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

20020626: McIDAS ETA ABV vs calculated ABV question



>From:  "Jennie L. Moody" <address@hidden>
>Organization:  UVa
>Keywords:  200206270315.g5R3F6u03265 McIDAS GRDCOPY XCD GRID

Jennie,

>I am still trying to track down what might be an error in some of
>the PV fields that we calculated previously, based on a
>correction factor (might turn out to be a misnomer!) that Owen
>applied.  Anyway, in my process I am relearning mcidas (good),
>and confusing myself (bad).  First, I was trying to see if the 
>quantity ABV (absolute vorticity) in some of the ETA files was
>already there, or if it was placed there by Owen.

ABV is a parameter that is available from the ETA runs.  Here
is an example that shows this:

DATALOC ADD RTGRIDS ADDE.UCAR.EDU
GRDLIST RTGRIDS/ETA FORM=FILE ALL
<pick out the 0Z run for forecast hours 0 to 24; in dataset position 8>

GRDLIST RTGRIDS/ETA.8 PAR=ABV NUM=ALL
Dataset position 8      Directory Title= ALL  00Z ETA 0  HR<=VT<=24 HR
PAR    LEVEL       DAY        TIME    SRC FHOUR     FDAY       FTIME  GRID  PRO
---- --------- ------------ -------- ---- ----- ------------ -------- ----- ----
ABV  1000 MB   27 JUN 02178 00:00:00  ETA     0 27 JUN 02178 00:00:00   230 LAMB
ABV   850 MB   27 JUN 02178 00:00:00  ETA     0 27 JUN 02178 00:00:00   231 LAMB
ABV   700 MB   27 JUN 02178 00:00:00  ETA     0 27 JUN 02178 00:00:00   232 LAMB
ABV   500 MB   27 JUN 02178 00:00:00  ETA     0 27 JUN 02178 00:00:00   233 LAMB
ABV   250 MB   27 JUN 02178 00:00:00  ETA     0 27 JUN 02178 00:00:00   234 LAMB
ABV  1000 MB   27 JUN 02178 00:00:00  ETA     6 27 JUN 02178 06:00:00   387 LAMB
ABV   850 MB   27 JUN 02178 00:00:00  ETA     6 27 JUN 02178 06:00:00   388 LAMB
ABV   700 MB   27 JUN 02178 00:00:00  ETA     6 27 JUN 02178 06:00:00   389 LAMB
ABV   500 MB   27 JUN 02178 00:00:00  ETA     6 27 JUN 02178 06:00:00   390 LAMB
ABV   250 MB   27 JUN 02178 00:00:00  ETA     6 27 JUN 02178 06:00:00   391 LAMB
ABV  1000 MB   27 JUN 02178 00:00:00  ETA    12 27 JUN 02178 12:00:00   740 LAMB
ABV   850 MB   27 JUN 02178 00:00:00  ETA    12 27 JUN 02178 12:00:00   741 LAMB
ABV   700 MB   27 JUN 02178 00:00:00  ETA    12 27 JUN 02178 12:00:00   742 LAMB
ABV   500 MB   27 JUN 02178 00:00:00  ETA    12 27 JUN 02178 12:00:00   743 LAMB
ABV   250 MB   27 JUN 02178 00:00:00  ETA    12 27 JUN 02178 12:00:00   748 LAMB
ABV  1000 MB   27 JUN 02178 00:00:00  ETA    18 27 JUN 02178 18:00:00  1738 LAMB
ABV   500 MB   27 JUN 02178 00:00:00  ETA    18 27 JUN 02178 18:00:00  1739 LAMB
ABV   250 MB   27 JUN 02178 00:00:00  ETA    18 27 JUN 02178 18:00:00  1740 LAMB
ABV   700 MB   27 JUN 02178 00:00:00  ETA    18 27 JUN 02178 18:00:00  1744 LAMB
ABV   850 MB   27 JUN 02178 00:00:00  ETA    18 27 JUN 02178 18:00:00  1746 LAMB
ABV  1000 MB   27 JUN 02178 00:00:00  ETA    24 28 JUN 02179 00:00:00  1890 LAMB
ABV   850 MB   27 JUN 02178 00:00:00  ETA    24 28 JUN 02179 00:00:00  1894 LAMB
ABV   700 MB   27 JUN 02178 00:00:00  ETA    24 28 JUN 02179 00:00:00  1896 LAMB
ABV   500 MB   27 JUN 02178 00:00:00  ETA    24 28 JUN 02179 00:00:00  1898 LAMB
ABV   250 MB   27 JUN 02178 00:00:00  ETA    24 28 JUN 02179 00:00:00  1899 LAMB
Number of grids listed = 25
GRDLIST - done

