def detrend(var, ax=None, lcopy=True, ldetrend=True, ltrend=False, degree=1, rcond=None, w=None,
lsmooth=False, lresidual=False, window_len=11, window='hanning'):
''' subtract a linear trend from a time-series array (operation is in-place) '''
# check input
if not isinstance(var,np.ndarray): raise NotImplementedError # too many checks
if lcopy: var = var.copy() # make copy - not in-place!
# fit over entire array (usually not what we want...)
if ax is None and ldetrend: ax = np.arange(var.size) # make dummy axis, if necessary
if var.ndim != 1:
shape = var.shape
var = var.ravel() # flatten array, if necessary
else: shape = None
# apply optional detrending
if ldetrend or ltrend:
# fit linear trend
trend = np.polyfit(ax, var, deg=degree, rcond=rcond, w=w, full=False, cov=False)
# evaluate and subtract linear trend
if ldetrend and ltrend: raise ArgumentError("Can either return trend/polyfit or residuals, not both.")
elif ldetrend and not ltrend: var -= np.polyval(trend, ax) # residuals
elif ltrend and not ldetrend: var = np.polyval(trend, ax) # residuals
# apply optional smoothing
if lsmooth and lresidual: raise ArgumentError("Can either return smoothed array or residuals, not both.")
elif lsmooth: var = smooth(var, window_len=window_len, window=window)
elif lresidual: var -= smooth(var, window_len=window_len, window=window)
# return detrended and/or smoothed time-series
if shape is not None: var = var.reshape(shape)
return var
def detrend(var,
ax=None,
lcopy=True,
ldetrend=True,
ltrend=False,
degree=1,
rcond=None,
w=None,
lsmooth=False,
lresidual=False,
window_len=11,
window='hanning'):
''' subtract a linear trend from a time-series array (operation is in-place) '''
# check input
if not isinstance(var, np.ndarray):
raise NotImplementedError # too many checks
if lcopy: var = var.copy() # make copy - not in-place!
# fit over entire array (usually not what we want...)
if ax is None and ldetrend:
ax = np.arange(var.size) # make dummy axis, if necessary
if var.ndim != 1:
shape = var.shape
var = var.ravel() # flatten array, if necessary
else:
shape = None
# apply optional detrending
if ldetrend or ltrend:
# fit linear trend
trend = np.polyfit(ax,
var,
deg=degree,
rcond=rcond,
w=w,
full=False,
cov=False)
# evaluate and subtract linear trend
if ldetrend and ltrend:
raise ArgumentError(
"Can either return trend/polyfit or residuals, not both.")
elif ldetrend and not ltrend:
var -= np.polyval(trend, ax) # residuals
elif ltrend and not ldetrend:
var = np.polyval(trend, ax) # residuals
# apply optional smoothing
if lsmooth and lresidual:
raise ArgumentError(
"Can either return smoothed array or residuals, not both.")
elif lsmooth:
var = smooth(var, window_len=window_len, window=window)
elif lresidual:
var -= smooth(var, window_len=window_len, window=window)
# return detrended and/or smoothed time-series
if shape is not None: var = var.reshape(shape)
return var
def getPlotValues(var, checkunits=None, checkname=None, lsmooth=False, lperi=False, laxis=False):
''' Helper function to check variable/axis, get (scaled) values for plot, and return appropriate units. '''
# figure out units
if var.plot is not None:
varname = var.plot.name
if checkname is not None and varname != checkname: # only check plotname!
raise VariableError, "Expected variable name '{}', found '{}'.".format(checkname,varname)
else: varname = var.atts['name']
val = var.getArray(unmask=True, copy=True) # the data to plot
if var.plot is not None:
if var.units != var.plot.units:
val *= var.plot.scalefactor
val += var.plot.offset
varunits = var.plot.units
else:
varunits = var.atts['units']
if checkunits is not None and varunits != checkunits:
raise VariableError, "Units for variable '{}': expected {}, found {}.".format(var.name,checkunits,varunits)
# some post-processing
val = val.squeeze()
if lsmooth: val = smooth(val)
if lperi:
if laxis:
delta = np.diff(val)
val = np.concatenate((val[:1]-delta[:1],val,val[-1:]+delta[-1:]))
else: val = np.concatenate((val[-1:],val,val[:1]))
# return values, units, name
return val, varunits, varname
def getPlotValues(var, checkunits=None, checkname=None, lsmooth=False, lperi=False, laxis=False):
''' Helper function to check variable/axis, get (scaled) values for plot, and return appropriate units. '''
# figure out units
if var.plot is not None:
varname = var.plot.name
if checkname is not None and varname != checkname: # only check plotname!
