Merge pull request #30 from dcherian/panel_3

Panel 3

Plots/Panels/README.rst

@@ -0,0 +1,6 @@
+.. _panels_examples:
+
+.. _panels-examples-index:
+
+Panels
+======

Plots/Panels/panel_3.py

@@ -0,0 +1,203 @@
+"""
+panel_3.py
+===============
+Two panel image with shared colorbar and title
+
+NCL docs: https://www.ncl.ucar.edu/Applications/panel.shtml
+
+NCL script: https://www.ncl.ucar.edu/Applications/Scripts/panel_3.ncl
+
+NCL output: https://www.ncl.ucar.edu/Applications/Images/panel_3_lg.png
+"""
+
+###############################################################################
+# Concepts illustrated:
+#   - Paneling plots vertically on a page ``plt.subplots``
+#   - Adding a common title to paneled plots ``matplotlib.Figure.suptitle``
+#   - Adding a common labelbar (or colorbar) to paneled plots ``matplotlib.Figure.colorbar``
+#   - Subsetting a color map
+
+###############################################################################
+# Lets read the netCDF dataset using xarray and choose the second timestamp.
+import cartopy.crs as ccrs
+import matplotlib as mpl
+import matplotlib.pyplot as plt
+import numpy as np
+import xarray as xr
+
+ds = xr.open_dataset("../../data/netcdf_files/uv300.nc").isel(time=1)
+
+###############################################################################
+# Now we will define a few utility functions.
+# The plot requires filled colored contours, with black labelled contours, so we'll
+# make utility function that does all that.
+#
+# We will then call this function twice, once with `ds.U` and then with `ds.V`
+
+levels = np.arange(-10, 46, 5)
+
+
+###############################################################################
+# The NCL example uses a truncated colormap. Here's a small function that does that
+
+
+def truncate_colormap(cmap, minval=0.0, maxval=1.0, n=100):
+    """
+    Utility function that truncates a colormap. Copied from  https://stackoverflow.com/questions/18926031/how-to-extract-a-subset-of-a-colormap-as-a-new-colormap-in-matplotlib
+    """
+
+    new_cmap = mpl.colors.LinearSegmentedColormap.from_list(
+        name="trunc({n},{a:.2f},{b:.2f})".format(n=cmap.name, a=minval, b=maxval),
+        colors=cmap(np.linspace(minval, maxval, n)),
+    )
+    return new_cmap
+
+
+###############################################################################
+# This is the main plotting function. We do this so as not to repeat many lines of code since we
+# need to make the same figure with two different variables.
+
+
+def plot_labelled_filled_contours(data, ax=None):
+    """
+    A utility function for convenience that plots labelled, filled contours with black contours
+    marking each level.It will return a dictionary containing three objects corresponding to the
+    filled contours, the black contours, and the contour labels.
+    """
+
+    cmap = truncate_colormap(mpl.cm.rainbow, 0.2, 0.9)
+
+    handles = dict()
+    handles["filled"] = data.plot.contourf(
+        ax=ax,  # this is the axes we want to plot to
+        cmap=cmap,  # our special colormap
+        levels=levels,  # contour levels specified outside this function
+        xticks=np.arange(-180, 181, 30),  # nice x ticks
+        yticks=np.arange(-90, 91, 30),  # nice y ticks
+        transform=ccrs.PlateCarree(),  # data projection
+        add_colorbar=False,  # don't add individual colorbars for each plot call
+        add_labels=False,  # turn off xarray's automatic Lat, lon labels
+    )
+
+    # matplotlib's contourf doesn't let you specify the "edgecolors" (MATLAB terminology)
+    # instead we plot black contours on top of the filled contours
+    handles["contour"] = data.plot.contour(
+        ax=ax,
+        levels=levels,
+        colors="k",  # note plurals in this and following kwargs
+        linestyles="-",
+        linewidths=0.5,
+        add_labels=False,  # again turn off automatic labels
+    )
+
+    # Label the contours
+    ax.clabel(
+        handles["contour"], fontsize="small", fmt="%.0f",  # Turn off decimal points
+    )
+
+    # make a nice title
+    title = f"{data.attrs['long_name']} [{data.attrs['units']}]"
+    ax.set_title(title, loc="left", y=1.05)
+
+    return handles
+
+
+###############################################################################
+# Here's how this function works
+
+ax = plt.gca(projection=ccrs.PlateCarree())
+plot_labelled_filled_contours(ds.U, ax)
+
+###############################################################################
+# These next two functions add nice axes decorations and make the plot look more
+# like NCL
+
+
+def add_lat_lon_ticklabels(ax):
+    """
+    Nice latitude, longitude tick labels
+    """
+    from cartopy.mpl.ticker import LongitudeFormatter, LatitudeFormatter
+
+    lon_formatter = LongitudeFormatter(
+        zero_direction_label=False, dateline_direction_label=False
+    )
+    lat_formatter = LatitudeFormatter()
+    ax.xaxis.set_major_formatter(lon_formatter)
+    ax.yaxis.set_major_formatter(lat_formatter)
+
+
+def nclize_axis(ax):
+    """
+    Utility function to make plots look like NCL plots
+    """
+    import matplotlib.ticker as tic
+
+    ax.tick_params(labelsize="small")
+    ax.minorticks_on()
+    ax.xaxis.set_minor_locator(tic.AutoMinorLocator(n=3))
+    ax.yaxis.set_minor_locator(tic.AutoMinorLocator(n=3))
+
+    # length and width are in points and may need to change depending on figure size etc.
+    ax.tick_params(
+        "both",
+        length=8,
+        width=1.5,
+        which="major",
+        bottom=True,
+        top=True,
+        left=True,
+        right=True,
+    )
+    ax.tick_params(
+        "both",
+        length=5,
+        width=0.75,
+        which="minor",
+        bottom=True,
+        top=True,
+        left=True,
+        right=True,
+    )
+
+
+###############################################################################
+# Now we'll make two panels (subplots in matplotlib terminology) using ``plt.subplots``
+# We'll specify ``constrained_layout=True`` which will attempt to automatically
+# layout panels, colorbars and axes decorations nicely.
+# https://matplotlib.org/tutorials/intermediate/constrainedlayout_guide.html
+
+f, ax = plt.subplots(
+    2,  # 2 rows
+    1,  # 1 column
+    constrained_layout=True,  # "magic"
+    subplot_kw={"projection": ccrs.PlateCarree()},  # specify plot projection
+)
+
+# first U, save handles so that we can make a nice colorbar later
+handles = plot_labelled_filled_contours(ds.U, ax=ax[0])
+
+# Now V
+plot_labelled_filled_contours(ds.V, ax=ax[1])
+
+cbar = f.colorbar(
+    handles["filled"],  # make colorbar appropriate for this object
+    ax=ax,  # a list of *two* axes, matplotlib will steal space from both these to fit the colorbar
+    orientation="horizontal",  # horizontal colorbars are on the bottom by default
+    aspect=30,  # aspect ratio of colorbar, just because we can.
+)
+cbar.set_ticks(levels)  # set the tick labels on the colorbar
+
+# make axes look nice and add coastlines
+[nclize_axis(axes) for axes in ax.flat]
+[add_lat_lon_ticklabels(axes) for axes in ax.flat]
+[axes.coastlines(linewidth=0.5) for axes in ax.flat]
+
+# nice figure size in inches
+f.set_size_inches((6, 7))
+
+# a common title
+f.suptitle("A plot with a common label bar (colorbar)")
+
+# show the plot!
+plt.show()