20050331: Icing GRIB/netCDF file displays

>From: "Petty Kevin" <address@hidden>
>Organization: NTSB
>Keywords: 200503311749.j2VHnpv2003613 IDV grib reading

Hi Kevin-

>Let me see if I can do a good job of explaining myself. Greg Salottolo =
>forwarded me your response to the error he found associated IDV's =
>ability to display Total Icing Potential GRIB data. I downloaded the new =
>nc2.2.jar file, and I have been examining Total Icing Potential GRIB and =
>netCDF files. I think there are a couple of other issues. I am attaching =
>files that I have been using as a test case. The GRIB file is the =
>original file, and netCDF file was produced by converting the GRIB data =
>utilizing "gribtonc." Because of file
>size limitations, I will attach the GRIB file to this message and  send =
>the netCDF file separately.
>
>The icing potential product is based on a 40km grid and should have =
>values from 0 to 1. However, when IDV loads the GRIB file, it defaults =
>to a range from 0 to 15 (on the color scale). I think this is the result =
>of values at 7010 meters. If you use the data probe at this level on =
>areas over northern California, you will see values at or near 15.  This =
>may be as simple as a decimal point placement problem (i.e. 15 should be =
>0.15). A display of the netCDF data shows values near 0.15 in this =
>region. I just don't know why it would only be happening at one level. I =
>have seen this problem in another GRIB file I loaded, but it was at a =
>different level. It was still near the top of the icing layer.

Okay, this was a bug.  I put a new nc2.2.jar at:

ftp://ftp.unidata.ucar.edu/pub/idv/untested/nc2.2.jar

Download it and replace the other one in your IDV directory.  Let me
know if this doesn't fix the problem.

>I have also noticed other small differences between the GRIB display and =
>the netCDF display. For example, if you look at the supercooled large =
>droplet potential (SLD) over southwest Michigan at 2134 meters (I used a =
>color-shaded plan view), you'll see that there are very small =
>differences in the display. Specifically, the SLD display from the =
>netCDF file covers a slightly larger area when compared to the display =
>from the GRIB file. Any thoughts why this would occur?

I'm not sure what's going on here, but I can recreate what you are seeing.
I'm thinking it's some slight difference between the way the values
are read from grib versus netcdf.  Perhaps it's a float/double precision
issue.  If you put a data probe over the values that appear to be
changing and use the Nearest Neighbor sampling, you'll see that the
values are exactly the same.  I'm kind of stumped on this but am
still investigating.

>Thanks for your help,

Thanks for your clear explanations.

Don Murray
NOTE: All email exchanges with Unidata User Support are recorded in the
Unidata inquiry tracking system and then made publicly available
through the web.  If you do not want to have your interactions made
available in this way, you must let us know in each email you send to us.