fix hycom download

[veenstrajelmer]

test_download_hycom fails for newer python versions, with "RuntimeError: NetCDF: file not found"

Warning messages from test_download_hycom:

xarray opening opendap dataset like: https://tds.hycom.org/thredds/dodsC/GLBu0.08/expt_19.1/2010.html (1 urls/years)
oc_open: server error retrieving url: code=? message="Error {
    code = 500;
    message = "-1008324796";
}"xarray subsetting data (variable 'surf_el' and lon/lat extents)
available time range in dataset from 2010-01-01 00:00:00 to 2010-12-31 00:00:00
xarray subsetting data per <Day>: 2010-01-01
xarray writing netcdf file: hycom_surf_el_2010-01-01.nc
oc_open: server error retrieving url: code=? message="Error {
    code = 500;
    message = "-1008324796";
}

Reproducible code:

import dfm_tools as dfmt
import pandas as pd

longitude_min, longitude_max, latitude_min, latitude_max =    2,   3,  51, 52 #test domain
date_min = '2010-01-01'
date_max = '2010-01-02'
year = pd.Timestamp(date_min).year
dataset_url = [f'https://tds.hycom.org/thredds/dodsC/GLBu0.08/expt_19.1/{year}']
varkey = 'surf_el'
dfmt.download_OPeNDAP(dataset_url=dataset_url,
                      varkey=varkey,
                      longitude_min=longitude_min, longitude_max=longitude_max, latitude_min=latitude_min, latitude_max=latitude_max,
                      date_min=date_min, date_max=date_max,
                      dir_output=".", overwrite=True)

Interestingly enough, this issue does not occur when passing dataset_url as a string instead of a list of one string.

Some different versions (py39 vs py311):

• dask 2024.8.0 vs 2024.12.1
• erddapy 2.2.0 vs 2.2.3
• pydap 3.5 vs 3.5.3
• pystac 1.10.1 vs 1.11.0
• scipy 1.13.1 vs 1.15.0
• urllib3 1.26.20 vs 2.3.0
• xarray 2024.7.0 vs 2025.1.0
• zarr 2.18.2 vs 2.18.4

py39_success.txt
py311_fail.txt

[veenstrajelmer]

Was suddenly resolved with fresh environment created on 21-1-2025, changes:

• dask 2025.01
• scipy 1.15.1
• xarray 2025.1.1

Not sure what the cause was, but if it works it works