Re: Fill values differ- NCARG 5.1 to 5.2

Hi Minna,

NCL 5.2.0 and later decodes this GRIB data correctly. Between 5.1 and 5.2 there was a bug fix involving files that had both a bit map (that indicates which points are valid) and constant data (only one value for the whole grid). Prior to the fix, the bitmap was not being checked if the data was constant. Since the fix, it is. The bitmap for this data is set to almost all 0's indicating there is very little valid data. You should find that there are 684 "good" points that have valid data. Since the data is constant, all valid points are set to the single value 0.0. The other points are set to the default fill value for GRIB data: 1e20 -- note that this is an arbitrary fill value chosen by the NCL developers for GRIB data. The GRIB specification does not indicate what value should be used for fill values. It was changed within the last couple of years from the previous value -999 to avoid accidental equality of real data with the fill value.

 -dave

On Feb 9, 2011, at 10:05 AM, Minna Win wrote:

> Hello,
>
> I am currently running two versions of NCL: 5.1.0 and 5.2.0. I've
> observed that fill values for specified variables are different between
> these versions. For instance, when performing the following NCL commands:
> ----------------
> ncl 0> f=addfile("201102081942_NCWD2-L.grb","r")
> ncl 1> print(f)
> ncl 2> l=f->LAVNI_GDS3_EATM
> ncl 3> print(l)
>
> The NCL 5.1.0 output looks like:
> ---------------------------------------------------------------
> Variable: l
> Type: float
> Total Size: 6750000 bytes
> 1687500 values
> Number of Dimensions: 2
> Dimensions and sizes: [g3_x_0 | 1125] x [g3_y_1 | 1500]
> Coordinates:
> Number Of Attributes: 14
> sub_center : Aviation Weather Center
> center : US National Weather Service - NCEP (WMC)
> long_name : Low-level aviation interest
> units : non-dim
> _FillValue : 1e+20
> coordinates : g3_lat_0 g3_lon_1
> level_indicator : 200
> gds_grid_type : 3
> parameter_table_version : 129
> parameter_number : 177
> level : 0
> forecast_time : 0
> forecast_time_units : minutes
> initial_time : 02/08/2011 (19:42)
> (0,0) 0
> (0,1) 0
> (0,2) 0
> (0,3) 0
> (0,4) 0
> (0,5) 0
> (0,6) 0
> (0,7) 0
> (0,8) 0
> (0,9) 0
>
> ---------------------------------------------------------------
>
> NCL 5.2.0 output:
> ---------------------------------------------------------------
> Variable: l
> Type: float
> Total Size: 6750000 bytes
> 1687500 values
> Number of Dimensions: 2
> Dimensions and sizes: [g3_x_0 | 1125] x [g3_y_1 | 1500]
> Coordinates:
> Number Of Attributes: 14
> sub_center : Aviation Weather Center
> center : US National Weather Service - NCEP (WMC)
> long_name : Low-level aviation interest
> units : non-dim
> _FillValue : 1e+20
> coordinates : g3_lat_0 g3_lon_1
> level_indicator : 200
> gds_grid_type : 3
> parameter_table_version : 129
> parameter_number : 177
> level : 0
> forecast_time : 0
> forecast_time_units : minutes
> initial_time : 02/08/2011 (19:42)
> (0,0) 1e+20
> (0,1) 1e+20
> (0,2) 1e+20
> (0,3) 1e+20
> (0,4) 1e+20
> (0,5) 1e+20
> (0,6) 1e+20
> (0,7) 1e+20
> (0,8) 1e+20
> (0,9) 1e+20
> ---------------------------------------------------------------
>
>
> The different versions of NCL are running on separate machines with
> similar hardware and versions of Linux:
> 1) Linux strut 2.6.26-2-amd64 #1 SMP Thu Sep 16 15:56:38 UTC 2010 x86_64
> GNU/Linux (this host has NCL 5.1.0)
>
> 2) Linux trim 2.6.26-2-amd64 #1 SMP Thu Nov 25 04:30:55 UTC 2010 x86_64
> GNU/Linux (this host has NCL 5.2.0)
>
> The 1e+20 fill values observed with NCL 5.2.0 are causing some problems
> with our plotting scripts. I've attached the grib file that illustrates
> this behavior. Is there some work-around I can employ?
>
> Thanks in advance for any assistance.
>
> --
> Minna Win
> Software Engineer
> Aviation Digital Data Service (ADDS)
> National Center for Atmospheric Research (NCAR)
> Research Applications Laboratory (RAL)
> 3450 Mitchell Lane
> Boulder, CO 80301
> minnawin@ucar.edu
> +1(303)497-8423
>
> <201102081942_NCWD2-L.grb>_______________________________________________
> ncl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Thu Feb 10 13:24:21 2011