Request: evaluation_tools.nwis_client CLI about hydrotools

[aaraney]

With old wizards in mind, is the intended functionality to use a long data format as adhered to currently, or pivot output to a wide format? Just curious your thoughts on handling multiple stations.

[jarq6c]

With old wizards in mind, is the intended functionality to use a long data format as adhered to currently, or pivot output to a wide format? Just curious your thoughts on handling multiple stations.

Ah yes, the wide format (one station per column) would more closely match the USGS rdb tab-separated output. That could be an option later. However, I suspect an output format with a fixed number of columns might be easier to use. If we assume old wizards prefer to manually parse CSV output line-by-line in C or FORTRAN, I think they will have an easer time if the number of columns is fixed.

[aaraney]

I agree, I think a long format would be best to implement from both our perspective and any perspective sorcerers.

[jarq6c]

Possible CSV output format:

# NCOLS 7
# NROWS 5
# HEADR 1
#
# value_date: Datetime of measurement (UTC) (character string)
# variable: USGS variable name (character string)
# usgs_site_code: USGS Gage Site Code (character string)
# measurement_unit: Units of measurement(character string)
# value: Measurement value (float)
# qualifiers: Qualifier string (character string)
# series: Series number in case multiple time series are returned (integer)
#
# Generated using nwis_client v1.3.0
# https://github.com/NOAA-OWP/evaluation_tools
# Retrieved 2021-02-18T13:51:02
# 
value_date,variable_name,usgs_site_code,measurement_unit,value,qualifiers,series
2019-09-29T23:00:00,streamflow,02146470,ft3/s,0.72,['A'],0
2019-09-30T00:00:00,streamflow,02146470,ft3/s,0.79,['A'],0
2019-09-30T01:00:00,streamflow,02146470,ft3/s,0.79,['A'],0
2019-09-30T02:00:00,streamflow,02146470,ft3/s,0.72,['A'],0
2019-09-30T03:00:00,streamflow,02146470,ft3/s,0.79,['A'],0

[aaraney]

Good addition including metadata. I was not familiar with pd.read_csv(comment="#") option until now. I like this functionality. With this in mind, I think including a csv writer in the nwis_client is a good idea to support users who may want to distribute data in a canonical evaluation_tools csv format? Maybe my definition of format is a little loose here, but I hope you get what I mean.