def make_plot(cubes, rings, wcb, theta_level, n):
# Plot a map at verification time
levels = ('air_potential_temperature', [theta_level])
pv = convert.calc('ertel_potential_vorticity', cubes, levels=levels)[0]
plot.pcolormesh(pv, vmin=-2, vmax=2, cmap='coolwarm')
iplt.contour(pv, [0, 2], colors='k', linewidths=2)
add_map()
# Load the trajectories
x = wcb['grid_longitude'] - 360
y = wcb['grid_latitude']
c = wcb['air_potential_temperature']
# Plot the 3d trajectory positions
plt.scatter(x[:, -n],
y[:, -n],
c=c[:, -n],
vmin=300,
vmax=340,
cmap='coolwarm')
# Plot the isentropic boundary
x = rings['grid_longitude'] - 360
y = rings['grid_latitude']
plt.plot(x[:, -n], y[:, -n], color='r', linewidth=3)
return
def bottom_level(tracers, mslp, fig, **kwargs):
# Plot each cube separately
for n, cube in enumerate(tracers):
row = n / ncols
col = n - row * ncols
ax = plt.subplot2grid((nrows, ncols), (row, col))
# Make the plot
im = iplt.pcolormesh(cube[0], **kwargs)
mslp.convert_units('hPa')
cs = iplt.contour(mslp, plevs, colors='k', linewidths=2)
plt.clabel(cs, fmt='%1.0f')
add_map()
plt.title(second_analysis.all_diagnostics[cube.name()].symbol)
# Label the cross section points
if n == 0:
plt.plot([xs, xf], [ys, yf], '-kx')
plt.text(xs, ys, 'A', fontsize=20)
plt.text(xf, yf, 'B', fontsize=20)
# Add letter labels to panels
for n, ax in enumerate(fig.axes):
multilabel(ax, n)
# Add colorbar at bottom of figure
cbar = plt.colorbar(im,
ax=fig.axes,
orientation='horizontal',
fraction=0.05)
cbar.set_label('PVU')
return
def main(cubes, levels, *args, **kwargs):
# Initialise the plot
fig = plt.figure(figsize=(18, 20))
for n, name in enumerate(names):
row = n / ncols
col = n - row * ncols
print(row, col)
ax = plt.subplot2grid((nrows, ncols), (row, col))
cube = convert.calc(name, cubes, levels=levels)[0]
im = iplt.pcolormesh(cube, *args, **kwargs)
add_map()
plt.title(second_analysis.all_diagnostics[name].symbol, fontsize=30)
for n, ax in enumerate(fig.axes):
multilabel(ax, n, fontsize=25)
cbar = plt.colorbar(im, ax=fig.axes, orientation='horizontal',
fraction=0.05, spacing='proportional')
cbar.set_label('PVU')
# cbar.set_ticks(np.array(levels)[::2])
plt.savefig(plotdir + '../../iop8_bl_map.png')
# plt.show()
return
def make_plot(forecast, analysis, variable, levels, units, errlevs, clevs,
cmap='coolwarm', mask=None):
# Extract data and errors
f = convert.calc(variable, forecast, levels=levels)[0]
a = convert.calc(variable, analysis, levels=levels)[0]
err = f - a
for cube in [f, a, err]:
cube.convert_units(units)
if mask is not None:
for cube in [f, a]:
cube.data = np.ma.masked_where(mask, cube.data)
# Plot error
iplt.contourf(err, errlevs, cmap=cmap, extend='both')
add_map()
cbar = plt.colorbar(orientation='horizontal', spacing='proportional')
errlevs.append(0)
cbar.set_ticks(errlevs)
cbar.set_label(units)
# Overlay forecast and analysis
cs = iplt.contour(f, clevs, colors='k', linewidths=2)
iplt.contour(a, clevs, colors='k', linewidths=2, linestyles='--')
plt.clabel(cs, fmt='%1.0f', colors='k')
return
