Welcome, Guest
Username: Password: Remember me
General topics

TOPIC: CAPE in AROME

CAPE in AROME 4 years 1 month ago #1551

  • Lisa Bengtsson
  • Lisa Bengtsson's Avatar
  • OFFLINE
  • Administrator
  • Posts: 60
  • Thank you received: 5
Dear all,

Reports of excessively high CAPE values from Harmonie-AROME compared with for instance values given from the ECMWF model have been reported from forecasters at SMHI and Met-Eeireann (also seen at FMI).

Since we are currently in the season of convection we propose here a few alternatives that can be used in operations during this summer, and we will continue to evaluate the most appropriate method(s) to use and have a more long term solution implemented in fullpos on top of cycle 43.

Alternative CAPE computations:

1) Current method uses a combination of surface based CAPE and “most unstable CAPE”. E.g within a defined layer depth above surface, the parcel (level) resulting in the largest CAPE values are used. It is activated by selecting NFPCAPE => '5' in the script Select_postp.pl. (Computed by fullpos). The result is shown in the attached plot labelled “Reference AROME - most unstable CAPE.”

2) An alternative method to be computed in fullpos has been presented by ZAMG (Austria) – however, this code is not yet phased to cycle 43 (and thus not part of any of our “h” cycles either). This method is based on “mixed layer CAPE”, (or mean layer CAPE) – no entrainment assumed in the ascent. E.g the parcel entering CAPE computations (starting from the surface) is characterized by the temperature/humidity which is valid for a well-mixed layer (typically 50-100hPa) depth above the surface. The result of this method is shown in the attached plot labelled “Austria mixed layer CAPE” - source code for this method based on harmonie cycle 40h1.1.beta5 is attached.

Two options have also been added to the “gl” tool, and post-processing of CAPE can be done after the forecast.

3) First Trygve Aspelien at MET Norway have added a routine computing surface based CAPE (no entrainment assumed). The result of this method is shown in the attached plot labelled “Surface based CAPE no entrainment”.

4) In the final alternative I have added a routine to gl where we compute CAPE using the exact same method as done at ECMWF. This computation is based on a combination of mixed layer CAPE and most unstable CAPE – no entrainment is assumed during the ascent. The result of this method is shown in the attached plot labelled “Mixed layer CAPE ECMWF method”. I can confirm that this method used in gl on ECMWF data reproduces the CAPE field given as output from ECMWF (seen in the bottom middle and bottom right figures). I have also added computations of CIN based on the same method.

To post-process CAPE using one of the “gl” methods please find the modified code attached (gl.tar). Code is based on cycle 40h1beta5, and to use surface based CAPE set cape_version = 0 in your namelist and to use ECMWF based CAPE set cape_version = 1 in your namelist.

For CAPE:

&NAMINTERP
PPPKEY(1)%ggg = 253,
PPPKEY(1)%ppp = 160,
PPPKEY(1)%lll = 000,
PPPKEY(1)%ttt = 105,
PPPKEY(1)%tri = 000,
cape_version = 1,
LWRITE_PPONLY = T,
/

For CIN:

&NAMINTERP
PPPKEY(1)%ggg = 253,
PPPKEY(1)%ppp = 165,
PPPKEY(1)%lll = 000,
PPPKEY(1)%ttt = 105,
PPPKEY(1)%tri = 000,
LWRITE_PPONLY = T,
/


At SMHI the last option (4) will be used for both CAPE and CIN. If you use any of these methods during the summer it would be great if you could provide some feedback on its usefulness.

CAPE.png
Last Edit: 4 years 1 month ago by Lisa Bengtsson.

CAPE in AROME 4 years 1 month ago #1552

  • Lisa Bengtsson
  • Lisa Bengtsson's Avatar
  • OFFLINE
  • Administrator
  • Posts: 60
  • Thank you received: 5
Ok, maybe I cannot attach .tar files? Maybe just e-mail me if you are interested in the code from Austria, I will update gl in trunk with the ECMWF method. Trygves method already exists in gl since cycle 38.
The following user(s) said Thank You: Carl Fortelius

CAPE in AROME 3 years 11 months ago #1638

Hi,

maybe this is stating the obvious, but mixed layer CAPE values are typically much smaller during daytime conditions than surface CAPE or most unstable CAPE (when the most unstable parcel typically equals the surface parcel), so in this case it would be expected that "ECMWF CAPE" (which utilizes a mixed-layer parcel) would be lower than "Harmonie-CAPE" (not utilizing a mixed-layer parcel).

Your figures do show very large differences between the Harmonie CAPE and ECMWF CAPE, I would be curious to know how exactly the latter is formulated :)

Another point, mixed-layer parcels implicitly assume entrainment. Whether its the best way to account for it is another matter; in my own work where I compared thunderstorm predictors using ERA-Interim, mixed-layer CAPE fared worse than most unstable cape for predicting +0-6h lightning occurrence, even for daytime pseudo-soundings.

I'm currently doing some 12 km ALARO Harmonie-Climate simulations for some thesis work, might also do nested 2.5 km AROME runs if we get it to work, and eventually I hope to do some evaluation of convective parameters in harmonie.
Last Edit: 3 years 11 months ago by Peter Ukkonen .

CAPE in AROME 3 years 11 months ago #1639

  • Lisa Bengtsson
  • Lisa Bengtsson's Avatar
  • OFFLINE
  • Administrator
  • Posts: 60
  • Thank you received: 5
Hello,

See my answer below each question below:

maybe this is stating the obvious, but mixed layer CAPE values are typically much smaller during daytime conditions then surface CAPE or most unstable CAPE (when the most unstable parcel typically equals the surface parcel), so in this case it would be expected that "ECMWF CAPE" (which utilizes a mixed-layer parcel) would be lower than "Harmonie-CAPE" (not utilizing a mixed-layer parcel).

LB: Yes.

Your figures do show very large differences between the Harmonie CAPE and ECMWF CAPE, I would be curious to know how exactly the latter is formulated :)

LB: You can find some information on page 87 in the IFS documentation here:
www.ecmwf.int/sites/default/files/elibra...ysical-processes.pdf

and the source code I implemented in gl here:
hirlam.org/trac/browser/trunk/harmonie/util/gl/grb/cape_ec.f90


Another point, mixed-layer parcels implicitly assume entrainment. Whether its the best way to account for it is another matter; in my own work where I compared thunderstorm predictors using ERA-Interim, mixed-layer CAPE fared worse than most unstable cape for predicting +0-6h lightning occurrence, even for daytime pseudo-soundings.

I'm currently doing some 12 km ALARO Harmonie-Climate simulations for some thesis work, might also do nested 2.5 km AROME runs if we get it to work, and eventually I hope to do some evaluation of convective parameters in harmonie.

LB: It will be very interesting to hear what you find. Forecasters here at SMHI currently look at both the most unstable CAPE from fulpos and the ECMWF method CAPE from gl and they have promised to report back on their usefulness after the summer.

One thing that I think is useful with the latter is that you can get the corresponding CIN values using the same method. Currently the CIN values in Fulpos are corrupt... another thing to look into.

/Lisa
Time to create page: 0.086 seconds