[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

19990129: GEMPAK Question

John,

If you can show me what you are doing in gddiag, that might help.
I'll step through gddiag to give you a feeling of whats going on.

You can calculate d/dz(variable) in gddiag. By default, GEMPAK
will create a name based on the functions, parameter, and levels.
To make the output grid more usable, you might want to control
the naming more.

Take for instance, calculating dT/dz in the 500 to 700 mb layer.

So the grids you need are:
LEVL1 LEVL2   VCORD PARM
500           PRES  TMPC
700           PRES  TMPC
500           PRES  HGHT
700           PRES  HGHT

Your gddiag function would look like:
glevel = 500:700
gfunc = quo(ldf(tmpc),ldf(hght))
gvcord = pres

you could also do the calculation like:
glevel = 500
gfunc = quo(sub(tmpc,tmpc@700),sub(hght,hght@700))


Now, if you let gddiag determine the name by itself, by leaving
GRDNAM = 
then you get a grid named:
LEVL1 LEVL2   VCORD PARM
 500  700     PRES QUOLDFTLDFH

That PARM name is pretty ugly. You can use GRDNAM = LAPSE
for instance, and get:
LEVL1 LEVL2   VCORD PARM
500  700     PRES LAPSE 

You notice that your input grids were defined on individual
constant pressure levels, 500 and 700 mb, but the output
grid is defined over a layer, using both LEVL1 and LEVL2.
Remember, them refering to a level in using the output function
later, your will have to use 500:700 as the layer.

If you decide you want the above calculation to be written as
the lapse rate at a single level between 500 and 700 mb, say
600mb, you can do that with GRDNAM = LAPSE@600 to get:
LEVL1 LEVL2   VCORD PARM
600          PRES LAPSE

GEMPAK uses up to 2 levels to define a parameter, and up to 2 times
to define a parameter. For accumulating/averaging a quantity over time,
you will get gdattimes like f000:f024 similar to the way
layers are denoted. The "^" character is used to modify
the gdattim variable, just as @ for glevel and % for gvcord
as described in the GPARM (3.36) section of the user guide.

How you chose to name the output grids is probably dependent on
what is most convenient for using in your later calculations.

The grid diagnostic excersises in the tutorial at:
http://www/packages/gempak/tutorial/diagnostic_exercises.html
Give some examples of some calculations that use the %,^ and @
modifiers to override the gvcord, gdattim and glevel parameters.
When you get ready to do your calculation that uses the individual
terms you calculate in gddiag, you may have to use the inline
modifiers since quite oftem you are using layer and level quantities
at the same time.

Steve Chiswell
Unidata User Support

>From: "John H. E. Clark" <address@hidden>
>Organization: .
>Keywords: 199901291333.GAA26601

>
>Steve,
>
>Some time ago I asked you about what to do if a line of GEMPAK code is too
>long. In particular I was trying to calculate a variable in GDPLOT and its
>definition in GFUNC was too long for a single line.
>
>You had suggested using GDDIAG to do the computation in steps.
>
>I have tried to use GDDIAG without success.
>
>I am working with a 24-hour data set of MM5 model output. At one particular
>time and at a few constant pressure levels, I wish to find the product of
>vertical velocity and the vertical derivative of a complicated function. As
>I understand GDDIAG, it can be used to calculate, say, the vertical
>derivative of the function and add it to my MM5 data set. I could then call
>the variable and multiply by vertical velocity in GDPLOT to get the desired
>result.
>
>The manual only offers a brief explanation of GDDIAG and every time I run
>it there are errors. Also even if I get GDDIAG to work such that it adds
>the new variable to my MM5 data set, I do not see how to call the new
>variable in GDPLOT to get the final result.
>
>I would appreciate any help you can offer.
>
>John Clark
>
>