Running a hydrological model with multiple timeseries files

In this notebook, we show how to provide the hydrological model with multiple timeseries files. For example, one file could contain meteorological data and the other contain streamflow data, or all variables could be separate (i.e. precip, temperature, streamflow, etc.). The following instructions should make it easier to understand how to do this. for this example, we actually start from a netcdf file containing all information, and from there divide it into multiple time series netcdf files. We then use the split files to drive the model. In most user cases, different files will be provided directly by the user so no need to pre-split your files!

SECTION TO SEPARATE DISCHARGE AND MET DATA TO RECOMBINE LATER¶

[5]:

# Define the 2 new files, i.e. the meteorological data and the streamflow data
filepathMet = os.getcwd()+"/CANOPEX_Met.nc"
filepathQobs=os.getcwd()+"/CANOPEX_Qobs.nc"

# Do the extraction for the selected catchment
newBasin=ds.isel(watershed=watershedID)

# Generate the streamflow time-series netcdf
Qobsfile = newBasin['discharge']
Qobsfile.to_netcdf(filepathQobs)

# Generate the meteorological time-series netcdf
newBasin=newBasin[['drainage_area','pr','tasmax','tasmin']]
newBasin.to_netcdf(filepathMet)

Here is where we run the model with multiple input time series¶

[6]:

# The model parameters. We are forcing values here just so the model runs, the models are probably very bad choices!

params = '9.5019, 0.2774, 6.3942, 0.6884, 1.2875, 5.4134, 2.3641, 0.0973, 0.0464, 0.1998, 0.0222, -1.0919, ' \
         '2.6851, 0.3740, 1.0000, 0.4739, 0.0114, 0.0243, 0.0069, 310.7211, 916.1947'

# Model configuration parameters. Please see "canopex.ipynb" for more details.
# This remains unchanged with multiple timeseries!
config = dict(
    start_date=start,
    end_date=stop,
    area=basin_area,
    elevation=basin_elevation,
    latitude=basin_latitude,
    longitude=basin_longitude,
    run_name='test_hmets_NRCAN',
    rain_snow_fraction='RAINSNOW_DINGMAN',
    nc_spec=json.dumps({'tasmax': {'linear_transform': (1.0, -273.15)},'tasmin': {'linear_transform': (1.0, -273.15)},'pr': {'linear_transform': (86400.0, 0.0)}},),
)


# Here is where we must tell the model that there are multiple input files. The idea is to combine them into a list of strings,
# with each string representing a path to a netcdf file. So we could do something like this:
ts_combined=[str(filepathMet),str(filepathQobs)]
resp = wps.raven_hmets(ts=ts_combined, params=params, **config)

# And get the response
# With `asobj` set to False, only the reference to the output is returned in the response.
# Setting `asobj` to True will retrieve the actual files and copy the locally.
[hydrograph, storage, solution, diagnostics, rv] = resp.get(asobj=True)
print(diagnostics)

observed data series,filename,DIAG_NASH_SUTCLIFFE,DIAG_RMSE,
HYDROGRAPH,/tmp/pywps_process_fufyza3l/CANOPEX_Qobs.nc,-0.12679,30.5267,

You can even invert the order of the netcdf files. Raven will detect which files contain which variables, so the order is not important!¶

[7]:

# Test with reversed timeseries files:
ts_combined=[str(filepathQobs),str(filepathMet)]
resp = wps.raven_hmets(ts=ts_combined, params=params, **config)

# And get the response
# With `asobj` set to False, only the reference to the output is returned in the response.
# Setting `asobj` to True will retrieve the actual files and copy the locally.
[hydrograph, storage, solution, diagnostics, rv] = resp.get(asobj=True)
print(diagnostics)

observed data series,filename,DIAG_NASH_SUTCLIFFE,DIAG_RMSE,
HYDROGRAPH,/tmp/pywps_process_6r8q6yk0/CANOPEX_Qobs.nc,-0.12679,30.5267,

As you can see, the NSE values and RMSE values are identical. You can pass as many NetCDF files as you have variables in any order and it will still work.¶

Running a hydrological model with multiple timeseries files¶

SECTION TO SEPARATE DISCHARGE AND MET DATA TO RECOMBINE LATER¶

Here is where we run the model with multiple input time series¶

You can even invert the order of the netcdf files. Raven will detect which files contain which variables, so the order is not important!¶

As you can see, the NSE values and RMSE values are identical. You can pass as many NetCDF files as you have variables in any order and it will still work.¶

Raven

Navigation

Related Topics