
• Reference is STRACO (Based on Sundqvist)
• New/optional is Kain-Fritsch Rasch-Kristjansson
• Background of new moist physics components
• Review of the results of the new schemes in
  Hirlam
• Conclusions

3
Moist physics schemes for HIRLAM (1)

• Operational scheme is called STRACO (Sass, 2002)
• Sundquist type (Kuo convection and Sundquist
  microphysics)
• Smooth transitions convective-large scale regimes
• Treatment of shallow convection
• Tuning for high horizontal resolutions (5-10 km)
• Improvement of diagnostic cloud scheme
  (statistical type)
• Most HIRLAM components developed with this scheme
  as reference.
• Long operational use.

4
Moist physics schemes for HIRLAM (2)

• A scheme based on Kain-Fritsch convection and
  Rasch/Kristjansson is in the HIRLAM reference
  system since September 2004.
• Three new components
• Kain-Fritsch convection
• Rasch-Kristjansson condensation
• Diagnostic cloud fraction based on RH

5
Background of the new schemes
Kain-Fritsch Convection

• Mostly developed and used within MM5 community
• Specially suitable for mesoscale middle-latitudes
  simulations (10-30 km resolution) including
  severe phenomena.
• Operational use over the USA (up to 10 km
  resolution)
• MM5 Model
• Eta Model
• WRF Model
• Bechtolds version of KF
• MESO-NH Extensively tested. Many cases studies
  from synoptic scale to CRM scales. Realistic
  systems
• ARPEGE/ALADIN. Not able to improve the reference
  convection
• Hirlam
• RCA, Rossby Center Regional Climate Model 20-50
  km
• Operational at SMHI since 2003
• Hirlam reference as option since September, 2004

6
Hirlam modifications to the new schemes
Kain-Fritsch Convection

• Based on new version of KF (Kain, JAM, 2004)
• Implementation similar to the one in WRF model
• In HIRLAM, implemented by Colin Jones for RCA
  model
• Several updates mainly concerning shallow
  convection

7
Background of the new schemes
Rasch-Kristjansson large scale condensation

• Based on Rasch and Kristjansson, J. Climate, 1998
• Developed for the Community Climate Model (CAM3)
  where is the reference since 2004
• First introduced in Hirlam by Odegard, 1999 (HNL
  33)

8
Hirlam modifications to the new schemes
Rasch-Kristjansson large scale condensation

• The scheme is divided in two parts
• Condensation parameterization (water vapor ?
  condensate) following Sundquist ideas but
  completely recoded.
• Bulk microphysics (condensate to precipitation)
  with and approach similar to CRM formulae.
• Different physical processes are well separated
• Updates and improvements more easy
• For HIRLAM mainly tuning updates (Mainly by C.
  Jones).
• Inclusion of Hirlam thermodynamics
• Make the autoconversion process independent of
  models vertical resolution

9
Hirlam modifications to the new schemes
Diagnostic cloud fraction

• Large scale clouds basically depend on Relative
  Humidity. Initially based on Slingo (1987) but
  completely recoded.
• Convective clouds are function of convective
  mass-flux following Su and Krueguer (1991).
• Shallow clouds are based on Albrecht (1981) as
  described by Jones and Sánchez (2002)
• Depend on resolved RH and Cu updrafts water vapor
  and condensate.
• There is a memory of shallow clouds that may
  generate precipitation through large scale
  microphysics.

10
Review of the results of the new moist physics
(KFRK) in Hirlam

• Single Column Model case studies
• Models climate Hadley circulation
• Rossby Center Climate model.
• Complete model case studies Mainly severe
  weather
• Parallel runs comparison with reference STRACO
  moist scheme and observations.
• Operational use at SMHI.

11
Review of the results of the new moist physics
(KFRK) in Hirlam

• Single Column Model case studies
• EUROCS Project
• Models climate Hadley circulation
• Rossby Center Climate model.
• Complete model case studies Mainly severe
  weather
• Parallel runs comparison with reference STRACO
  moist scheme and observations.
• Operational use at SMHI.

