# This software is open source software available under the BSD-3 license.
#
# Copyright (c) 2022 Triad National Security, LLC. All rights reserved.
# Copyright (c) 2022 Lawrence Livermore National Security, LLC. All rights
# reserved.
# Copyright (c) 2022 UT-Battelle, LLC. All rights reserved.
#
# Additional copyright and license information can be found in the LICENSE file
# distributed with this code, or at
# https://raw.githubusercontent.com/MPAS-Dev/MPAS-Analysis/main/LICENSE
"""
Functions for plotting remapped horizontal fields (and comparing with reference
data sets)
"""
# Authors
# -------
# Xylar Asay-Davis, Milena Veneziani, Luke Van Roekel, Greg Streletz
import matplotlib
import matplotlib.pyplot as plt
import matplotlib.colors as cols
import matplotlib.ticker as mticker
import matplotlib.patches as mpatches
import numpy as np
from mpl_toolkits.axes_grid1 import make_axes_locatable
from mpl_toolkits.axes_grid1.inset_locator import inset_axes
import cartopy
from cartopy.util import add_cyclic_point
from mpas_analysis.shared.plot.colormap import setup_colormap
from mpas_analysis.shared.plot.title import limit_title
from mpas_analysis.shared.plot.save import savefig
from mpas_analysis.shared.projection import get_cartopy_projection
[docs]
def plot_polar_comparison(
config,
lon,
lat,
modelArray,
refArray,
diffArray,
colorMapSectionName,
fileout,
title=None,
plotProjection='npstere',
latmin=50.0,
lon0=0,
modelTitle='Model',
refTitle='Observations',
diffTitle='Model-Observations',
cbarlabel='units',
titleFontSize=None,
defaultFontSize=None,
figsize=None,
dpi=None,
vertical=False,
maxTitleLength=None):
"""
Plots a data set around either the north or south pole.
Parameters
----------
config : tranche.Tranche
the configuration, containing a [plot] section with options that
control plotting
lon, lat : float arrays
longitude and latitude arrays
modelArray, refArray : numpy.ndarray
model and observational or control run data sets
diffArray : float array
difference between modelArray and refArray
colorMapSectionName : str
section name in ``config`` where color map info can be found.
fileout : str
the file name to be written
title : str, optional
the subtitle of the plot
plotProjection : {'npstere', 'spstere'}, optional
projection for the plot (north or south pole)
modelTitle : str, optional
title of the model panel
refTitle : str, optional
title of the observations or control run panel
diffTitle : str, optional
title of the difference (bias) panel
cbarlabel : str, optional
label on the colorbar
titleFontSize : int, optional
size of the title font
defaultFontSize : int, optional
the size of text other than the title
figsize : tuple of float, optional
the size of the figure in inches. If ``None``, the figure size is
``(8, 22)`` if ``vertical == True`` and ``(22, 8)`` otherwise.
dpi : int, optional
the number of dots per inch of the figure, taken from section ``plot``
option ``dpi`` in the config file by default
vertical : bool, optional
whether the subplots should be stacked vertically rather than
horizontally
maxTitleLength : int or None, optional
the maximum number of characters in the title, beyond which it is
truncated with a trailing ellipsis. The default is from the
``maxTitleLength`` config option.
"""
# Authors
# -------
# Xylar Asay-Davis, Milena Veneziani
def _do_subplot(ax, field, title, colormap, norm, levels, ticks, contours,
lineWidth, lineColor, **kwargs):
"""
Make a subplot within the figure.
"""
data_crs = cartopy.crs.PlateCarree()
ax.set_extent(extent, crs=data_crs)
title = limit_title(title, maxTitleLength)
ax.set_title(title, y=1.06, **plottitle_font)
gl = ax.gridlines(crs=data_crs, color='k', linestyle=':', zorder=5,
draw_labels=True)
gl.xlocator = mticker.FixedLocator(np.arange(-180., 181., 20.))
gl.ylocator = mticker.FixedLocator(np.arange(-80., 81., 10.))
