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

[netCDFJava #QAX-163810]: Data Value Discrepancies Between Netcdf 4.2 and 4.3



We do have a bug fix in version 4.4 about vertical levels sometimes getting 
flipped. Can you send me a file where you are seeing this problem (previous 
attachment didnt come through). Also, what version 4.3 are you using?

thanks,
John



> John,
> I was wondering if you had a chance to diagnose the RUC grib2 encoded file 
> sent last week for data discrepancies when read using netcdf-java version 4.2 
> and 4.3.
> 
> I was able to find and install the degrib command line utility provided 
> publicly by the NWS. This provided an independent grib parser for comparing 
> data values.
> 
> Using the degrib output as a reference lead to the following conclusions:
> 
> 1) If the provided RUC file is read using a 4.2 jar then the data values and 
> variable names match those of the degrib output
> 2) If the grib2 original RUC file is copied using the netcdf 4.3 Tools jar or 
> the netcdfAll-4.3.jar then when read using a 4.2 or 4.3 jar
> 2.a) Data values for the 3-D wind components 'flip' with respect to the 
> altitude index
> Ex: uWindComponentDegrib(alt=n, x=m, y=k) == 
> uWindComponentNetcdfCopied(alt=n, x=m, y=k) => false
> uWindComponentDegrib (alt=n, x=m, y=k) == uWindComponentNetcdfCopied 
> (alt=50-n+1, x=m, y=k) => true
> 2.b) Variable names change
> 
> If you could please provide some insight as to whether these conclusions are 
> plausible I would very much appreciate it. The degrib utility is independent 
> of netcdf and its results indicate that the 4.2 jar is providing the correct 
> data values, in contradiction to our discussion of 4.2/4.3 grib handling. I 
> can provide additional data upon request, thanks
> 
> Kevin
> 
> 
> -----Original Message-----
> From: Unidata netCDF Java Support [mailto:address@hidden]
> Sent: Friday, January 17, 2014 10:32 AM
> To: Ray, Kevin M
> Cc: address@hidden
> Subject: [netCDFJava #QAX-163810]: Data Value Discrepancies Between Netcdf 
> 4.2 and 4.3
> 
> Very likely!
> 
> > John,
> > I have found some projects that were utilizing a netcdf java 4.0 jar. Would 
> > that version be subject to the same behavior we are discussing? thanks,
> >
> > Sincerely
> > Kevin
> >
> > -----Original Message-----
> > From: Ray, Kevin M
> > Sent: Thursday, January 16, 2014 12:43 PM
> > To: 'address@hidden'
> > Subject: RE: [netCDFJava #QAX-163810]: Data Value Discrepancies Between 
> > Netcdf 4.2 and 4.3
> >
> > John,
> > Here is a RUC file and associated gbx file for which data values were found 
> > to differ. Please let me know if you require anything else,
> >
> > Kevin
> >
> > -----Original Message-----
> > From: Unidata netCDF Java Support [mailto:address@hidden]
> > Sent: Thursday, January 16, 2014 12:41 PM
> > To: Ray, Kevin M
> > Cc: address@hidden
> > Subject: [netCDFJava #QAX-163810]: Data Value Discrepancies Between Netcdf 
> > 4.2 and 4.3
> >
> > > John,
> > > Thank you for getting back.
> > >
> > > I will take your advice and start the process of updating our netcdf-java 
> > > dependencies to version 4.3.
> > >
> > > In the meantime would you mind examining a specific file to hopefully 
> > > diagnose the nature of the discrepancies? I have been copying the file 
> > > using custom java code (adhering to the previously referenced tutorials) 
> > > and the netcdf tools UI jar (versions 4.2 and 4.3). My verification 
> > > process has been bringing the file's data into MATLAB using the freely 
> > > available nctoolbox. The data values brought into MATLAB were compared to 
> > > the data values parsed by the custom java code by dumping data values 
> > > into a csv file, and found to match in all instances. What did not match 
> > > was the data values between original files and those that had been copied 
> > > using a 4.3 netcdf jar.
> > >
> > > I'm afraid that I do not have immediate access to an independent 
> > > verification process for the files under consideration, but I will look 
> > > around and try to see if there is one available. In the meantime what 
> > > would be the best way for me to get these files to you? Thanks!
> >
> > if too big for email, could use an ftp or web server or dropbox?
> >
> >
> > Ticket Details
> > ===================
> > Ticket ID: QAX-163810
> > Department: Support netCDF Java
> > Priority: Normal
> > Status: Open
> >
> >
> 
> 
> Ticket Details
> ===================
> Ticket ID: QAX-163810
> Department: Support netCDF Java
> Priority: Normal
> Status: Open
> 
> 


Ticket Details
===================
Ticket ID: QAX-163810
Department: Support netCDF Java
Priority: Normal
Status: Open