def pv(forecast, analysis, variable, levels, **kwargs):
pv_f = convert.calc(variable, forecast, levels=levels)[0]
pv_a = convert.calc(variable, analysis, levels=levels)[0]
err = pv_f - pv_a
axes = []
# Initialise the plot
fig = plt.figure(figsize=(18, 20))
axes.append(plt.subplot2grid((25, 4), (0, 0), colspan=2, rowspan=10))
im = make_pv_plot(pv_f, **kwargs)
plt.title('Forecast')
axes.append(plt.subplot2grid((25, 4), (0, 2), colspan=2, rowspan=10))
im = make_pv_plot(pv_a, **kwargs)
plt.title('Analysis')
ax = plt.subplot2grid((25, 4), (10, 1), colspan=2, rowspan=1)
cbar = plt.colorbar(im, cax=ax, orientation='horizontal')
axes.append(plt.subplot2grid((25, 4), (13, 1), colspan=2, rowspan=10))
im = iplt.pcolormesh(err, vmin=-5, vmax=5, cmap='coolwarm')
iplt.contour(pv_f, [2], colors='k', linewidths=2)
iplt.contour(pv_a, [2], colors='k', linewidths=2, linestyles='--')
add_map()
plt.title('Forecast Minus Analysis')
ax = plt.subplot2grid((25, 4), (23, 1), colspan=2, rowspan=1)
cbar = plt.colorbar(im, cax=ax, orientation='horizontal')
for n, ax in enumerate(axes):
ax.set_title(ascii_lowercase[n].ljust(20))
plt.show()
def make_plot(cubes, name, mass, mask, **kwargs):
pv = convert.calc(name, cubes)
pv.data = np.ma.masked_where(mask, pv.data)
z = grid.extract_dim_coord(pv, 'z')
pv_ave = pv.collapsed(z.name(), MEAN, weights=mass.data)
im = iplt.pcolormesh(pv_ave, **kwargs)
add_map()
return im
def main(cubes):
pv = convert.calc('ertel_potential_vorticity', cubes, levels=levels)[0]
adv = convert.calc('advection_only_pv', cubes, levels=levels)[0]
for n, name in enumerate([
'sum_of_physics_pv_tracers', 'epsilon',
'dynamics_tracer_inconsistency', 'residual_pv'
]):
cube = convert.calc(name, cubes, levels=levels)[0]
m = n / 2
ax = plt.subplot2grid((2, 2), (n - 2 * m, m))
iplt.contourf(cube, clevs, cmap=cmap)
add_map()
iplt.contour(pv, [2], colors='k', linestyles='-')
iplt.contour(adv, [2], colors='k', linestyles='-')
plt.show()
def main(cubes, levels, *args, **kwargs):
# Initialise the plot
fig = plt.figure(figsize=(18, 20))
pv = convert.calc('ertel_potential_vorticity', cubes, levels=levels)[0]
theta = convert.calc('equivalent_potential_temperature', cubes,
levels=levels)[0][slices]
rh = convert.calc('relative_humidity', cubes, levels=levels)[0][slices]
lon, lat = grid.true_coords(theta)
lonmask = np.logical_or(lon < -15, lon > 5)
latmask = np.logical_or(lat < 47.5, lat > 62.5)
areamask = np.logical_or(lonmask, latmask)
mask = theta.copy(data=areamask.astype(int))
for n, name in enumerate(names):
row = n / ncols
col = n - row * ncols
print(row, col)
ax = plt.subplot2grid((nrows, ncols), (row, col))
cube = convert.calc(name, cubes, levels=levels)[0] # [slices]
im = iplt.contourf(cube, *args, **kwargs)
#iplt.contour(pv, [2], colors='k', linewidths=2)
iplt.contour(mask, [0.5], colors='k', linestyles='--')
add_map()
plt.title(second_analysis.all_diagnostics[name].symbol)
iplt.contour(theta, [300], colors='k', linewidths=2)
iplt.contour(rh, [0.8], colors='w', linewidths=2)
iplt.contourf(rh, [0.8, 2], colors='None', hatches=['.'])