gl.n_steps = 100
gl.right_labels = False
gl.xformatter = cartopy.mpl.gridliner.LONGITUDE_FORMATTER
gl.yformatter = cartopy.mpl.gridliner.LATITUDE_FORMATTER
gl.rotate_labels = False
fieldPeriodic, lonPeriodic = add_cyclic_point(field, lon)
LonsPeriodic, LatsPeriodic = np.meshgrid(lonPeriodic, lat)
if levels is None:
plotHandle = ax.pcolormesh(LonsPeriodic, LatsPeriodic,
fieldPeriodic, cmap=colormap,
norm=norm, transform=data_crs,
zorder=1, rasterized=True)
else:
plotHandle = ax.contourf(LonsPeriodic, LatsPeriodic,
fieldPeriodic, cmap=colormap,
norm=norm, levels=levels,
transform=data_crs,
zorder=1)
_add_land_lakes_coastline(ax)
if contours is not None:
matplotlib.rcParams['contour.negative_linestyle'] = 'solid'
ax.contour(LonsPeriodic, LatsPeriodic, fieldPeriodic,
levels=contours, colors=lineColor,
linewidths=lineWidth, transform=data_crs)
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1,
axes_class=plt.Axes)
cbar = plt.colorbar(plotHandle, cax=cax)
cbar.set_label(cbarlabel)
if ticks is not None:
cbar.set_ticks(ticks)
cbar.set_ticklabels(['{}'.format(tick) for tick in ticks])
if maxTitleLength is None:
maxTitleLength = config.getint('plot', 'maxTitleLength')
if defaultFontSize is None:
defaultFontSize = config.getint('plot', 'defaultFontSize')
matplotlib.rc('font', size=defaultFontSize)
if dpi is None:
dpi = config.getint('plot', 'dpi')
dictModelRef = setup_colormap(config, colorMapSectionName, suffix='Result')
dictDiff = setup_colormap(config, colorMapSectionName, suffix='Difference')
if refArray is None:
if figsize is None:
figsize = (8, 8.5)
subplots = [111]
elif vertical:
if figsize is None:
figsize = (8, 22)
subplots = [311, 312, 313]
else:
if figsize is None:
figsize = (22, 7.5)
subplots = [131, 132, 133]
fig = plt.figure(figsize=figsize, dpi=dpi)
if (title is not None):
if titleFontSize is None:
titleFontSize = config.get('plot', 'titleFontSize')
title_font = {'size': titleFontSize,
'color': config.get('plot', 'titleFontColor'),
'weight': config.get('plot', 'titleFontWeight')}
fig.suptitle(title, y=0.95, **title_font)
plottitle_font = {'size': config.get('plot',
'threePanelPlotTitleFontSize')}
if plotProjection == 'npstere':
projection = cartopy.crs.NorthPolarStereo()
extent = [-180, 180, latmin, 90]
elif plotProjection == 'spstere':
projection = cartopy.crs.SouthPolarStereo()
extent = [-180, 180, -90, latmin]
else:
raise ValueError('Unexpected plot projection {}'.format(
plotProjection))
ax = plt.subplot(subplots[0], projection=projection)
_do_subplot(ax=ax, field=modelArray, title=modelTitle, **dictModelRef)
if refArray is not None:
ax = plt.subplot(subplots[1], projection=projection)
_do_subplot(ax=ax, field=refArray, title=refTitle, **dictModelRef)
ax = plt.subplot(subplots[2], projection=projection)
_do_subplot(ax=ax, field=diffArray, title=diffTitle, **dictDiff)
fig.canvas.draw()
plt.tight_layout(pad=4.)
if vertical:
plt.subplots_adjust(top=0.9)
if fileout is not None:
savefig(fileout, config)
plt.close()
[docs]
def plot_global_comparison(
config,
Lons,
Lats,
modelArray,
refArray,
diffArray,
colorMapSectionName,
fileout,
title=None,
modelTitle='Model',
refTitle='Observations',
diffTitle='Model-Observations',
cbarlabel='units',
titleFontSize=None,
defaultFontSize=None,
figsize=None,
dpi=None,
lineWidth=1,
lineColor='black',
maxTitleLength=None,
extend='both'):
"""
Plots a data set as a longitude/latitude map.