raise VariableError, "Expected variable name '{}', found '{}'.".format(checkname,varname)
else: varname = var.atts['name']
if np.issubdtype(var.dtype, np.datetime64): val = var.data_array.copy() # need to preserve dates
else: val = var.getArray(unmask=True, fillValue=np.NaN, dtype=np.float, copy=True) # the data to plot
# N.B.: matplotlib does not understand masked arrays, therefor we have to convert masked values to NaN's
# (and convert the data to float in the process...)
if var.plot is not None:
if var.units != var.plot.units:
val *= var.plot.scalefactor
val += var.plot.offset
varunits = var.plot.units
else:
varunits = var.atts['units']
if checkunits is not None and varunits != checkunits:
raise VariableError, "Units for variable '{}': expected {}, found {}.".format(var.name,checkunits,varunits)
# some post-processing
if val.size > 1: val = val.squeeze()
if lsmooth: val = smooth(val)
if lperi:
if laxis:
delta = np.diff(val)
val = np.concatenate((val[:1]-delta[:1],val,val[-1:]+delta[-1:]))
else: val = np.concatenate((val[-1:],val,val[:1]))
# return values, units, name
return val, varunits, varname
def getPlotValues(var,
checkunits=None,
checkname=None,
lsmooth=False,
lperi=False,
pseudo_axis=None,
laxis=False):
''' Helper function to check variable/axis, get (scaled) values for plot, and return appropriate units. '''
# figure out units
if var.plot is not None:
varname = var.plot.name
if checkname is not None and varname != checkname: # only check plotname!
raise VariableError(
"Expected variable name '{}', found '{}'.".format(
checkname, varname))
else:
varname = var.atts['name']
if np.issubdtype(var.dtype, np.datetime64):
val = var.data_array.copy() # need to preserve dates
else:
val = var.getArray(unmask=True,
fillValue=np.NaN,
dtype=np.float,
copy=True) # the data to plot
# N.B.: matplotlib does not understand masked arrays, therefor we have to convert masked values to NaN's
# (and convert the data to float in the process...)
if var.plot is not None:
if var.units != var.plot.units:
val *= var.plot.scalefactor
val += var.plot.offset
varunits = var.plot.units
else:
varunits = var.atts['units']
if checkunits is not None and varunits != checkunits:
raise VariableError(
"Units for variable '{}': expected {}, found {}.".format(
var.name, checkunits, varunits))
# some post-processing
if val.size > 1: val = val.squeeze()
if lsmooth: val = smooth(val)
if lperi:
if laxis:
delta = np.diff(val)
val = np.concatenate(
(val[:1] - delta[:1], val, val[-1:] + delta[-1:]))
else:
val = np.concatenate((val[-1:], val, val[:1]))
# return values, units, name
return val, varunits, varname
def getPlotValues(var,
checkunits=None,
checkname=None,
lsmooth=False,
lperi=False,
laxis=False):
''' Helper function to check variable/axis, get (scaled) values for plot, and return appropriate units. '''
# figure out units
if var.plot is not None:
varname = var.plot.name
if checkname is not None and varname != checkname: # only check plotname!
raise VariableError, "Expected variable name '{}', found '{}'.".format(
checkname, varname)
else:
varname = var.atts['name']
val = var.getArray(unmask=True, copy=True) # the data to plot
if var.plot is not None:
if var.units != var.plot.units:
val *= var.plot.scalefactor
val += var.plot.offset
varunits = var.plot.units
else:
varunits = var.atts['units']
if checkunits is not None and varunits != checkunits:
raise VariableError, "Units for variable '{}': expected {}, found {}.".format(
var.name, checkunits, varunits)
# some post-processing
val = val.squeeze()
if lsmooth: val = smooth(val)
if lperi:
if laxis:
delta = np.diff(val)
val = np.concatenate(
(val[:1] - delta[:1], val, val[-1:] + delta[-1:]))
else:
val = np.concatenate((val[-1:], val, val[:1]))
# return values, units, name
return val, varunits, varname
def linePlot(varlist,
ax=None,
fig=None,
linestyles=None,
varatts=None,
legend=None,
xline=None,
yline=None,
title=None,
flipxy=None,
xlabel=None,
ylabel=None,
xlim=None,
ylim=None,
lsmooth=False,
lprint=False,
**kwargs):
''' A function to draw a list of 1D variables into an axes, and annotate the plot based on variable properties. '''
warn('Deprecated function: use Figure or Axes class methods.')
