def change_variable_attributes(cfg, ds):
"""
Purpose:
Clean up the variable attributes.
Usage:
Author: PRI
Date: November 2018
"""
# rename existing long_name to description, introduce a
# consistent long_name attribute and introduce the group_name
# attribute
vattr_list = list(cfg["variable_attributes"].keys())
series_list = list(ds.series.keys())
descr = "description_" + ds.globalattributes["nc_level"]
for label in series_list:
variable = pfp_utils.GetVariable(ds, label)
variable["Attr"][descr] = copy.deepcopy(variable["Attr"]["long_name"])
for item in vattr_list:
if label[:len(item)] == item:
for key in list(cfg["variable_attributes"][item].keys()):
variable["Attr"][key] = cfg["variable_attributes"][item][
key]
pfp_utils.CreateVariable(ds, variable)
# parse variable attributes to new format, remove deprecated variable attributes
# and fix valid_range == "-1e+35,1e+35"
tmp = cfg["variable_attributes"]["deprecated"]
deprecated = pfp_cfg.cfg_string_to_list(tmp)
series_list = list(ds.series.keys())
for label in series_list:
variable = pfp_utils.GetVariable(ds, label)
# parse variable attributes to new format
variable["Attr"] = parse_variable_attributes(variable["Attr"])
# remove deprecated variable attributes
for vattr in deprecated:
if vattr in list(variable["Attr"].keys()):
del variable["Attr"][vattr]
# fix valid_range == "-1e+35,1e+35"
if "valid_range" in variable["Attr"]:
valid_range = variable["Attr"]["valid_range"]
if valid_range == "-1e+35,1e+35":
d = numpy.ma.min(variable["Data"])
mn = pfp_utils.round2significant(d, 4, direction='down')
d = numpy.ma.max(variable["Data"])
mx = pfp_utils.round2significant(d, 4, direction='up')
variable["Attr"]["valid_range"] = repr(mn) + "," + repr(mx)
pfp_utils.CreateVariable(ds, variable)
return
def rpLL_plot(pd, ds, output, drivers, target, l6_info, si=0, ei=-1):
""" Plot the results of the Lasslop run. """
iel = l6_info["ERUsingLasslop"]
ieli = l6_info["ERUsingLasslop"]["info"]
ielo = l6_info["ERUsingLasslop"]["outputs"]
# get a local copy of the datetime series
if ei == -1:
dt = ds.series['DateTime']['Data'][si:]
else:
dt = ds.series['DateTime']['Data'][si:ei + 1]
xdt = numpy.array(dt)
#Hdh, f, a = pfp_utils.GetSeriesasMA(ds, 'Hdh', si=si, ei=ei)
Hdh = numpy.array(
[d.hour + (d.minute + d.second / float(60)) / float(60) for d in xdt])
# get the observed and modelled values
obs, f, a = pfp_utils.GetSeriesasMA(ds, target, si=si, ei=ei)
mod, f, a = pfp_utils.GetSeriesasMA(ds, output, si=si, ei=ei)
# make the figure
if iel["gui"]["show_plots"]:
plt.ion()
else:
plt.ioff()
fig = plt.figure(pd["fig_num"], figsize=(13, 8))
fig.clf()
fig.canvas.set_window_title(target + " (LL): " + pd["startdate"] + " to " +
pd["enddate"])
plt.figtext(0.5, 0.95, pd["title"], ha='center', size=16)
# XY plot of the diurnal variation
rect1 = [0.10, pd["margin_bottom"], pd["xy_width"], pd["xy_height"]]
ax1 = plt.axes(rect1)
# get the diurnal stats of the observations
mask = numpy.ma.mask_or(obs.mask, mod.mask)
obs_mor = numpy.ma.array(obs, mask=mask)
dstats = pfp_utils.get_diurnalstats(xdt, obs_mor, ieli)
ax1.plot(dstats["Hr"], dstats["Av"], 'b-', label="Obs")
# get the diurnal stats of all predictions
dstats = pfp_utils.get_diurnalstats(xdt, mod, ieli)
ax1.plot(dstats["Hr"], dstats["Av"], 'r-', label="LL(all)")
mod_mor = numpy.ma.masked_where(numpy.ma.getmaskarray(obs) == True,
mod,
copy=True)
dstats = pfp_utils.get_diurnalstats(xdt, mod_mor, ieli)
ax1.plot(dstats["Hr"], dstats["Av"], 'g-', label="LL(obs)")
plt.xlim(0, 24)
plt.xticks([0, 6, 12, 18, 24])
ax1.set_ylabel(target)
ax1.set_xlabel('Hour')
ax1.legend(loc='upper right', frameon=False, prop={'size': 8})
# XY plot of the 30 minute data
rect2 = [0.40, pd["margin_bottom"], pd["xy_width"], pd["xy_height"]]
ax2 = plt.axes(rect2)
ax2.plot(mod, obs, 'b.')