Parameters
----------
config : tranche.Tranche
the configuration, containing a [plot] section with options that
control plotting
Lons, Lats : numpy.ndarray
longitude and latitude arrays
modelArray, refArray : numpy.ndarray
model and observational or control run data sets
diffArray : float array
difference between modelArray and refArray
colorMapSectionName : str
section name in ``config`` where color map info can be found.
fileout : str
the file name to be written
title : str, optional
the subtitle of the plot
modelTitle : str, optional
title of the model panel
refTitle : str, optional
title of the observations or control run panel
diffTitle : str, optional
title of the difference (bias) panel
cbarlabel : str, optional
label on the colorbar
titleFontSize : int, optional
size of the title font
defaultFontSize : int, optional
the size of text other than the title
figsize : tuple of float, optional
the size of the figure in inches
dpi : int, optional
the number of dots per inch of the figure, taken from section ``plot``
option ``dpi`` in the config file by default
lineWidth : int, optional
the line width of contour lines (if specified)
lineColor : str, optional
the color of contour lines (if specified)
maxTitleLength : int or None, optional
the maximum number of characters in the title, beyond which it is
truncated with a trailing ellipsis. The default is from the
``maxTitleLength`` config option.
extend : {'neither', 'both', 'min', 'max'}, optional
Determines the ``contourf``-coloring of values that are outside the
range of the levels provided if using an indexed colormap.
"""
# Authors
# -------
# Xylar Asay-Davis, Milena Veneziani
def _plot_panel(ax, title, array, colormap, norm, levels, ticks, contours,
lineWidth, lineColor, **kwargs):
ax.set_extent(extent, crs=projection)
title = limit_title(title, maxTitleLength)
ax.set_title(title, y=1.02, **plottitle_font)
gl = ax.gridlines(crs=projection, color='k', linestyle=':', zorder=5,
draw_labels=True)
gl.right_labels = False
gl.top_labels = False
gl.xlocator = mticker.FixedLocator(np.arange(-180., 181., 60.))
gl.ylocator = mticker.FixedLocator(np.arange(-80., 81., 20.))
gl.xformatter = cartopy.mpl.gridliner.LONGITUDE_FORMATTER
gl.yformatter = cartopy.mpl.gridliner.LATITUDE_FORMATTER
if levels is None:
plotHandle = ax.pcolormesh(Lons, Lats, array, cmap=colormap,
norm=norm, transform=projection,
zorder=1, rasterized=True)
else:
plotHandle = ax.contourf(Lons, Lats, array, cmap=colormap,
norm=norm, levels=levels, extend=extend,
transform=projection, zorder=1)
_add_land_lakes_coastline(ax)
if contours is not None:
matplotlib.rcParams['contour.negative_linestyle'] = 'solid'
ax.contour(Lons, Lats, array, levels=contours, colors=lineColor,
linewidths=lineWidth, transform=projection)
cax = inset_axes(ax, width='5%', height='60%', loc='center right',
bbox_to_anchor=(0.08, 0., 1, 1),
bbox_transform=ax.transAxes, borderpad=0)
cbar = plt.colorbar(plotHandle, cax=cax)
cbar.set_label(cbarlabel)
if ticks is not None:
cbar.set_ticks(ticks)
cbar.set_ticklabels(['{}'.format(tick) for tick in ticks])
if maxTitleLength is None:
maxTitleLength = config.getint('plot', 'maxTitleLength')
if defaultFontSize is None:
defaultFontSize = config.getint('plot', 'defaultFontSize')
matplotlib.rc('font', size=defaultFontSize)
# set up figure
if dpi is None:
dpi = config.getint('plot', 'dpi')
if figsize is None:
# set the defaults, depending on if we have 1 or 3 panels
if refArray is None:
figsize = (8, 5)
else:
figsize = (8, 13)
fig = plt.figure(figsize=figsize, dpi=dpi)
if (title is not None):
if titleFontSize is None:
titleFontSize = config.get('plot', 'titleFontSize')
title_font = {'size': titleFontSize,
'color': config.get('plot', 'titleFontColor'),
'weight': config.get('plot', 'titleFontWeight')}
fig.suptitle(title, y=0.935, **title_font)
plottitle_font = {'size': config.get('plot',
'threePanelPlotTitleFontSize')}
if refArray is None:
subplots = [111]
else:
subplots = [311, 312, 313]
projection = cartopy.crs.PlateCarree()
extent = [-180, 180, -85, 85]
dictModelRef = setup_colormap(config, colorMapSectionName, suffix='Result')
dictDiff = setup_colormap(config, colorMapSectionName, suffix='Difference')
axes = []
ax = plt.subplot(subplots[0], projection=projection)
_plot_panel(ax, modelTitle, modelArray, **dictModelRef)
axes.append(ax)
if refArray is not None:
ax = plt.subplot(subplots[1], projection=projection)
_plot_panel(ax, refTitle, refArray, **dictModelRef)
axes.append(ax)
ax = plt.subplot(subplots[2], projection=projection)
_plot_panel(ax, diffTitle, diffArray, **dictDiff)
axes.append(ax)
_add_stats(modelArray, refArray, diffArray, Lats, axes)
if fileout is not None:
savefig(fileout, config, pad_inches=0.2)
plt.close()
def plot_projection_comparison(
config,
x,
y,
landMask,
modelArray,
refArray,
diffArray,
fileout,
colorMapSectionName,
projectionName,
title=None,
modelTitle='Model',
refTitle='Observations',
diffTitle='Model-Observations',
cbarlabel='units',
titleFontSize=None,
cartopyGridFontSize=None,
defaultFontSize=None,
vertical=False,
maxTitleLength=None,
extend='both'):
"""
Plots a data set as a projection map.