# create axes, if necessary
if ax is None:
if fig is None: fig, ax = getFigAx(1) # single panel
else: ax = fig.axes[0]
# varlist is the list of variable objects that are to be plotted
#print varlist
if isinstance(varlist, Variable): varlist = [varlist]
elif not isinstance(varlist, (tuple, list)) or not all(
[isinstance(var, Variable) for var in varlist]):
raise TypeError
for var in varlist:
var.squeeze() # remove singleton dimensions
# linestyles is just a list of line styles for each plot
if isinstance(linestyles, (basestring, NoneType)):
linestyles = [linestyles] * len(varlist)
elif not isinstance(linestyles, (tuple, list)):
if not all([isinstance(linestyles, basestring) for var in varlist]):
raise TypeError
if len(varlist) != len(linestyles):
raise ListError, "Failed to match linestyles to varlist!"
# varatts are variable-specific attributes that are parsed for special keywords and then passed on to the
if varatts is None: varatts = [dict()] * len(varlist)
elif isinstance(varatts, dict):
tmp = [
varatts[var.name] if var.name in varatts else dict()
for var in varlist
]
if any(tmp): varatts = tmp # if any variable names were found
else:
varatts = [varatts] * len(
varlist
) # assume it is one varatts dict, which will be used for all variables
elif not isinstance(varatts, (tuple, list)):
raise TypeError
if not all([isinstance(atts, dict) for atts in varatts]): raise TypeError
# check axis: they need to have only one axes, which has to be the same for all!
if len(varatts) != len(varlist):
raise ListError, "Failed to match varatts to varlist!"
for var in varlist:
if var.ndim > 1:
raise AxisError, "Variable '{}' has more than one dimension; consider squeezing.".format(
var.name)
elif var.ndim == 0:
raise AxisError, "Variable '{}' is a scalar; consider display as a line.".format(
var.name)
# loop over variables
plts = []
varname = None
varunits = None
axname = None
axunits = None # list of plot handles
for var, linestyle, varatt in zip(varlist, linestyles, varatts):
varax = var.axes[0]
# scale axis and variable values
axe, axunits, axname = getPlotValues(varax,
checkunits=axunits,
checkname=None)
val, varunits, varname = getPlotValues(var,
checkunits=varunits,
checkname=None)
# variable and axis scaling is not always independent...
if var.plot is not None and varax.plot is not None:
if 'preserve' in var.plot and 'scalefactor' in varax.plot:
if varax.units != axunits and var.plot.preserve == 'area':
val /= varax.plot.scalefactor
# figure out keyword options
kwatts = kwargs.copy()
kwatts.update(varatt) # join individual and common attributes
if 'label' not in kwatts:
kwatts['label'] = var.name # default label: variable name
# N.B.: other scaling behavior could be added here
if lprint: print varname, varunits, val.mean()
if lsmooth: val = smooth(val)
# figure out orientation
if flipxy: xx, yy = val, axe
else: xx, yy = axe, val
# call plot function
if linestyle is None: plts.append(ax.plot(xx, yy, **kwatts)[0])
else: plts.append(ax.plot(xx, yy, linestyle, **kwatts)[0])
# set axes limits
if isinstance(xlim, (list, tuple)) and len(xlim) == 2: ax.set_xlim(*xlim)
elif xlim is not None: raise TypeError
if isinstance(ylim, (list, tuple)) and len(ylim) == 2: ax.set_ylim(*ylim)
elif ylim is not None: raise TypeError
# set title
if title is not None:
ax.set_title(title, dict(fontsize='medium'))
pos = ax.get_position()
pos = pos.from_bounds(x0=pos.x0,
y0=pos.y0,
width=pos.width,
height=pos.height - 0.03)
ax.set_position(pos)
# set axes labels
if flipxy:
xname, xunits, yname, yunits = varname, varunits, axname, axunits
else:
xname, xunits, yname, yunits = axname, axunits, varname, varunits
if not xlabel:
xlabel = '{0:s} [{1:s}]'.format(
xname, xunits) if xunits else '{0:s}'.format(xname)
else:
xlabel = xlabel.format(xname, xunits)
if not ylabel:
ylabel = '{0:s} [{1:s}]'.format(
yname, yunits) if yunits else '{0:s}'.format(yname)
else:
ylabel = ylabel.format(yname, yunits)
# a typical custom label that makes use of the units would look like this: 'custom label [{1:s}]',
# where {} will be replaced by the appropriate default units (which have to be the same anyway)
xpad = 2
xticks = ax.get_xaxis().get_ticklabels()
ypad = -2
yticks = ax.get_yaxis().get_ticklabels()
# len(xticks) > 0 is necessary to avoid errors with AxesGrid, which removes invisible tick labels
if len(xticks) > 0 and xticks[-1].get_visible():
ax.set_xlabel(xlabel, labelpad=xpad)
elif len(yticks) > 0 and not title:
yticks[0].set_visible(False) # avoid overlap
if len(yticks) > 0 and yticks[-1].get_visible():
ax.set_ylabel(ylabel, labelpad=ypad)
elif len(xticks) > 0:
xticks[0].set_visible(False) # avoid overlap
# make monthly ticks
if axname == 'time' and axunits == 'month':
ax.xaxis.set_minor_locator(
mpl.ticker.AutoMinorLocator(2)) # ax.minorticks_on()
# add legend
if legend:
legatts = dict()
if ax.get_yaxis().get_label():
legatts['fontsize'] = ax.get_yaxis().get_label().get_fontsize()
if isinstance(legend, dict): legatts.update(legend)
elif isinstance(legend, (int, np.integer, float, np.inexact)):
legatts['loc'] = legend
ax.legend(**legatts)
# add orientation lines
if isinstance(xline, (int, np.integer, float, np.inexact)):
ax.axhline(y=xline, color='black')
elif isinstance(xline, dict):
ax.axhline(**xline)
if isinstance(yline, (int, np.integer, float, np.inexact)):
ax.axvline(x=yline, color='black')
elif isinstance(xline, dict):
ax.axvline(**yline)
# return handle
return plts
def linePlot(varlist, ax=None, fig=None, linestyles=None, varatts=None, legend=None,
xline=None, yline=None, title=None, flipxy=None, xlabel=None, ylabel=None, xlim=None,
ylim=None, lsmooth=False, lprint=False, **kwargs):
''' A function to draw a list of 1D variables into an axes, and annotate the plot based on variable properties. '''
warn('Deprecated function: use Figure or Axes class methods.')
# create axes, if necessary
if ax is None:
if fig is None: fig,ax = getFigAx(1) # single panel
else: ax = fig.axes[0]
# varlist is the list of variable objects that are to be plotted
#print varlist
if isinstance(varlist,Variable): varlist = [varlist]
elif not isinstance(varlist,(tuple,list)) or not all([isinstance(var,Variable) for var in varlist]): raise TypeError
for var in varlist: var.squeeze() # remove singleton dimensions
# linestyles is just a list of line styles for each plot
if isinstance(linestyles,(basestring,NoneType)): linestyles = [linestyles]*len(varlist)
elif not isinstance(linestyles,(tuple,list)):
if not all([isinstance(linestyles,basestring) for var in varlist]): raise TypeError
if len(varlist) != len(linestyles): raise ListError, "Failed to match linestyles to varlist!"