for n, ax in enumerate(fig.axes):
multilabel(ax, n)
cbar = plt.colorbar(im, ax=fig.axes, orientation='horizontal',
fraction=0.05, spacing='proportional')
cbar.set_label('PVU')
cbar.set_ticks(np.linspace(-2, 2, 17)[::2])
plt.savefig(plotdir + 'iop5_pv_tracers_24h_' +
str(levels[1][0])[0:3] + 'hpa.pdf')
# plt.show()
return
def plot_overview(mslp, pv):
iplt.contourf(pv,
even_cscale(2),
cmap='coolwarm',
spacing='proportional',
extend='both')
add_map()
cbar = plt.colorbar(orientation='horizontal',
fraction=0.05,
spacing='proportional')
cbar.set_label('PVU')
cbar.set_ticks(np.linspace(-2, 2, 9))
cs = iplt.contour(mslp, plevs, colors='k', linewidths=1)
plt.clabel(cs, fmt='%1.0f', colors='k')
plt.plot(track[:, 0], track[:, 1], '-y')
plt.plot(track[time, 0], track[time, 1], 'yx', markersize=15)
return
def main(forecast, name, levels, *args, **kwargs):
nt = len(forecast)
rows = (nt / columns) + 1
fig = plt.figure(figsize=(18, 10 * float(rows) / columns))
for n, cubes in enumerate(forecast):
row = n / columns
column = n - row * columns
print(row, column)
ax = plt.subplot2grid((rows, columns), (row, column))
cube = convert.calc(name, cubes, levels=levels)[0]
im = iplt.pcolormesh(cube, *args, **kwargs)
add_map()
ax = plt.subplot2grid((rows, columns), (row, column + 1))
cbar = plt.colorbar(im, cax=ax, orientation='horizontal')
plt.savefig(plotdir + name + '_' + str(levels[0]) + '_' +
str(levels[1][0]) + '.png')
return
def bl_categories():
# Initialise the plot
fig = plt.figure(figsize=(18, 10))
for n, forecast in enumerate(forecasts):
cubes = forecast.set_lead_time(hours=24)
ax = plt.subplot2grid((1, 2), (0, n))
if n == 0:
plt.title('IOP5')
elif n == 1:
plt.title('IOP8')
multilabel(ax, n + 2)
bl_type = convert.calc('boundary_layer_type', cubes)
cmap = mpl.colors.ListedColormap([t[1] for t in types])
iplt.pcolormesh(bl_type, cmap=cmap)
overlay_pressure(cubes)
add_map()
# Add category map legend
handles = []
for bl_type, colour in types:
handles.append(mpatches.Patch(color=colour, label=bl_type))
ax = fig.axes[1]
ax.legend(handles=handles,
ncol=2,
loc='upper_right',
bbox_to_anchor=(0.95, -0.1))
fig.subplots_adjust(bottom=0.5)
plt.savefig(plotdir + 'bl_categories')
return
def wind_speed(cubes, levs, cubes_a=None, **kwargs):
u = convert.calc('x_wind', cubes, levels=levels)[0]
v = convert.calc('y_wind', cubes, levels=levels)[0]
w = convert.calc('upward_air_velocity', cubes, levels=levels)[0]
wind = (u.data**2 + v.data**2 + w.data**2)**0.5
wind = u.copy(data=wind)
if cubes_a is not None:
u_a = convert.calc('x_wind', cubes_a, levels=levels)[0]
v_a = convert.calc('y_wind', cubes_a, levels=levels)[0]
w_a = convert.calc('upward_air_velocity', cubes_a, levels=levels)[0]
wind_a = (u_a.data**2 + v_a.data**2 + w_a.data**2)**0.5
wind.data -= wind_a
u -= u_a
v -= v_a
w -= w_a
print(wind.data.max(), wind.data.min())
im = iplt.contourf(wind, levs, **kwargs)
add_map()
plot.overlay_winds(u, v, 33, 57, scale=1500)
return im