Parameters
----------
config : tranche.Tranche
the configuration, containing a [plot] section with options that
control plotting
x, y : numpy.ndarrays
1D x and y arrays of the corners of grid cells on the projection grid
landMask : numpy.ndarrays
model and observational or control run data sets
modelArray, refArray : numpy.ndarrays
model and observational or control run data sets
diffArray : float array
difference between modelArray and refArray
fileout : str
the file name to be written
colorMapSectionName : str
section name in ``config`` where color map info can be found.
title : str, optional
the subtitle of the plot
modelTitle : str, optional
title of the model panel
refTitle : str, optional
title of the observations or control run panel
diffTitle : str, optional
title of the difference (bias) panel
cbarlabel : str, optional
label on the colorbar
titleFontSize : int, optional
size of the title font
cartopyGridFontSize : int, optional
the size of text used by cartopy to label lon and lat
defaultFontSize : int, optional
the size of text other than the title
vertical : bool, optional
whether the subplots should be stacked vertically rather than
horizontally
projectionName : str, optional
the name of projection that the data is on, one of the projections
available via
:py:func:`mpas_analysis.shared.projection.get_cartopy_projection()`.
maxTitleLength : int or None, optional
the maximum number of characters in the title, beyond which it is
truncated with a trailing ellipsis. The default is from the
``maxTitleLength`` config option.
extend : {'neither', 'both', 'min', 'max'}, optional
Determines the ``contourf``-coloring of values that are outside the
range of the levels provided if using an indexed colormap.
"""
# Authors
# -------
# Xylar Asay-Davis
def _add_arrow_to_line_2d(ax, poly, arrow_spacing, arrow_width):
"""
https://stackoverflow.com/a/27637925/7728169
Add arrows to a matplotlib.lines.Line2D at selected locations.
Polygons instead of paths, following
https://gis.stackexchange.com/a/246861/143986
"""
x = poly[:, 0]
y = poly[:, 1]
arrows = []
delta = np.sqrt(np.diff(x) ** 2 + np.diff(y) ** 2)
s = np.cumsum(delta)
indices = np.searchsorted(
s, arrow_spacing*np.arange(1, int(s[-1]/arrow_spacing)))
for n in indices:
dx = np.mean(x[n-2:n]) - x[n]
dy = np.mean(y[n-2:n]) - y[n]
p = mpatches.FancyArrow(
x[n], y[n], dx, dy, length_includes_head=False,
width=arrow_width, facecolor='k')
ax.add_patch(p)
arrows.append(p)
return arrows
def _plot_panel(ax, title, array, colormap, norm, levels, ticks, contours,
lineWidth, lineColor, arrowSpacing, arrowWidth):
title = limit_title(title, maxTitleLength)
ax.set_title(title, **plottitle_font)
ax.set_extent(extent, crs=projection)
gl = ax.gridlines(crs=cartopy.crs.PlateCarree(), color='k',
linestyle=':', zorder=5, draw_labels=True)
gl.xlocator = mticker.FixedLocator(lonLines)
gl.ylocator = mticker.FixedLocator(latLines)
gl.n_steps = 100
gl.right_labels = False
gl.left_labels = left_labels
gl.xformatter = cartopy.mpl.gridliner.LONGITUDE_FORMATTER
gl.yformatter = cartopy.mpl.gridliner.LATITUDE_FORMATTER
gl.xlabel_style['size'] = cartopyGridFontSize
gl.ylabel_style['size'] = cartopyGridFontSize
gl.rotate_labels = False
if levels is None:
plotHandle = ax.pcolormesh(x, y, array, cmap=colormap, norm=norm,
rasterized=True)
else:
plotHandle = ax.contourf(xCenter, yCenter, array, cmap=colormap,
norm=norm, levels=levels, extend=extend)
if useCartopyCoastline:
_add_land_lakes_coastline(ax, ice_shelves=False)
else:
# add the model coastline
plt.pcolormesh(x, y, landMask, cmap=landColorMap)
plt.contour(xCenter, yCenter, landMask.mask, (0.5,), colors='k',
linewidths=0.5)
if contours is not None:
# add arrows to streamlines
arrows = arrowSpacing is not None and arrowWidth is not None
special_zero = arrows and 0. in contours
if special_zero:
# treat the zero contour as a special case
contours = [contour for contour in contours if contour != 0.]
