siphon-ncss

Extracting data with the NetCDF subset service (`siphon.ncss`)

We got the url by navigating the THEDDS catalog because we could not find it navigating the metadata with siphon. See siphon-explore-catalog.ipynb.



In [1]:

    
from siphon.ncss import NCSS

subset = NCSS('http://thredds.cencoos.org/thredds/ncss/'
              'G1_SST_US_WEST_COAST.nc')

Let's inspect the `subset` object.



In [2]:

    
[value for value in dir(subset) if not value.startswith('__')]









    Out[2]:





['_base',
 '_get_metadata',
 '_session',
 'get',
 'get_data',
 'get_data_raw',
 'get_path',
 'get_query',
 'metadata',
 'query',
 'unit_handler',
 'url_path',
 'validate_query',
 'variables']

`subset.metadata` seems to be what we want.



In [3]:

    
[value for value in dir(subset.metadata) if not value.startswith('__')]









    Out[3]:





['_get_handler',
 '_parse_AcceptList',
 '_parse_LatLonBox',
 '_parse_TimeSpan',
 '_parse_axis',
 '_parse_coordTransform',
 '_parse_element',
 '_parse_featureDataset',
 '_parse_gridset',
 '_parse_variable',
 '_types',
 '_types_methods',
 'accept_list',
 'axes',
 'featureDataset',
 'gridsets',
 'lat_lon_box',
 'time_span',
 'variables']

Before making our own request we need to know the data time and space domains.



In [4]:

    
subset.metadata.time_span









    Out[4]:





{'begin': '2010-12-01T00:00:00Z', 'end': '2013-12-31T00:00:00Z'}



In [5]:

    
subset.metadata.lat_lon_box









    Out[5]:





{'east': -105.0049, 'north': 54.9949, 'south': 23.995, 'west': -140.005}

And what variables are available.



In [6]:

    
subset.variables









    Out[6]:





{'analysed_sst'}



In [7]:

    
subset.metadata.variables









    Out[7]:





{'analysed_sst': {'attributes': {'coordinates': 'time lat lon lat lon',
   'coverage_content_type': 'modelResult',
   'long_name': 'analysed sea surface temperature',
   'missing_value': '-273.15',
   'standard_name': 'sea_surface_temperature',
   'type': 'foundation',
   'units': 'degree_Celsius'},
  'desc': 'analysed sea surface temperature',
  'shape': 'time lat lon',
  'type': 'float'}}

Now we can create a query.



In [8]:

    
query = subset.query()

Add space and time constraints.



In [9]:

    
query.lonlat_box(east=-120, north=50, south=35, west=-135)









    Out[9]:





west=-135&north=50&south=35&east=-120



In [10]:

    
from datetime import datetime

query.time(datetime(2013, 12, 31))









    Out[10]:





time=2013-12-31T00%3A00%3A00&west=-135&north=50&south=35&east=-120

And the variable constraint.



In [11]:

    
variable = 'analysed_sst'

query.variables(variable)









    Out[11]:





var=analysed_sst&time=2013-12-31T00%3A00%3A00&west=-135&north=50&south=35&east=-120

`query` prints the constraints we have so far but you can also inspect them.



In [12]:

    
query.spatial_query, query.time_query









    Out[12]:





(OrderedDict([('west', -135), ('north', 50), ('south', 35), ('east', -120)]),
 OrderedDict([('time', '2013-12-31T00:00:00')]))

To request the data we need to pass the `query` object back to the `subset` object.



In [13]:

    
import time

start_time = time.time()

data = subset.get_data(query)

elapsed = time.time() - start_time
print('{:.2f} s'.format(elapsed))



In [14]:

    
type(data)









    Out[14]:





netCDF4._netCDF4.Dataset

`siphon.ncss` returns by default a `netCDF4-python` object. Nice!

But note that it took more time to get the data than we would expect if doing the same using `OPeNDAP`. The reason for that is because siphon is downloading all the data and saving it in the disk.



In [15]:

    
data.filepath()









    Out[15]:





'/tmp/tmpglbcyjbg'

Does the data real limits make sense with our constraints?



In [16]:

    
import numpy.ma as ma

lon = data['lon'][:]
lat = data['lat'][:]

temp = data[variable][:]
temp = ma.masked_invalid(temp.squeeze())

time = data['time']
time = ''.join(time[:][0].astype(str).tolist())



In [17]:

    
lon[0], lon[-1], lat[0], lat[-1], time









    Out[17]:





(-134.995, -119.995, 34.994999, 50.005001, '2013-12-31T00:00:00Z')

They are all within the queried constraints as expected.

Let's end the example with a plot!



In [18]:

    
%matplotlib inline

import cartopy.crs as ccrs
import matplotlib.pyplot as plt

from cmocean import cm
from cartopy.mpl.gridliner import LONGITUDE_FORMATTER, LATITUDE_FORMATTER

fig, ax = plt.subplots(figsize=(9, 9),
                       subplot_kw=dict(projection=ccrs.PlateCarree()))

ax.coastlines(resolution='50m')
cs = ax.pcolormesh(lon, lat, temp, cmap=cm.thermal)
cbar = fig.colorbar(cs, extend='both', shrink=0.75)

gl = ax.gridlines(draw_labels=True)
gl.xlabels_top = gl.ylabels_right = False
gl.xformatter = LONGITUDE_FORMATTER
gl.yformatter = LATITUDE_FORMATTER

siphon-ncss

Extracting data with the NetCDF subset service (siphon.ncss)

We got the url by navigating the THEDDS catalog because we could not find it navigating the metadata with siphon. See siphon-explore-catalog.ipynb.

Let's inspect the subset object.

subset.metadata seems to be what we want.

Before making our own request we need to know the data time and space domains.

And what variables are available.

Now we can create a query.

Add space and time constraints.

And the variable constraint.

query prints the constraints we have so far but you can also inspect them.

To request the data we need to pass the query object back to the subset object.

siphon.ncss returns by default a netCDF4-python object. Nice!

But note that it took more time to get the data than we would expect if doing the same using OPeNDAP. The reason for that is because siphon is downloading all the data and saving it in the disk.

Does the data real limits make sense with our constraints?

They are all within the queried constraints as expected.

Let's end the example with a plot!

Extracting data with the NetCDF subset service (`siphon.ncss`)

Let's inspect the `subset` object.

`subset.metadata` seems to be what we want.

`query` prints the constraints we have so far but you can also inspect them.

To request the data we need to pass the `query` object back to the `subset` object.

`siphon.ncss` returns by default a `netCDF4-python` object. Nice!

But note that it took more time to get the data than we would expect if doing the same using `OPeNDAP`. The reason for that is because siphon is downloading all the data and saving it in the disk.