12
EUROCS project

• EUROCS tried to improve the representation of
  clouds in GCM and NWP models.
• Identify systematic errors in the models.
• Design ideal cases based on observations.
• Compare SCM and CRM with observations
• Evaluate improvements in the GCM.
• Philosophy similar to GCSS intercomparisons
  except that EUROCS also includes simulations with
  the complete GCMs.

13
Diurnal Cycle of Sc over ocean
Evolution of Liquid water

• KF/RK and STRACO show similar behaviour
• Lack of or insuficient cloud top entrainment
• Drizzle acts to control Liquid Water
• Sensitivity to microphysics formulation.

14
Diurnal Cycle of Shallow Cu over land
Evolution of Liquid water

• The mass-flux approach (KF) describes better the
  growth and daily evolution of the clouds.

• Passive cloud formulation for KFRK
• As active clouds tipically not larger than 5 it
  is very important to include the passive clouds
  which have a significant impact on radiation

15
Review of the results of the new moist physics
(KFRK) in Hirlam

• Single Column Model case studies
• EUROCS Project
• Models climate Hadley circulation
• Rossby Center Climate model.
• Complete model case studies Mainly severe
  weather
• Parallel runs comparison with reference STRACO
  moist scheme and observations.
• Operational use at SMHI.

16
Representation of the Hadley Circulation
Climate of KF/RK and STRACO

• Hirlam run in NWP mode
• Compute monthly means of H24 forecasts
• Compare with satellite observations
• Comparison with other models

17
Cross sections along the Hadley Circulation
18
Precipitation Comparison with TRMM satellite
KF/RK
STR
TRMM
19
Precipitation Comparison with TRMM satellite
Both Hirlam schemes show too much precipitation
in the Sc and Shallow Cu regions
20
Cloud Cover Comparison with ERA ECMWF H6
KF/RK
STR
ERA ECMWF H6
21
Deficiencies on cloud cover and LWP specially
underestimation in the Sc regions and
overestimationi n the sahllow Cu
KF/RK
Cloud cover
STR
LWP
22
Review of the results of the new moist physics
(KFRK) in Hirlam

• Single Column Model case studies
• EUROCS Project
• Models climate Hadley circulation
• Rossby Center Climate model.
• Complete model case studies Mainly severe
  weather
• Parallel runs comparison with reference STRACO
  moist scheme and observations.
• Operational use at SMHI.

23
Hirlam-Rossby Center Climate Model(Willen, Jones
and Wyser, 2004)
RCA3 44km 24L ERA40

• Seasonal precipitation from RCA compared to
  observations
• the model captures the amplitude and spatial
  distribution of the precipitation for all seasons

24
Review of the results of the new moist physics
(KFRK) in Hirlam

• Single Column Model case studies
• EUROCS Project
• Models climate Hadley circulation
• Rossby Center Climate model.
• Case studies Mainly severe weather
• Parallel runs comparison with reference STRACO
  moist scheme and observations.
• Operational use at SMHI.

25
Cases Studies

• Niemela and Fortelius, 2002 (HNL 41)
• Frontal system. Experiment at 5.5 km resolutions
  and 40L
• KFRK shows stronger updradrafts and downdrafts
  than STRACO
• No clear adventage on using one of the schemes.

26
A typhoon case study (COMPARE III)

• H72 integration using different initial
  conditions.
• Comparison with observations and other models

• Errors in the track prediction are small
• Little impact of the resolution on the track
  forecast

27
Typhoon case study Intensity
Very rapid intensification (100 hPa/ 3 days) -
Very sensitive to initial conditions and model
resolution - Kain-Fritsch performed quite well
for this case (BOLAM and HIRLAM model)
STR
KF/RK
28
Typhoon case study STRACO precipitation (H60)
20 km
50 km
29
Typhoon case study KF/RK precipitation (H60)
20 km
50 km
30
Precipitation as function of model resolution in
the core region
As expected, in the inner area (200 km around
the typhoon center), convective precipitation
decreases as resolution increases whereas large
scale precipitation increases. Total
precipitation increases increasing model
resolution.
31
Review of the results of the new moist physics
(KFRK) in Hirlam

• Single Column case studies.
• Models climate Hadley circulation
• Case studies Mainly severe weather
• Parallel runs comparison with reference STRACO
  moist scheme.
• Operational use at SMHI.

