[RegCNET] max and min variables in postproc

Rauscher Sara_A_ srausche at ictp.it
Thu Jul 3 00:39:30 CEST 2008


Hi,

It seems that the current version of PostProc is not correctly calculating
the daily min and max variables averages (like tamax, tgmax, psmin, etc.)
from the SRF output. Right now if you do a daily average in postproc (by
selecting "continuous averaging" and the daily option), you get just that,
a daily average. However, on daily timescales generally you want to have
the daily maximum or minimum value of max/min variables, not the time
average. So if you have 3 hourly output, the program should find the max
or min value in those 8 time slices (8x3= 24 hours = 1 day) and this
should be value for the "daily average." If you compare the post-processed
max/min data with observations right now with results created with this
version of postproc, you might note that TMAX is too low, and TMIN might
be too high.

Anyway - this seems to have happened when we added additional output
variables like psmin and tgmax and tgmin to the SRF output some time ago.
If you have used the post-processor and have used max/min values, you
should check them against your original output. You can find the max or
min easily in grads:

define tm = max(tmax,t=1,t=8) *over a day by going from 1 to 8

and compare that with output from the post-processor.

We will do our best to fix this problem as soon as possible. You can,
however, easily calculate these fields using grads, ferret, NCL, or any
other software, or even fortran. Bi has his own set of postprocessors that
work and do not have this bug, but they are not as user-friendly as
postproc.

IT IS IMPORTANT TO NOTE THAT THIS GLITCH AFFECTS ONLY THE MAX/MIN
VARIABLES LIKE MAX TEMP, MIN TEMP, MIN SFC PRESSURE, ETC. VARIABLES LIKE
SURFACE AIR TEMPERATURE, SFC ENERGY FLUXES, AND PRECIP ARE -NOT- AFFECTED.

If you have questions or need assistance with postproc, please don't
hesitate to contact RegCNET or me or Bi to learn more about his
postprocessor.

cheers,
sara




More information about the RegCNET mailing list