def bl_heights(**kwargs):
# Initialise the plot
fig = plt.figure(figsize=(18, 10))
for n, forecast in enumerate(forecasts):
cubes = forecast.set_lead_time(hours=24)
ax = plt.subplot2grid((1, 2), (0, n))
z_bl = convert.calc('atmosphere_boundary_layer_thickness', cubes)
z_bl.convert_units('km')
im = iplt.pcolormesh(z_bl, **kwargs)
cs = overlay_pressure(cubes)
plt.clabel(cs, fmt='%1.0f')
add_map()
if n == 0:
plt.title('IOP5')
elif n == 1:
plt.title('IOP8')
multilabel(ax, n)
# Add BL Height colourscale
cbar = plt.colorbar(im,
ax=fig.axes,
fraction=0.05,
orientation='horizontal',
extend='max')
cbar.set_label('Boundary layer thickness (km)')
cbar.set_ticks(np.linspace(0, 3, 7))
plt.savefig(plotdir + 'bl_heights')
return
def plot_overview(fp, rp, **kwargs):
plt.figure(figsize=(12, 9))
row_title = 2
row_fig = 8
col_fig = 8
row_cbar = 3
col_cbar = 6
nrows = row_title + 2 * (row_fig + row_cbar)
ncols = 2 * col_fig
row_fig1 = row_title
row_cbar1 = row_fig1 + row_fig
row_fig2 = row_cbar1 + row_cbar
row_cbar2 = row_fig2 + row_fig
col_fig1 = 0
col_fig2 = col_fig1 + col_fig
col_fig3 = col_fig1 + int(col_fig / 2)
col_cbars = int((ncols - col_fig1 - col_cbar) / 2)
plt.subplot2grid((nrows, ncols), (row_fig1, col_fig1),
rowspan=row_fig,
colspan=col_fig)
im = iplt.pcolormesh(fp, **kwargs)
plt.title('Full Precision')
add_map()
plt.subplot2grid((nrows, ncols), (row_fig1, col_fig2),
rowspan=row_fig,
colspan=col_fig)
im = iplt.pcolormesh(rp, **kwargs)
plt.title('Reduced Precision')
add_map()
ax = plt.subplot2grid((nrows, ncols), (row_cbar1, col_cbars),
rowspan=row_cbar,
colspan=col_cbar)
cbar = plt.colorbar(im, cax=ax, orientation='horizontal')
plt.subplot2grid((nrows, ncols), (row_fig2, col_fig3),
rowspan=row_fig,
colspan=col_fig)
vmax = 0.1 * max(abs(kwargs['vmin']), abs(kwargs['vmax']))
im = iplt.pcolormesh(rp - fp, vmin=-vmax, vmax=vmax, cmap='bwr')
plt.title('Difference')
add_map()
ax = plt.subplot2grid((nrows, ncols), (row_cbar2, col_cbars),
rowspan=row_cbar,
colspan=col_cbar)
cbar = plt.colorbar(im, cax=ax, orientation='horizontal')
return
def make_plots(forecast, analysis, variable, levels, units, errlevs, clevs,
cmap1='plasma', cmap='coolwarm', mask=None):
# Extract data and errors
f = convert.calc(variable, forecast, levels=levels)[0]
a = convert.calc(variable, analysis, levels=levels)[0]
err = f - a
for cube in [f, a, err]:
cube.convert_units(units)
if mask is not None:
for cube in [f, a]:
cube.data = np.ma.masked_where(mask, cube.data)
# Initialise the plot
plt.figure(figsize=(18, 15))
# Plot absolute values
plt.subplot2grid((25, 4), (0, 0), colspan=2, rowspan=10)
im = iplt.contourf(f, clevs, extend='both', cmap=cmap1)
#iplt.contour(f, [2], colors='k')
plt.title('(a)'.ljust(28) + 'Forecast'.ljust(35))
add_map()
plt.subplot2grid((25, 4), (0, 2), colspan=2, rowspan=10)
im = iplt.contourf(a, clevs, extend='both', cmap=cmap1)
#iplt.contour(a, [2], colors='k', linestyles='--')