matplotlib.rcParams['contour.negative_linestyle'] = 'solid'
x_center = 0.5*(x[0:-1] + x[1:])
y_center = 0.5*(y[0:-1] + y[1:])
cs = ax.contour(x_center, y_center, array, levels=contours,
colors=lineColor, linewidths=lineWidth)
for path in cs.get_paths():
for poly in path.to_polygons():
_add_arrow_to_line_2d(ax, poly, arrowSpacing, arrowWidth)
if special_zero:
ax.contour(x_center, y_center, array, levels=[0.],
colors=lineColor, linewidths=1.5 * lineWidth)
# create an axes on the right side of ax. The width of cax will be 5%
# of ax and the padding between cax and ax will be fixed at 0.05 inch.
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.05,
axes_class=plt.Axes)
cbar = plt.colorbar(plotHandle, cax=cax)
cbar.set_label(cbarlabel)
if ticks is not None:
cbar.set_ticks(ticks)
cbar.set_ticklabels(['{}'.format(tick) for tick in ticks])
if maxTitleLength is None:
maxTitleLength = config.getint('plot', 'maxTitleLength')
if defaultFontSize is None:
defaultFontSize = config.getint('plot', 'defaultFontSize')
matplotlib.rc('font', size=defaultFontSize)
if cartopyGridFontSize is None:
cartopyGridFontSize = config.getint('plot', 'cartopyGridFontSize')
# set up figure
dpi = config.getint('plot', 'dpi')
section = f'plot_{projectionName}'
useCartopyCoastline = config.getboolean(section, 'useCartopyCoastline')
if refArray is None:
figsize = config.getexpression(section, 'onePanelFigSize')
subplots = [111]
elif vertical:
figsize = config.getexpression(section, 'threePanelVertFigSize')
subplots = [311, 312, 313]
else:
figsize = config.getexpression(section, 'threePanelHorizFigSize')
subplots = [131, 132, 133]
latLines = config.getnumpy(section, 'latLines')
lonLines = config.getnumpy(section, 'lonLines')
# put latitude labels on the left unless we're in a polar projection
left_labels = projectionName not in ['arctic', 'antarctic']
dictModelRef = setup_colormap(config, colorMapSectionName, suffix='Result')
dictDiff = setup_colormap(config, colorMapSectionName, suffix='Difference')
fig = plt.figure(figsize=figsize, dpi=dpi)
if title is not None:
if titleFontSize is None:
titleFontSize = config.get('plot', 'titleFontSize')
title_font = {'size': titleFontSize,
'color': config.get('plot', 'titleFontColor'),
'weight': config.get('plot', 'titleFontWeight')}
fig.suptitle(title, y=0.95, **title_font)
plottitle_font = {'size': config.get('plot',
'threePanelPlotTitleFontSize')}
# set up land colormap
if not useCartopyCoastline:
colorList = [(0.8, 0.8, 0.8), (0.8, 0.8, 0.8)]
landColorMap = cols.LinearSegmentedColormap.from_list('land',
colorList)
# locations of centers for contour plots
xCenter = 0.5 * (x[1:] + x[0:-1])
yCenter = 0.5 * (y[1:] + y[0:-1])
projection = get_cartopy_projection(projectionName)
extent = [x[0], x[-1], y[0], y[-1]]
ax = plt.subplot(subplots[0], projection=projection)
_plot_panel(ax, modelTitle, modelArray, **dictModelRef)
if refArray is not None:
ax = plt.subplot(subplots[1], projection=projection)
_plot_panel(ax, refTitle, refArray, **dictModelRef)
ax = plt.subplot(subplots[2], projection=projection)
_plot_panel(ax, diffTitle, diffArray, **dictDiff)
