start = datetime.datetime(1980, 1, 1)
end = datetime.datetime(1990, 1, 2)
# make a time series
delta = datetime.timedelta(days=1)
times = [start + i * delta for i in range((end - start).days)]
# path to 7zip if using windows
processor.path_to_7z = r'C:/Program Files/7-Zip/7z.exe'
# download the data; this step will be skipped if this has already been done
# (from the last example) but is necessary to set the metadata
processor.download(bbox, start, end, output, datasets=['GSOD'])
# let's use the processor to aggregate the GSOD data together, including
# tmin, tmax, dewpoint, and wind speed. The GSOD database contains dew point
# and seems to be more complete than GHCND. It doesn't have snow or pan
# evaporation though--this is why they are both included.
tmax = processor.aggregate('GSOD', 'tmax', start, end)
tmin = processor.aggregate('GSOD', 'tmin', start, end)
dewpoint = processor.aggregate('GSOD', 'dewpoint', start, end)
wind = processor.aggregate('GSOD', 'wind', start, end)
# now these time series can be saved for later consistent with the structure
# used by PyHSPF's HSPFModel class (start date, time step in minutes, data)
# this way the data are easy to access later
bbox = -77.2056, 38.2666, -76.4008, 39.3539
# start and end dates (just look at a few months to highlight the concept)
start = datetime.datetime(1980, 4, 1)
end = datetime.datetime(1980, 10, 2)
# path to 7zip if using windows
processor.path_to_7z = r'C:/Program Files/7-Zip/7z.exe'
# download the data; this step will be skipped if this has already been done
# (from the last example) but is necessary to set the metadata
processor.download(bbox, start, end, output, datasets = ['GHCND'])
# let's use GHCND to work with temperature data. let's use the metadata to
# find the 10 stations with the longest temperature records and then aggregate
# the data together to get a mean time series across the period of interest.
n = 10
# make a list of the
templengths = []
for k, v in processor.metadata.ghcndstations.items():
templengths.append((v['tmax'], k))
# sort the list; the last 10 values are the longest records
bbox = -77.2056, 38.2666, -76.4008, 39.3539
# start and end dates (aggregate the whole 31 years)
start = datetime.datetime(1980, 1, 1)
end = datetime.datetime(2011, 1, 1)
# path to 7zip if using windows
processor.path_to_7z = r'C:/Program Files/7-Zip/7z.exe'
# download the data; this step will be skipped if this has already been done
# (from the last example); alternatively it will set the metadata
processor.download(bbox, start, end, output, datasets = ['precip3240'])
# aggregate the data -- it's important to keep in mind missing data at many
# of these stations and the high degree of spatial variability associated with
# precipitation. in this example, all the data are aggregating into one series;
# however, it may make sense to aggregate more specifically to capture this
# variability to some degree (lots of papers on this subject).
precip = processor.aggregate('precip3240', 'precip', start, end)
# now these time series can be saved for later consistent with the structure
# used by PyHSPF's HSPFModel class (start date, time step in minutes, data)
# this way the data are easy to access later
name = '{}/precip3240_aggregated_precip'.format(output)
ts = start, 60, precip
# you may need to change the path to 7zip for your system; can also be done
# when the processor is instantiated (should not be necessary on linux)
processor.path_to_7z = r'C:/Program Files/7-Zip/7z.exe'
# download the data--worth noting there are some issues with the databases
# sometimes. these are handled with warnings, but this will give you ALL the
# data available so it takes a while (obviously longer the bigger the area
# requested) and further processing will be needed to develop "final" datasets
# for a model simulation. if the subdirectories for the various databases in
# the output directory already exist (i.e., if you have already done the data
# download), this step will be skipped but it will tell the processor the
# location of the data files and read the metadata for further processing.
processor.download(bbox, start, end, output, datasets=datasets)
# because there are so many data files, it makes sense to keep track of some
# metadata for the files to use for parsing, etc. the following lines show
# how the metadata are organized. the filenames are dictionary keys, and the
# values are dictionaries containing the name of the station, the latitude,
# the longitude, the elevation, and the length of the datasets.
# GHCND (daily precip, tmax, tmin, snowdepth, snowfall, wind, pan evaporation
for k, v in list(processor.metadata.ghcndstations.items()):
print(('\n'.join(
['{} {} {}'.format(k, p, val) for p, val in list(v.items())])))
# GSOD (daily precip, tmax, tmin, wind, dewpoint)
# you may need to change the path to 7zip for your system; can also be done
# when the processor is instantiated (should not be necessary on linux)
processor.path_to_7z = r'C:/Program Files/7-Zip/7z.exe'
# download the data--worth noting there are some issues with the databases
# sometimes. these are handled with warnings, but this will give you ALL the
# data available so it takes a while (obviously longer the bigger the area
# requested) and further processing will be needed to develop "final" datasets
# for a model simulation. if the subdirectories for the various databases in
# the output directory already exist (i.e., if you have already done the data
# download), this step will be skipped but it will tell the processor the
# location of the data files and read the metadata for further processing.
processor.download(bbox, start, end, output, datasets = datasets)
# because there are so many data files, it makes sense to keep track of some
# metadata for the files to use for parsing, etc. the following lines show
# how the metadata are organized. the filenames are dictionary keys, and the
# values are dictionaries containing the name of the station, the latitude,
# the longitude, the elevation, and the length of the datasets.
# GHCND (daily precip, tmax, tmin, snowdepth, snowfall, wind, pan evaporation
for k, v in processor.metadata.ghcndstations.items():
print('\n'.join(['{} {} {}'.format(k, p, val) for p, val in v.items()]))
# GSOD (daily precip, tmax, tmin, wind, dewpoint)
bbox = -77.2056, 38.2666, -76.4008, 39.3539
# start and end dates (aggregate the whole 31 years)
start = datetime.datetime(1980, 1, 1)
end = datetime.datetime(2011, 1, 1)
# path to 7zip if using windows
processor.path_to_7z = r'C:/Program Files/7-Zip/7z.exe'
# download the data; this step will be skipped if this has already been done
# (from the last example); alternatively it will set the metadata
processor.download(bbox, start, end, output, datasets = ['NSRDB'])
# aggregate the data -- worth noting that prior to 1991 is a separate database
# that corresponds more closely to METSTAT than SUNY; requesting data prior
# to 1991 will give the same values either way.
metstat = processor.aggregate('NSRDB', 'metstat', start, end)
suny = processor.aggregate('NSRDB', 'suny', start, end)
# now these time series can be saved for later consistent with the structure
# used by PyHSPF's HSPFModel class (start date, time step in minutes, data)
# this way the data are easy to access later
for n, dataset in zip(('metstat','suny'), (metstat, suny)):
name = '{}/NSRDB_aggregated_{}'.format(output, n)