ax2.set_ylabel(target + '_obs')
ax2.set_xlabel(target + '_LL')
# plot the best fit line
coefs = numpy.ma.polyfit(numpy.ma.copy(mod), numpy.ma.copy(obs), 1)
xfit = numpy.ma.array([numpy.ma.min(mod), numpy.ma.max(mod)])
yfit = numpy.polyval(coefs, xfit)
r = numpy.ma.corrcoef(mod, obs)
ax2.plot(xfit, yfit, 'r--', linewidth=3)
eqnstr = 'y = %.3fx + %.3f, r = %.3f' % (coefs[0], coefs[1], r[0][1])
ax2.text(0.5,
0.875,
eqnstr,
fontsize=8,
horizontalalignment='center',
transform=ax2.transAxes)
# write the fit statistics to the plot
numpoints = numpy.ma.count(obs)
numfilled = numpy.ma.count(mod) - numpy.ma.count(obs)
diff = mod - obs
bias = numpy.ma.average(diff)
ielo[output]["results"]["Bias"].append(bias)
rmse = numpy.ma.sqrt(numpy.ma.mean((obs - mod) * (obs - mod)))
plt.figtext(0.725, 0.225, 'No. points')
plt.figtext(0.825, 0.225, str(numpoints))
ielo[output]["results"]["No. points"].append(numpoints)
plt.figtext(0.725, 0.200, 'No. filled')
plt.figtext(0.825, 0.200, str(numfilled))
plt.figtext(0.725, 0.175, 'Slope')
plt.figtext(0.825, 0.175, str(pfp_utils.round2significant(coefs[0], 4)))
ielo[output]["results"]["m_ols"].append(coefs[0])
plt.figtext(0.725, 0.150, 'Offset')
plt.figtext(0.825, 0.150, str(pfp_utils.round2significant(coefs[1], 4)))
ielo[output]["results"]["b_ols"].append(coefs[1])
plt.figtext(0.725, 0.125, 'r')
plt.figtext(0.825, 0.125, str(pfp_utils.round2significant(r[0][1], 4)))
ielo[output]["results"]["r"].append(r[0][1])
plt.figtext(0.725, 0.100, 'RMSE')
plt.figtext(0.825, 0.100, str(pfp_utils.round2significant(rmse, 4)))
ielo[output]["results"]["RMSE"].append(rmse)
var_obs = numpy.ma.var(obs)
ielo[output]["results"]["Var (obs)"].append(var_obs)
var_mod = numpy.ma.var(mod)
ielo[output]["results"]["Var (LL)"].append(var_mod)
ielo[output]["results"]["Var ratio"].append(var_obs / var_mod)
ielo[output]["results"]["Avg (obs)"].append(numpy.ma.average(obs))
ielo[output]["results"]["Avg (LL)"].append(numpy.ma.average(mod))
# time series of drivers and target
ts_axes = []
rect = [
pd["margin_left"], pd["ts_bottom"], pd["ts_width"], pd["ts_height"]
]
ts_axes.append(plt.axes(rect))
#ts_axes[0].plot(xdt,obs,'b.',xdt,mod,'r-')
ts_axes[0].scatter(xdt, obs, c=Hdh)
ts_axes[0].plot(xdt, mod, 'r-')
plt.axhline(0)
ts_axes[0].set_xlim(xdt[0], xdt[-1])
TextStr = target + '_obs (' + ds.series[target]['Attr']['units'] + ')'
ts_axes[0].text(0.05,
0.85,
TextStr,
color='b',
horizontalalignment='left',
transform=ts_axes[0].transAxes)
TextStr = output + '(' + ds.series[output]['Attr']['units'] + ')'
ts_axes[0].text(0.85,
0.85,
TextStr,
color='r',
horizontalalignment='right',
transform=ts_axes[0].transAxes)
for ThisOne, i in zip(drivers, list(range(1, pd["nDrivers"] + 1))):
this_bottom = pd["ts_bottom"] + i * pd["ts_height"]
rect = [
pd["margin_left"], this_bottom, pd["ts_width"], pd["ts_height"]
]
ts_axes.append(plt.axes(rect, sharex=ts_axes[0]))
data, flag, attr = pfp_utils.GetSeriesasMA(ds, ThisOne, si=si, ei=ei)
data_notgf = numpy.ma.masked_where(flag != 0, data)
data_gf = numpy.ma.masked_where(flag == 0, data)
ts_axes[i].plot(xdt, data_notgf, 'b-')
ts_axes[i].plot(xdt, data_gf, 'r-')
plt.setp(ts_axes[i].get_xticklabels(), visible=False)
TextStr = ThisOne + '(' + ds.series[ThisOne]['Attr']['units'] + ')'
ts_axes[i].text(0.05,
0.85,
TextStr,
color='b',
horizontalalignment='left',
transform=ts_axes[i].transAxes)
# save a hard copy of the plot
sdt = xdt[0].strftime("%Y%m%d")
edt = xdt[-1].strftime("%Y%m%d")
if not os.path.exists(ieli["plot_path"]):
os.makedirs(ieli["plot_path"])
figname = ieli["plot_path"] + pd["site_name"].replace(
" ", "") + "_LL_" + pd["label"]
figname = figname + "_" + sdt + "_" + edt + '.png'
fig.savefig(figname, format='png')
# draw the plot on the screen
if iel["gui"]["show_plots"]:
plt.draw()
pfp_utils.mypause(0.5)
plt.ioff()
else:
plt.close(fig)
plt.ion()