if fileout is not None:
savefig(fileout, config)
plt.close()
def _add_stats(modelArray, refArray, diffArray, Lats, axes):
""" compute the means, std devs. and Pearson correlation """
weights = np.cos(np.deg2rad(Lats))
model_weights = weights
model_mask = None
if isinstance(modelArray, np.ma.MaskedArray):
# make sure we're using the MPAS land mask for all 3 sets of stats
model_mask = modelArray.mask
model_weights = np.ma.array(weights, mask=model_mask)
modelMean = np.average(modelArray, weights=model_weights)
_add_stats_text(
names=['Min', 'Mean', 'Max'],
values=[np.amin(modelArray), modelMean, np.amax(modelArray)],
ax=axes[0], loc='upper')
if refArray is not None:
ref_weights = weights
ref_mask = None
if isinstance(modelArray, np.ma.MaskedArray):
# make sure we're using the MPAS land mask for all 3 sets of stats
if isinstance(refArray, np.ma.MaskedArray):
# mask invalid where either model or ref array is invalid
ref_mask = np.logical_or(model_mask, refArray.mask)
ref_weights = np.ma.array(weights, mask=ref_mask)
refArray = np.ma.array(refArray, mask=ref_mask)
modelAnom = modelArray - modelMean
modelVar = np.average(modelAnom ** 2, weights=ref_weights)
refMean = np.average(refArray, weights=ref_weights)
refAnom = refArray - refMean
refVar = np.average(refAnom**2, weights=ref_weights)
_add_stats_text(
names=['Min', 'Mean', 'Max'],
values=[np.amin(refArray), refMean, np.amax(refArray)],
ax=axes[1], loc='upper')
diffMean = np.average(diffArray, weights=ref_weights)
diffVar = np.average((diffArray - diffMean)**2, weights=ref_weights)
diffRMSE = np.sqrt(diffVar)
_add_stats_text(
names=['Min', 'Mean', 'Max'],
values=[np.amin(diffArray), diffMean, np.amax(diffArray)],
ax=axes[2], loc='upper')
covar = np.average(modelAnom*refAnom, weights=ref_weights)
corr = covar/np.sqrt(modelVar*refVar)
_add_stats_text(
names=['RMSE', 'Corr'],
values=[diffRMSE, corr],
ax=axes[2], loc='lower')
def _add_stats_text(names, values, ax, loc):
if loc == 'upper':
text_ax = inset_axes(ax, width='17%', height='20%', loc='upper right',
bbox_to_anchor=(0.2, 0.1, 1., 1.),
bbox_transform=ax.transAxes, borderpad=0)
else:
text_ax = inset_axes(ax, width='17%', height='20%', loc='lower right',
bbox_to_anchor=(0.2, 0.03, 1., 1.),
bbox_transform=ax.transAxes, borderpad=0)
text = '\n'.join(names)
text_ax.text(0., 0., text, fontsize=10, horizontalalignment='left')
text = '\n'.join(['{:6.4g}'.format(val) for val in values])
text_ax.text(1., 0., text, fontsize=10, horizontalalignment='right')
text_ax.axis('off')
def _add_land_lakes_coastline(ax, ice_shelves=True):
land_50m = cartopy.feature.NaturalEarthFeature(
'physical', 'land', '50m', edgecolor='k',
facecolor='#cccccc', linewidth=0.5)
lakes_50m = cartopy.feature.NaturalEarthFeature(
'physical', 'lakes', '50m', edgecolor='k',
facecolor='white',
linewidth=0.5)
ax.add_feature(land_50m, zorder=2)
if ice_shelves:
ice_50m = cartopy.feature.NaturalEarthFeature(
'physical', 'antarctic_ice_shelves_polys', '50m',
edgecolor='k', facecolor='lightgray', linewidth=0.5)
ax.add_feature(ice_50m, zorder=3)
ax.add_feature(lakes_50m, zorder=4)