def time_series_plot(df, extra_plots=[], column_name='hmix', ysc=1,
do_flight_data=True, ax=None,
draw_legend=True):
"""
plot dfsummary (above), based on the function in draw_figures.py
"""
#res = np.diff(df.index.to_pydatetime())[0].total_seconds()
# make a custom legend handler to get the error range into the legend
# ref: http://matplotlib.org/users/legend_guide.html#proxy-legend-handles
class HardWiredHandler(object):
def legend_artist(self, legend, orig_handle, fontsize, handlebox):
x0, y0 = handlebox.xdescent, handlebox.ydescent
width, height = handlebox.width, handlebox.height
patch = mpl.patches.Rectangle([x0, y0], width, height,
facecolor='k', alpha=0.2,
edgecolor='none',
transform=handlebox.get_transform())
line = mpl.lines.Line2D([x0, x0+width],
[y0+height/2.0, y0+height/2.0],
color='k',solid_capstyle='butt',
linewidth=1.5,
transform=handlebox.get_transform())
handlebox.add_artist(patch)
handlebox.add_artist(line)
return patch
if ax is None:
fig, ax = figure_template('acp-1')
else:
fig = ax.figure
# draw a plot with pandas to configure ax using pandas' date formatting
# df.lld.plot(ax=ax)
# del ax.lines[0]
#ysc = 1.0/res
t = df.index
low = df[column_name+'_p16'].values #_p16 or _p10
#low[low<10] = 10 # prevent zeros
# print(low.min())
high = df[column_name+'_p84'].values #_p84 or _p90
r1 = ax.fill_between(t, low*ysc, high*ysc, alpha=0.2, color='k', linewidth=0)
# for debugging
# ax.plot(t, low*ysc, color='k', alpha=0.2)
# ax.plot(t, high*ysc, color='k', alpha=0.2)
r2 = ax.plot(t, df[column_name+'_mean'].values*ysc, 'k')
leg_objs = []
leg_entries = []
leg_objs.append((r1,r2))
leg_entries.append('Box model with BMC deconvolution')
ret = ax.plot(t, df[column_name+'_scaled']*ysc)[0]
leg_objs.append(ret)
leg_entries.append('Box model with raw detector output')
ret = ax.plot(t, df[column_name+'_scaled_shifted']*ysc)[0]
leg_objs.append(ret)
leg_entries.append('Box model with lagged detector output')
if do_flight_data:
#x,y = 2011-11-06 07:56:59.650000 671.056073212
# (from 2014-10-goulburn-sbl-fluxes/extract_and_plot_soundings.py)
x = datetime.datetime(2011,11,6,8,0)
y = np.array([[583,671], [206,651]])
x = datetime.datetime(2011,11,6,8,0), datetime.datetime(2011,11,6,7,25)
ret = ax.errorbar(x, y.mean(axis=1), yerr=np.diff(y, axis=1).ravel()/2.0, fmt='none')
leg_objs.append(ret)
leg_entries.append('Aircraft measurements')
for xdata, ydata, label in extra_plots:
ret = ax.plot(xdata, ydata)[0]
leg_objs.append(ret[0])
leg_entries.append(label)
# make the legend
if draw_legend:
ax.legend(leg_objs, leg_entries,
handler_map={tuple: HardWiredHandler()},
loc='upper left',
frameon=False)
# format x-axis
#from micromet.plot import day_ticks
#day_ticks(ax) # TODO: switch to pandas convention
ax.xaxis.set_major_formatter(mpl.dates.DateFormatter('%H:%M'))
fancy_ylabel(ax, 'Boundary layer depth (m)')
return fig, ax
def newfig():
fig, ax = figure_template('acp-1')
return fig, ax
elif pos == 'right':
ann = ax.annotate(s,xy=(1,1), xycoords='axes fraction',
textcoords='offset points', xytext=(28+hoffset, 10),
fontsize=textsize,
horizontalalignment='right')
return ann
xvar = 'relhum'
yvar = 'a_to_c'
data = []
positions = []
for day_number, dfchain in chains_by_day:
dfss = peaklist[day_number-1]
data.append(dfchain[k].values)
positions.append(dfss[xvar][:60].mean())
fig, ax = figure_template('acp-1')
ax.boxplot(data, positions=positions,
showfliers=False, showcaps=False,
boxprops=dict(color="#0072B2"),
medianprops=dict(color="#0072B2"))
ax.xaxis.set_major_locator(mpl.ticker.AutoLocator())
ax.xaxis.set_major_formatter(mpl.ticker.ScalarFormatter())
ax.set_xlabel('Relative humidity (%)')
fancy_ylabel(ax, 'Polonium-218 counts / Polonium-214 counts')
savefig(fig, 'bayes-fit-to-peaks-relh-versus-po-count-ratio-boxplot', fdir='./figures/')
# error bar plot version
yplt = [np.mean(itm) for itm in data]
yerr = [np.std(itm) for itm in data]
fig, ax = figure_template('acp-1')