32
5 test periods on different seasons (3 months).
Using Hirlam 6.2.0
NWP mode at 20km Parallel runs
- Comparison of reference STRACO and KFRK-
Scores against observations

• 23 Dec- 7 Jan 2000
• 115 May 2001
• 1-15 Aug 2001
• 15-30 Sep 1994
• 1- 31 Oct 2002

33
Comparisons against EWGLAM soundings
Similar scores excepted in the humidity field
34
Comparisons against SYNOP stations
MSL Pressure
Cloud Cover
KF/RK
STR

• Only differences in MSLP and cloud cover
• Small deterioration of MSLP
• Some improvemnet in cloud cover

35
MSLP RMS against its own analysis (all periods)
KF/RK
REF
36
MSLP RMS against its analysis (all periods)
REF
KF/RK
37
Precipitation against SYNOPs (all periods)
38
October 2002 precipitation as seen by
SYNOP stations and High-resolution
climatic stations
39
Upscale high-resolution observations to model
resolution
40
October 2002 precipitation impact of the
resolution
41
October 2002 precipitationKFRK vs REF-STRACO at
0.2
KF/RK
REF
42
Review of the results of the new moist physics
(KFRK) in Hirlam

• Single Column case studies.
• Models climate Hadley circulation
• Case studies Mainly severe weather
• Parallel runs comparison with reference STRACO
  moist scheme.
• Operational experience at SMHI.

43
Operational experience at SMHI(from
Karl-Ivarsson, personal comunication)

• KF/RK is in the SMHI operational suit since Dec,
  2002
• At 0.4 and 0.2 resolutions with 40 levels
• Experimental 0.1 resolution.
• Only a few experiments comparing KF/RK with the
  reference STRACO so no conclusion on this.
• Duty forecasters seem to be very satisfied with
  the new scheme, specially for convective
  situations.
• Validation of the new operational suite gave a
  significant improvement of the humidity forecasts
  (but several new modules incorporated)
• It seems that Hirlam is able to add value to
  ECMWF forecasts in terms of clouds and
  precipitation.

44
Operational experience at SMHI(Karl-Ivarsson,
personal comunication)

• Comparison of Hirlam 0.2, Hirlam 0.1 and ECMWF
  model for the period July, 7-Ago,17, 2004 True
  Skill Statistic

45
(No Transcript)
46
Experiment at 10 km ppt acc. 6hr H06
47
Experiment at 10 km ppt acc. 6hr H12
48
Experiment at 10 km ppt acc. 6hr H18
49
Experiment at 10 km ppt acc. 6hr H24
50
(No Transcript)
51
High-resolution observations
52
Conclusions

• The new moist physics scheme based on
  Kain-Fritsch convection and Rasch-Kristjansson
  large scale condensation has been tested in
  Hirlam.
• In terms of standard scores no big difference
  with reference STRACO scheme a part from
• Slight deterioration in MSLP probably caused by a
  tendency to over-intensification of cyclonic
  systems.
• Slight improvement in cloud cover
• Improvement in the humidity profiles
• Similar scores of precipitation
• STRACO better for small ppt amounts (lt 1 mm/day)
• KF/RK probably better for ppt 3-30 mm/day
• More clear improvement in southern regions
  probably linked to the larger number of
  convective situations.
• From case studies it seems that KFRK tends to
  produce more realistic cloud and precipitation
  patterns.

53
Conclusions

• Operational experience at SMHI very satisfactory
• KFRK takes about 20 more computer time than the
  reference STRACO.
• Still pending the vectorization of the code
  external asistance is planned.
• Some clearly identified problems
• Excess of drizzle.
• Cloud cover formulation
• Statistical cloud scheme
• Moist turbulence
• We have started testing KFRK at 10 km
  resolutions.
• In the near future we should evaluate its
  performance at lower resolutions.
• Probably further tuning is needed

54
The End