[ncl-talk] HDF5 reading issues

[David Brown]

Hi Alex,
Work has been ongoing on the HDF file I/O code since 6.3.0 was
released. I can report that the read_var_test.ncl now works as
expected using the development version, but there do still seem to be
issues with reading the string variable in the file. That is a bit of
a surprise since we have spent some time improving the string
handling. Apparently there is still work to be done. So thanks for
providing this file. It will be a good test case.

Here is the output of your read_var_test.ncl with the dev version of NCL:

Variable: nlats
Type: integer
Total Size: 4 bytes
            1 values
Number of Dimensions: 1
Dimensions and sizes: [1]
Coordinates:

(0) 56476
(0) The number of non-zero lats: 56476
(0) The number of non-zero lons: 56476

Variable: backscatter_dims
Type: integer
Total Size: 8 bytes
            2 values
Number of Dimensions: 1
Dimensions and sizes: [2]
Coordinates:
(0) 56476
(1) 533
(0) 30101708

string_test.ncl produces a different set of error messages now from
what showed up in 6.3.0, but there are still issues. Regarding a
string being a scalar in HDF5, that is actually to be expected. (A
single string is considered to be a scalar in NCL as well.) Here is
what h5dump says about this variable:

h5dump -d /metadata_parameters/Granule_Start_DateTime
CATS-ISS_L1B_D-M7.2-V2-07.2015-08-26T16-27-44T17-14-51UTC.hdf5
HDF5 "CATS-ISS_L1B_D-M7.2-V2-07.2015-08-26T16-27-44T17-14-51UTC.hdf5" {
DATASET "/metadata_parameters/Granule_Start_DateTime" {
   DATATYPE  H5T_STRING {
      STRSIZE 27;
      STRPAD H5T_STR_NULLTERM;
      CSET H5T_CSET_ASCII;
      CTYPE H5T_C_S1;
   }
   DATASPACE  SCALAR
   DATA {
   (0): "2015-08-26T16-27-44.858999"
   }
 -dave



On Wed, Jun 29, 2016 at 8:16 AM, Alexander Schaefer <aschae11 at uncc.edu> wrote:
> Hi All-
>
> I am working on reading some LIDAR data and running into some issues with
> how NCL is handling some of the commands.
>
> The first is with regards to strings.  I try to read a string variable and
> it spits out an HDF error.  I know the variable is a single string that is
> supposed to be read in; this was checked with hdfview.  Google and the HDF
> forums were of no help.
>
> HDF5-DIAG: Error detected in HDF5 (1.8.14) thread 0:
>
>   #000: H5Shyper.c line 6595 in H5Sselect_hyperslab(): hyperslab doesn't
> support H5S_SCALAR space
>
>     major: Invalid arguments to routine
>
>     minor: Inappropriate type
>
> HDF5-DIAG: Error detected in HDF5 (1.8.14) thread 0:
>
>   #000: H5Shyper.c line 6595 in H5Sselect_hyperslab(): hyperslab doesn't
> support H5S_SCALAR space
>
>     major: Invalid arguments to routine
>
>     minor: Inappropriate type
>
>
> Why is NCL trying to read a string as a scalar?  Any ideas on whether this
> is a library issue or an NCL coding issue?
>
> My second question is with respect to how NCL is reading in float variables.
> I am reading in a variable that I wouldn’t expect NCL to struggle with but
> it is.  The variable being read should be about 57K x 533 point.  It reads
> in with no errors returned but checking the dimensions tells me that NCL has
> made it a ~57k x ~57k variable.  I thought maybe NCL was assuming a lat x
> lon grid; checking the data in the returned array gives me all 0s but
> printing the lats and lons I know there is data. Any ideas on what is going
> on?  Does the square array have to do with the hyperslab routines?
>
>
> I am running NCL 6.3.0 on OS X 10.11.5.
>
> I uploaded to the ftp the file for the tests:
> CATS-ISS_L1B_D-M7.2-V2-07.2015-08-26T16-27-44T17-14-51UTC.hdf5
>
> Attached are the tests that produce the string error and the variable issue.
>
> Thanks for any insight,
> -Alex
>
>
>
>
>
>
>
>
>
> _______________________________________________
> ncl-talk mailing list
> ncl-talk at ucar.edu
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>