How to load multi-species concentration file from an unstructured MODFLOW-USG Transport simulation?

[RyanConway91]

Based on the MF USG transport documentation, concentrations are saved with the same format as heads with headers conc01, conc02,... concNN

Therefore, I think this should work:
hds = fp.utils.binaryfile.HeadUFile(r'..\pht_usg_gwf.con',text='CONC01', verbose = True)
but is does not:

EOFError                                  Traceback (most recent call last)
Cell In[17], line 1
----> 1 hds = fp.utils.binaryfile.HeadUFile(r'..\pht_usg_gwf.con',text='CONC01', verbose = True)

File ~\AppData\Local\miniforge3\envs\gwmodel_2025\Lib\site-packages\flopy\utils\binaryfile\__init__.py:732, in HeadUFile.__init__(self, filename, text, precision, verbose, **kwargs)
    730         raise Exception()
    731 self.header_dtype = BinaryHeader.set_dtype(bintype="Head", precision=precision)
--> 732 super().__init__(filename, precision, verbose, **kwargs)

File ~\AppData\Local\miniforge3\envs\gwmodel_2025\Lib\site-packages\flopy\utils\binaryfile\__init__.py:314, in BinaryLayerFile.__init__(self, filename, precision, verbose, **kwargs)
    313 def __init__(self, filename: Union[str, os.PathLike], precision, verbose, **kwargs):
--> 314     super().__init__(filename, precision, verbose, **kwargs)

File ~\AppData\Local\miniforge3\envs\gwmodel_2025\Lib\site-packages\flopy\utils\datafile.py:217, in LayerFile.__init__(self, filename, precision, verbose, **kwargs)
    214     raise Exception(f"LayerFile error: unrecognized kwargs: {args}")
    216 # read through the file and build the pointer index
--> 217 self._build_index()
    219 # now that we read the data and know nrow and ncol,
    220 # we can make a generic mg if needed
    221 if self.mg is None:

File ~\AppData\Local\miniforge3\envs\gwmodel_2025\Lib\site-packages\flopy\utils\binaryfile\__init__.py:342, in BinaryLayerFile._build_index(self)
    340 ipos = 0
    341 while ipos < self.totalbytes:
--> 342     header = self._get_header()
    343     self.recordarray.append(header)
    344     if self.text.upper() not in header["text"]:

File ~\AppData\Local\miniforge3\envs\gwmodel_2025\Lib\site-packages\flopy\utils\binaryfile\__init__.py:399, in BinaryLayerFile._get_header(self)
    394 def _get_header(self):
    395     """
    396     Read the file header
    397 
    398     """
--> 399     header = binaryread(self.file, self.header_dtype, (1,))
    400     return header[0]

File ~\AppData\Local\miniforge3\envs\gwmodel_2025\Lib\site-packages\flopy\utils\binaryfile\__init__.py:210, in binaryread(file, vartype, shape, charlen)
    208 result = np.fromfile(file, vartype, nval)
    209 if result.size < nval:
--> 210     raise EOFError
    211 if nval != 1:
    212     result = np.reshape(result, shape)

EOFError: 

Any one done this before?

[mwtoews]

Out of curiosity, does flopy.utils.binaryfile.UcnFile() work? If no, and the file is small, post it somewhere to take a deeper look.

[RyanConway91]

No it does not

[RyanConway91]

Found the answer here: #374
here is a generalization of that solution that dumps the whole mess to a 4d dataframe

'''
process concentrations
'''
print('processing concentrations...')

def get_all_concentrations(ucnobj, nlay, ncpl):
    """
    Returns a 4D array of concentration data from a MODFLOW USG Transport binary output file.
    
    Dimensions: (ncomp, ntimes, nlay, nnodes)
    
    Parameters
    ----------
    ucnobj : flopy.utils.binaryfile.HeadUFile
        A MF USG Transport binary file object containing concentration data.
    Returns
    -------
    conc_data : ndarray
        4D array of shape (ncomp, ntimes, nlay, nnodes)
    text_labels : list
        List of component text labels used for each component.
    times : list
        List of simulation times used for each timestep.
    """
    times = ucnobj.get_times()
    text_labels = sorted(set(ucnobj.recordarray['text']))
    ncomp = len(text_labels)
    ntimes = len(times)
    conc_data = np.zeros((ncomp, ntimes, nlay, ncpl), dtype=ucnobj.realtype)

    for icomp, text in enumerate(text_labels):
        for itime, totim in enumerate(times):
            indices = np.where(
                (ucnobj.recordarray["totim"] == totim) & 
                (ucnobj.recordarray["text"] == text)
            )[0]
            for idx in indices:
                ilay = ucnobj.recordarray["ilay"][idx] - 1
                ipos = ucnobj.iposarray[idx]
                ucnobj.file.seek(ipos, 0)
                conc_layer = bf.binaryread(ucnobj.file, ucnobj.realtype, shape=(ncpl,))
                conc_data[icomp, itime, ilay, :] = conc_layer

    return conc_data, text_labels, times

ucnobj = fp.utils.binaryfile.HeadUFile('..\pht_usg_gwf.con', text='conc')
conc_data, components, times = get_all_concentrations(ucnobj,nlay,ncpl)

[mwtoews]

Ah, setting text='conc' seems to be the fix.