[ncl-talk] Incorrect reading of dimensions from he5

[Rebecca Buchholz]

Thanks Wei - it looks like updating has fixed both of my problems. The nPrs
now is loaded with correct dimensions and the averaging kernel matrix has
correctly named dimensions (nTime x nPrs2 x nPrs2). The dimensions of nTime
are still different to the "2" I see in hdfview, but I think the way NCL
loads it is correct (i.e. it correctly reads as the number of entries in
the he5 file).
Thanks again for your help!
Cheers,
Rebecca

On Thu, Nov 20, 2014 at 12:05 PM, Wei Huang <huangwei at ucar.edu> wrote:

> ​Rebecca,
>
> In the past NCL team has done a lots of bug fixes and added quite some
> new features. Some of the bug fixes are directly related to he5.
> Please update your NCL to 6.2.1 to see if the problem persist.
>
> Thanks,
>
> Wei​
>
> ================================================
> 1850 Table Mesa Dr.
> Boulder, CO 80307
> Phone: 303-497-8924
>
> On Thu, Nov 20, 2014 at 11:28 AM, Rebecca Buchholz <buchholz at ucar.edu>
> wrote:
>
>> Dear NCL team,
>>
>> I'm having difficulty reading some MOPITT V6 profile data. I am finding
>> zeros where there should not be any. The issue seems to be coming from NCL
>> incorrectly reading the dimensions of the hdf 5 file (he5). It happens for
>> every file but here is an example:
>>
>> The file:
>> ftp://l5eil01.larc.nasa.gov/MOPITT/MOP02T.006/2010.09.07/MOP02T-20100907-L2V16.2.1.he5
>>
>> I have attached an example ncl file script (test3.ncl) that shows how I
>> read in the file.
>>
>> Differences:
>>
>> ---------------------------------------------------------------------------
>> *Using hdfview these are the correct values for the profile dimension:*
>> OBJECT = Dimension_4
>>                      DimensionName="nPrs"
>>                      Size =9
>> END_OBJECT=Dimension_4
>>
>>
>> *Dimensions from an ncl_filedump gives:*
>> nPrs_MOP02 = 10
>>
>> -----------------------------------------------------------------------------
>>
>> All the other dimensions are identical in size, except for nTime, which
>> in hdfview has a size of 2 and ncl_filedump gives a size of 208903 (the
>> number of time values).
>>
>> Also, the named dimensions of some variables are incorrect. Specifically
>> for "RetrievalAveragingKernelMatrix" and "RetrievalErrorCovarianceMatrix"
>> the named dimensions are not the same:
>> *hdfview:* (ntime, nPrs2, nPrs2)
>> *ncl_filedump:* (ntime_MOP02, nPrs2_MOP02, nPrs_MOP02)
>> The averaging kernel and co-variance matrices should be 10x10 i.e. nPrs2,
>> nPrs2.
>>
>> Does anyone have any idea why NCL is incorrectly reading in the
>> dimensions of nPrs and how I can fix it?
>>
>> Specs: I'm running NCL version 6.1.2
>> uname-a gives
>> Linux mopfl 2.6.32-431.29.2.el6.x86_64 #1 SMP Tue Sep 9 13:45:55 CDT 2014
>> x86_64 x86_64 x86_64 GNU/Linux
>>
>> Kind Regards,
>> Rebecca
>>
>> _______________________________________________
>> ncl-talk mailing list
>> List instructions, subscriber options, unsubscribe:
>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20141120/8f8afbe6/attachment.html