Cloud water schemes incorporated into ECPC G-RSM


Name
define in def/physics.h
Cloud water prognostic variables (#)
Stratiform clouds
References
Efficiency (SLINGO =100%)
SLINGO
CLD1
None (0)
Diagnosed from RH
Slingo (1987)
100%
ZC
CLD2
qc/qi (1)
Diagnosed from RH and qc/qi
Zhao and Carr (1997), Sundqvist et al. (1987), Randall (1995)
110%
IS
CLDT93
qc/qi (1)
Predicted
Tiedtke (1993), Iacobellis and Somerville (2000)
150%
HONG3
CLD3
qc/qi, qr/qs (2)
Diagnosed from RH and qc/qi
Dudhia (1989), Hong et al. (1998, 2004), Randall (1995)
240%
HONG5
CLD5
qc, qi, qr, qs (4)
Diagnosed from RH and qc/qi

340%
HONG6
CLD6
qc, qi, qr, qs, qg (5)
Diagnosed from RH and qc/qi

510%
Cloud water prognostic variables: qc: cloud (liquid) water, qc: cloud ice, qr: rain water, qs: snow water, qg: graupel
Efficiency was estimated using GSM T62.


How to change cloud (water) scheme


"def/physics.h" should be modified when a different cloud scheme is used.
Default cloud scheme is CLD1

Use CLD2 : define CLD1 -> undef CLD1 / undef CLD2 -> define CLD2
Use CLDT93 : define CLD1 -> undef CLD1 / undef CLDT93 -> define CLDT93
Use CLD3 : define CLD1 -> undef CLD1 / undef CLD3 -> define CLD3
Use CLD5 : define CLD1 -> undef CLD1 / undef CLD5 -> define CLD5
Use CLD6 : define CLD1 -> undef CLD1 / undef CLD6 -> define CLD6

"Warning!" Convection schemes available to use with cloud water schemes


RAS with RASC1 or RASC2 : Relaxed Arakawa-Schubert scheme with detrained cloud water
ZM2 : modified Zhang McFarlane scheme
KF2 : Kain Fritsch convection scheme
Other convection schemes (ex. SAS) are not available to use with cloud water schemes in ECPC G-RSM, since detrained cloud water is not treated in those convection schemes.

Options


ICE : use saturation specific humidity (qs) with respect to ice (T<0C)
  • RH in pgb files is calculated using qs with respect to water for all schemes as default. If you would like to get RH with respect to ice, please modify fcst/getrh.F to make "#ifdef ICE" available.
  • Many codes use "function fpvs" for qs with respect to ice, however, some codes use their own formula. (e.g. gscond.F and precpd.F (CLD2) )
ICECLOUD : use cloud water/ice to calculate cloud optical properties
INTERACTIVE_STRATUS : cloudiness calculated from cloud water and RH, based on Randall(1995)
VCI : Has to be defined with CLD3 through CLD6.
NEW_MPHYS :
T93PR2 :


ntrac and ncldg values



CLD1 ntrac=1 and nldg=0
CLD2 ntrac=1 and ncldg=1
CLDT93 ntrac=3 and ncldg=1
CLD3 ntrac=1 and ncldg=2
CLD5 ntrac=1 and ncldg=4
CLD6 ntrac=1 and ncldg=5

The first ntrac is specific humidity (true for all cloud water schemes).
The second ntrac is O3 but it is not currently predicted in the model.
The third ntrac is cloud amount (CLDT93 only).
ncldg correspond to "cloud water prognostic variables" in the table above.


Horizontal diffusion of cloud water variables


You might find that a smaller time step is necessary to run the model (especially RSM) stably when using cloud water schemes. In this case, you might want to try an option DIFUC which can be turned on in def/physics.h. DIFUC introduces a stronger horizontal diffusion for cloud water variables (ntrac=2 and above and all ncldg). You may play with the value of difuc (default 0.5).

YOU NEED TO CHOSE DIFUC whenever you use all CLD water prediction schemes (i.e., all except CLD1).


Appendix


  • Comparison of cloud water schemes using 10-year averages (ppt)