plt.title('(b)'.ljust(28) + 'Analysis'.ljust(35))
add_map()
ax = plt.subplot2grid((25, 4), (10, 1), colspan=2, rowspan=1)
cbar = plt.colorbar(im, cax=ax, orientation='horizontal')
cbar.set_label(units)
# cbar.set_ticks(range(11))
# Plot error
plt.subplot2grid((25, 4), (13, 1), colspan=2, rowspan=10)
im = iplt.contourf(err, errlevs, cmap=cmap, extend='both')
#iplt.contour(f, [2], colors='k')
#iplt.contour(a, [2], colors='k', linestyles='--')
plt.title('(c)'.ljust(20) + 'Forecast Minus Analysis.'.ljust(38))
add_map()
ax = plt.subplot2grid((25, 4), (23, 1), colspan=2, rowspan=1)
cbar = plt.colorbar(im, cax=ax, orientation='horizontal',
spacing='proportional')
errlevs.append(0)
cbar.set_ticks(errlevs)
cbar.set_label(units)
return
def main(cubes, **kwargs):
# Pre-calculate parameters on every plot
mslp = convert.calc('air_pressure_at_sea_level', cubes)
mslp.convert_units('hPa')
theta = convert.calc('equivalent_potential_temperature',
cubes,
levels=levels)[0]
rh = convert.calc('relative_humidity', cubes)
fig = plt.figure(figsize=(18, 15))
lon, lat = grid.true_coords(theta)
lonmask = np.logical_or(lon < -15, lon > 5)
latmask = np.logical_or(lat < 47.5, lat > 62.5)
areamask = np.logical_or(lonmask, latmask)
# Plot overview
ax1 = plt.subplot2grid((2, 2), (0, 0))
iplt.contourf(theta, theta_levs, cmap='coolwarm', extend='both')
add_map()
cs = iplt.contour(mslp, plevs, colors='k', linewidths=2)
plt.clabel(cs, fmt='%1.0f')
mask = theta.copy(data=areamask.astype(int))
iplt.contour(mask, [0.5], colors='k', linestyles='--')
count = 0
for n, (xs, xf, ys, yf) in enumerate(points, start=1):
# Label the cross section points
plt.plot([xs, xf], [ys, yf], '-kx')
plt.text(xs, ys, ascii_uppercase[count], color='k', fontsize=20)
count += 1
plt.text(xf, yf, ascii_uppercase[count], color='k', fontsize=20)
count += 1
multilabel(plt.gca(), 0)
theta = convert.calc('equivalent_potential_temperature', cubes)
# Plot cross sections
titles = ['AB', 'CD', 'EF']
coords = ['grid_longitude', 'altitude']
for n, (xs, xf, ys, yf) in enumerate(points, start=1):
row = n / ncols
col = n - row * ncols
ax = plt.subplot2grid((nrows, ncols), (row, col))
theta_cs = cs_cube(theta, xs, xf, ys, yf)
rh_cs = cs_cube(rh, xs, xf, ys, yf)
im = iplt.contourf(theta_cs,
theta_levs,
coords=coords,
cmap='coolwarm',
extend='both')
iplt.contour(rh_cs, rh_levs, coords=coords, colors='w')
iplt.contourf(rh_cs, [0.8, 2],
coords=coords,
colors='None',
hatches=['.'])
ax.set_ylim(0, 7)
if xs > xf:
ax.invert_xaxis()
multilabel(ax, n, xreversed=True)
else:
multilabel(ax, n)
ax.set_title(titles[n - 1])
ax.set_ylabel('Height (km)')
ax.set_xlabel('Grid Longitude')
# Add colorbar at bottom of figure
cbar = plt.colorbar(im,
ax=fig.axes,
orientation='horizontal',
fraction=0.05,
spacing='proportional')
cbar.set_label('PVU')
fig.savefig(filename)
plt.show()
def make_pv_plot(pv, **kwargs):
im = iplt.pcolormesh(pv, **kwargs)
iplt.contour(pv, [2], colors='k', linewidths=2)
add_map()
return im