To get some stats on one of these grids, run:

GRDINFO RTGRIDS/ETA.8 STAT GRID=230     <- for 1000 MB ABV
Statistics of

Parameter: ABV
Units: PS
Level of data: 1000 MB
Time of data:      0 UTC
Day of data:   2002178
Data came from: ETA
0-hour forecast

Dataset: RTGRIDS/ETA

Minimum:  -0.4100000E+01 occurred at [row,col]: [   62,   83]
                                   [lat,lon]: [  17.88,  72.03]
Maximum:   0.3564000E+02 occurred at [row,col]: [    1,   93]
                                   [lat,lon]: [  57.31,  49.37]
Mean:   0.8647677E+01
SD  :   0.3556637E+01
Number of points analyzed:   6045
Number of points missing:       0

Row range:     1 to    65
Col range:     1 to    93
GRDINFO Done, Number of grids statistically analyzed=1
GRDINFO - done

The question is what the unit PS represents; per second?

>I made a new
>grid with the derived quantity ABV, and compared it with the old
>grid.  The grids look the same (place contours in the same
>location), but they have different units.  This matters to me,
>we are doing MATH on grids, and we have to be sure that the
>units are working out!

Absolutely.  The first thing to check in any calculation is units.

>Anyway, this raises one question...the
>units on these two grids (one that I made with the DERIVE=ABV
>command, the other one that was in my ETA grid directory, but
>could have been made by Owen in some step I haven't seen) the
>units of these two grids are different by 6 orders of magnitude
>(one has a scale factor of 2, the other has a scale factor of
>8!)....but, they list units of PS and 1/S respectively.

So, I think that the discrepency may be cleared up by correct
interpretation of exactly what PS means.  The GRIB table used
by XCD decoders defines absolute vorticity as follows:

041 | Absolute vorticity                  | ABV  | 1/s            | PS    | 7

The units are 'per second' and the scale factor is 10**7.  This tells
me that the values listed above for ABV range from -4.15 * 10**-7
to 35.64 * 10**-7.

>Their
>is no listing for frequencies in the units table (cu.k ), but
>this looks like its trying to say these are both per second???
>Does this make sense to you Tom?

Yes, but the values in the XCD decoded grid have a scale factor applied
to them.  This seems to be what is causing the interpretation difference.

>I have appended the long
>form listing of these two grids, from two different grid files,
>so you might see whats up:
>
>windfall: /home/jlm8h/mcidas/data $ grdlist.k MYDATA/GRIDS.9001
>PAR=ABV FOR=ALL
>Dataset position 9001   Directory Title=
>PAR    LEVEL       DAY        TIME    SRC FHOUR     FDAY
>FTIME  GRID  PRO
>---- --------- ------------ -------- ---- ----- ------------
>-------- ----- ----
>ABV  1000 MB   08 MAR 00068 12:00:00  ETA     0 08 MAR 00068
>12:00:00   119 LAMB
>Total pts=  6045 Num rows=  65  Num columns=  93    received:
>2000097 160201Z
>Absolute vorticity
>GRIB ID numbers: Geographic =211; PAR = 41; Model ID = 89; Level
>type =100
>Units of gridded variable are PS   Scale of variable is:   2
>Lambert Conformal Tangent Cone Projection
>Row num of pole= -113.34  Col num of pole=   53.00  Col spacing
>(m)= 81270.0
>Standard Latitudes=   25.00     25.00   Standard Longitude=
>95.00
>Number of grids listed = 1
>GRDLIST - done

OK, this is ETA output equivalent to what I listed above.

