*****************************************************************************
*****************************************************************************
***** THIS IS A HELP FILE FOR THE FORTRAN SUBROUTINE CONVECT *****
***** VERSION 4.1 *****
***** September 15, 1997 *****
*****************************************************************************
*****************************************************************************
-----------------------------------------------------------------------------
----- TO BE SURE THAT YOU ARE ON THE MAILING LIST FOR UPDATES, NEWS, _____
----- ETC., PLEASE SEND EMAIL TO emanuel@texmex.mit.edu STATING THAT _____
----- YOU INTEND TO USE THIS SOFTWARE _____
-----------------------------------------------------------------------------
CONVECT accepts the one-dimensional input arrays
T,Q,QS,P, and PH,
and the scalars TG, QG, and PG,
and returns the arrays FT and FQ as well as the scalars
PRECIP, WD, TPRIME AND QPRIME
THE FLOATING POINT SCALAR CBMF MUST BE INITIALIZED
TO ZERO AT THE BEGINNING OF A TIME INTEGRATION AND MUST BE "REMEMBERED"
BETWEEN CALLS TO CONVECT AT THE SAME HORIZONTAL GRID POINT. That is, the
calling program must return the same values of CBMF that it
received from CONVECT at the same horizontal position at the previous time
step.
The first element of each array represents the LOWEST model level
(i.e. near the surface). The input arrays are returned unaltered, UNLESS
the dry adiabatic adjustment part of the routine is performed, in which
case T and Q are adjusted to represent well mixed profiles in
superadiabatic regions, with a small increment at the lowest level to
represent the surface layer.
Here are the definitions of the input arrays. With the exception of PH,
they are dimensioned ND; PH is dimensioned ND+1:
T: Temperature in degrees Kelvin.
Q: Specific humidity (NOT mixing ratio) in gm/gm (NOT gm/Kg).
QS: Saturation specific humidity in gm/gm.
P: Pressure in millibars.
PH: Pressure (in millibars) at the half levels; i.e. halfway in between
the levels at which P,T,R and RS are defined. The first PH level
is BELOW (i.e. at a higher pressure than) the first P level.
The following are input scalars:
TG: The temperature of the underlying surface (K).
QG: The specific humidity characterizing the underlying surface (gm/gm).
PG: The surface pressure (mb).
One also needs to specify:
ND: The dimension of the arrays T,R,RS,P,FT and FQ. PH is dimensioned ND+1.
NL: The maximum number of model levels you wish to have checked for
moist convection. NL MUST be less than or equal to ND-1.
One should never expect moist convection to penetrate above NL.
DELT: The model's time step, in seconds.
------------------------------------------------------------------------------
On output one receives:
FT: The tendency of ACTUAL (not potential!) temperature in degrees/second.
FQ: The tendency of specific humidity (NOT mixing ratio) in (gm/gm)/second.
PRECIP: The rate of convective precipitation, in mm/day.
WD: A convective downdraft surface velocity scale, likewise for use in the
calling program's formulation of surface fluxes.
TPRIME: A convective downdraft surface temperature perturbation for use
in the surface flux formulation.
QPRIME: A convective downdraft surface specific humidity perturbation for use
in the surface flux formulation.
*****************************************************************************
* Note: Suppose the surface flux formulation has the form *
* C|V|(x* - x) *
* where C is the exchange coefficient, |V| is a surface wind speed scale, *
* x* is the surface value of temperature or specific humidity and x is the*
* boundary layer value of the temperature or specific humidity. When this *
* is averaged over an area that contains downdraft outflows the it can *
* be shown that the result of this averaging is given by *
* C[V2](x* - xm) - C[V2 - V1]x'. *
* Here V1 is defined as SQRT[u^2 + v^2 + w*^2], where u and v are *
* the x and y components of the mean flow velocity and w* is a dry *
* convective velocity scale; V2 is defined as SQRT[u^2+v^2+w*^2+wd^2], *
* where wd is the moist convective downdraft velocity scale; xm is the *
* mean value of x over the area in question and x' is the departure of *
* x in convective outflows from its mean value. In the output of CONVECT, *
* wd is given by WD, and x' is given by TPRIME or QPRIME, in the case *
* of temperature and specific humidity, respectively. It was found that *
* the contribution of these modifications to the surface fluxes is not *
* negligible in the TOGA/COARE single column tests. *
**********************************************************************************
IFLAG: A flag that denotes the following:
VALUE INTERPRETATION
----- --------------
0 No moist convection; atmosphere is not
unstable, or surface temperature is less
than 250 K or surface specific humidity
is non-positive.
1 Moist convection occurs.
2 No moist convection: lifted condensation
level is above the 200 mb level.
3 No moist convection: cloud base is higher
then the level NL-1.
4 Moist convection occurs, but a CFL condition
on the subsidence warming is violated. This
does not cause the scheme to terminate.
-------------------------------------------------------------------------------
GENERAL USAGE: CONVECT is designed for use in a time-marching model; it should
NOT be used for single time step evaluations. The routine takes
several time steps to "spin up" the convective mass fluxes
toward equilibrium with the forcing. In the event one desires the
convective tendencies associated with a single sounding, CONVECT
should be initialized with that sounding and run for enough time
steps that the tendencies have relaxed back to zero; the average
of the convective tendencies would then be the desired result.
This is not recommended, however, and may not yield results
that are independent of the time step and other numerical
parameters.
CONVECT should be called at least every 20 minutes of
integration. Longer intervals may result in noisy
distributions of FT and FQ.
-------------------------------------------------------------------------------
NOTES: 1. At the beginning of the subroutine there is a parameter NA which
---- should in general be set equal to ND+1 or greater.
2. All the adjustable parameters appear near the beginning of the
subroutine. The values provided have been optimized using the
December, 1992, forcing values deduced from the TOGA/COARE
sounding array.
3. The parameters ALPHA and DAMP control the smoothness of the mass
fluxes in time and may be altered to achieve a smoother integration.
ALPHA and DAMP control the rate of approach to quasi-equilibrium.
PPMIN is the minimum pressure below which the atmosphere is not
checked for unstable parcel origin. Convection can only
originate at the FIRST level above the surface at which the
air is unstable to upward displacement. CONVECT
can represent elevated convection, but not if the boundary
layer is also unstable in the same column.
4. Thermodynamic constants are defined at the beginning of the
routine and again at the beginning of the subroutine TLIFT.
These should be altered, if necessary, to be consistent with
those of the calling program. The value of the gravitational
acceleration is also defined here.
5. This version differs in several important respects from that
described in the report from the MIT Center for Global Change
Science.
6. Many modifications have been made to the scheme as described in
J. Atmos. Sci., 1991, pgs. 2313-2335. These will be described in
a forthcoming paper.
7. Please report any problems to Kerry Emanuel, Room 54-1620,
M.I.T., Cambridge, MA 02139
EMAIL address: emanuel@texmex.mit.edu (Internet)
telephone: (617) 253-2462
Fax: (617) 253-6208
Thanks.
-------------------------------------------------------------------------------