Multiprocessing satellite example from MetPy Monday #41.

multiprocessing_satellite.py

import multiprocessing as mp
from datetime import datetime

import cartopy.crs as ccrs
import cartopy.feature as cfeature
import matplotlib.pyplot as plt
from matplotlib import patheffects
from metpy.plots.ctables import registry
from siphon.catalog import TDSCatalog


def plot_dataset(dataset):
    print('Plotting dataset: ', dataset)

    # Get dataset
    ds = dataset.remote_access(service='OPENDAP')

    # Pull out what we need from the GOES netCDF file
    data_var = ds.variables['Sectorized_CMI']
    x = ds.variables['x'][:]
    y = ds.variables['y'][:]
    proj_var = ds.variables[data_var.grid_mapping]

    # Create a Globe specifying a spherical earth with the correct radius
    globe = ccrs.Globe(ellipse='sphere', semimajor_axis=proj_var.semi_major,
                       semiminor_axis=proj_var.semi_minor)

    proj = ccrs.LambertConformal(central_longitude=proj_var.longitude_of_central_meridian,
                                 central_latitude=proj_var.latitude_of_projection_origin,
                                 standard_parallels=[proj_var.standard_parallel],
                                 globe=globe)

    # Make the plot
    fig = plt.figure(figsize=(1.375 * 40, 40))
    ax = fig.add_subplot(1, 1, 1, projection=proj)
    plt.subplots_adjust(left=0, bottom=0, right=1, top=1, wspace=0, hspace=0)

    wv_norm, wv_cmap = registry.get_with_range('WVCIMSS_r', 195, 265)
    im = ax.imshow(data_var[:], extent=(x.min(), x.max(), y.min(), y.max()),
                   origin='upper', transform=proj)
    im.set_cmap(wv_cmap)
    im.set_norm(wv_norm)

    ax.add_feature(cfeature.BORDERS, linewidth=8, edgecolor='black')
    ax.add_feature(cfeature.STATES.with_scale('50m'), linestyle='-',
                   edgecolor='black', linewidth=4)

    timestamp = datetime.strptime(ds.start_date_time, '%Y%j%H%M%S')

    # Add text and save the returned object so we can manipulate it.
    text_time = ax.text(0.01, 0.01, timestamp.strftime('%d %B %Y %H%MZ'),
                        horizontalalignment='left', transform=ax.transAxes,
                        color='white', fontsize=100, weight='bold')

    outline_effect = [patheffects.withStroke(linewidth=15, foreground='black')]
    text_time.set_path_effects(outline_effect)

    ax.set_extent([-124.5, -105, 38.5, 50])
    ax.gridlines(linestyle=':', color='black', linewidth=2)

    filename = 'image_{}.png'.format(dataset)
    plt.savefig(filename)
    return filename

# Use Agg backend
plt.switch_backend('Agg')

# Get a catalog of GOES 16 data
date = datetime.utcnow()
channel = 8
region = 'CONUS'
cat = TDSCatalog('http://thredds-test.unidata.ucar.edu/thredds/'
                 'catalog/satellite/goes16/GOES16/'
                 '{}/Channel{:02d}/{:%Y%m%d}/catalog.xml'.format(region, channel, date))

with mp.Pool(processes=4) as pool:
    pool.map(plot_dataset, cat.datasets[-10:])
    pool.close()
    pool.join()

Thanks for the gist! But isn't pool being closed in the with context automatically?

I do like using multiprocessing to quickly make lots of plots.