>windfall: /home/jlm8h/mcidas/data $ grdlist.k MYDATA/GRIDS.9004
>PAR=ABV FOR=ALL
>Dataset position 9004   Directory Title=
>PAR    LEVEL       DAY        TIME    SRC FHOUR     FDAY
>FTIME  GRID  PRO
>---- --------- ------------ -------- ---- ----- ------------
>-------- ----- ----
>ABV   400 MB   08 MAR 00068 12:00:00  ETA     0 08 MAR 00068
>12:00:00     1 LAMB
>Total pts=  6045 Num rows=  65  Num columns=  93    received:
>2000097 160200Z
>Absolute vorticity
>GRIB ID numbers: Geographic =211; PAR = 33; Model ID = 89; Level
>type =100
>Units of gridded variable are 1/S  Scale of variable is:   8
>Lambert Conformal Tangent Cone Projection
>Row num of pole= -113.34  Col num of pole=   53.00  Col spacing
>(m)= 81270.0
>Standard Latitudes=   25.00     25.00   Standard Longitude=
>95.00
>Number of grids listed = 1
>GRDLIST - done

This must be the one you calculated since the PAR = 33 indicates that
the grid is the U component of the wind.  The ABV that comes with the
ETA run output specifies the parameter with PAR = 41.  For reference,
the GRIB tables used for the XCD grib decoder are gbtbpds001.2v[123].
You can find these in the ~mcidas/workdata directory (~mcidas/uvaworkdata
on windfall).

 -- deleted cu.k output --

>If you will bear with me a minute, this is only the first step
>that I was trying to follow.  One reason I decided to look more
>carefully at even a simple step like the ABV is that I had looked
>up the way ABV is defined in GRDCOPY and I found what I hope is
>just a typo in the definition table (surely this cannot really
>be an error in the mcidas code!).....ABV=VOR+COR, but you will
>see that this is not true in the table in the users guide (I am
>looking at Version 7.7, which is what I am running).  The
>derivatives are switched.  VOR is right, you get vorticity
>(circulation) by taking the variation of the u wind-component
>in the y direction, and the v wind-component in the x direction.
>The ABV in the table must be written incorrectly.  Anyway,
>this sidetracked me so I thought I would just look at ABV, 
>and then derive it myself by taking the derivatives of the
>U and V winds respectively and building ABV.  But, I didn't
>get this far and decided I won't waste too much time with this
>until I know how the units work.

OK.

>What is PS versus 1/S?

PS is the unit assigned to several parameters decoded by XCD.  It
apparently represents 'per second'.  The missing piece is the scaling
by 10**7.

>I am trying to say this clearly, but reading over what
>I wrote, is might not strike you that way.

Not to worry.

>Like I said
>at the beginning, this might work better if we had a
>conversation.  Let me know what you think, and if its
>better explained over the phone, maybe I can give you a
>call tomorrow sometime.

I think I understand your explanation.  Does the scaling of the XCD
grids rectify the situation for you?

>Just to state this one more time, this is just the first part
>of my effort to sort out what Owen did (and where there
>might be a mistake).

I understand.

>Ultimately, to get isentropic PV from
>data on constant pressure surfaces, you need to make a 
>correction which involves the vertical derivatives
>of wind and horizontal derivatives of Temperature.  I
>think he might have done this incorrectly, but I need
>to recalculate things, and I want to do it one step
>at a time.  So.....thanks for any light you can shine
>here at the beginning of this tunnel regarding making
>sense of SCALE and UNIT.

If the scaling of ABV in the ETA output doesn't jibe with what
you are seeing, we should definitely chat about this.

>I sure this must be more straight foward than it appears
>at the moment.  I am also pretty sure that there must just
>be a typo in the definition table of the derived parameters
>available to GRDCOPY.

A typo wouldn't suprise me at all.

>(I am sorry, because I know this is all ground I covered
>in one way or another previously....sigh)
>
>
>One last by the way....I know I failed to list the exact same
>grids above...that doesn't make any difference...but if you want
>them the same, here they are:

The grids being different didn't change the point you were making.
I believe that I understood what you are trying to reconcile.

Again, if what I wrote does not clear things up/make sense, let's talk
about the situation.

Tom
--
+-----------------------------------------------------------------------------+
* Tom Yoksas                                             UCAR Unidata Program *
* (303) 497-8642 (last resort)                                  P.O. Box 3000 *
* address@hidden                                   Boulder, CO 80307 *
* Unidata WWW Service                             http://www.unidata.ucar.edu/*
+-----------------------------------------------------------------------------+