# varatts are variable-specific attributes that are parsed for special keywords and then passed on to the
if varatts is None: varatts = [dict()]*len(varlist)
elif isinstance(varatts,dict):
tmp = [varatts[var.name] if var.name in varatts else dict() for var in varlist]
if any(tmp): varatts = tmp # if any variable names were found
else: varatts = [varatts]*len(varlist) # assume it is one varatts dict, which will be used for all variables
elif not isinstance(varatts,(tuple,list)): raise TypeError
if not all([isinstance(atts,dict) for atts in varatts]): raise TypeError
# check axis: they need to have only one axes, which has to be the same for all!
if len(varatts) != len(varlist): raise ListError, "Failed to match varatts to varlist!"
for var in varlist:
if var.ndim > 1: raise AxisError, "Variable '{}' has more than one dimension; consider squeezing.".format(var.name)
elif var.ndim == 0: raise AxisError, "Variable '{}' is a scalar; consider display as a line.".format(var.name)
# loop over variables
plts = []; varname = None; varunits = None; axname = None; axunits = None # list of plot handles
for var,linestyle,varatt in zip(varlist,linestyles,varatts):
varax = var.axes[0]
# scale axis and variable values
axe, axunits, axname = getPlotValues(varax, checkunits=axunits, checkname=None)
val, varunits, varname = getPlotValues(var, checkunits=varunits, checkname=None)
# variable and axis scaling is not always independent...
if var.plot is not None and varax.plot is not None:
if 'preserve' in var.plot and 'scalefactor' in varax.plot:
if varax.units != axunits and var.plot.preserve == 'area':
val /= varax.plot.scalefactor
# figure out keyword options
kwatts = kwargs.copy(); kwatts.update(varatt) # join individual and common attributes
if 'label' not in kwatts: kwatts['label'] = var.name # default label: variable name
# N.B.: other scaling behavior could be added here
if lprint: print varname, varunits, val.mean()
if lsmooth: val = smooth(val)
# figure out orientation
if flipxy: xx,yy = val, axe
else: xx,yy = axe, val
# call plot function
if linestyle is None: plts.append(ax.plot(xx, yy, **kwatts)[0])
else: plts.append(ax.plot(xx, yy, linestyle, **kwatts)[0])
# set axes limits
if isinstance(xlim,(list,tuple)) and len(xlim)==2: ax.set_xlim(*xlim)
elif xlim is not None: raise TypeError
if isinstance(ylim,(list,tuple)) and len(ylim)==2: ax.set_ylim(*ylim)
elif ylim is not None: raise TypeError
# set title
if title is not None:
ax.set_title(title, dict(fontsize='medium'))
pos = ax.get_position()
pos = pos.from_bounds(x0=pos.x0, y0=pos.y0, width=pos.width, height=pos.height-0.03)
ax.set_position(pos)
# set axes labels
if flipxy: xname,xunits,yname,yunits = varname,varunits,axname,axunits
else: xname,xunits,yname,yunits = axname,axunits,varname,varunits
if not xlabel: xlabel = '{0:s} [{1:s}]'.format(xname,xunits) if xunits else '{0:s}'.format(xname)
else: xlabel = xlabel.format(xname,xunits)
if not ylabel: ylabel = '{0:s} [{1:s}]'.format(yname,yunits) if yunits else '{0:s}'.format(yname)
else: ylabel = ylabel.format(yname,yunits)
# a typical custom label that makes use of the units would look like this: 'custom label [{1:s}]',
# where {} will be replaced by the appropriate default units (which have to be the same anyway)
xpad = 2; xticks = ax.get_xaxis().get_ticklabels()
ypad = -2; yticks = ax.get_yaxis().get_ticklabels()
# len(xticks) > 0 is necessary to avoid errors with AxesGrid, which removes invisible tick labels
if len(xticks) > 0 and xticks[-1].get_visible(): ax.set_xlabel(xlabel, labelpad=xpad)
elif len(yticks) > 0 and not title: yticks[0].set_visible(False) # avoid overlap
if len(yticks) > 0 and yticks[-1].get_visible(): ax.set_ylabel(ylabel, labelpad=ypad)
elif len(xticks) > 0: xticks[0].set_visible(False) # avoid overlap
# make monthly ticks
if axname == 'time' and axunits == 'month':
ax.xaxis.set_minor_locator(mpl.ticker.AutoMinorLocator(2)) # ax.minorticks_on()
# add legend
if legend:
legatts = dict()
if ax.get_yaxis().get_label():
legatts['fontsize'] = ax.get_yaxis().get_label().get_fontsize()
if isinstance(legend,dict): legatts.update(legend)
elif isinstance(legend,(int,np.integer,float,np.inexact)): legatts['loc'] = legend
ax.legend(**legatts)
# add orientation lines
if isinstance(xline,(int,np.integer,float,np.inexact)): ax.axhline(y=xline, color='black')
elif isinstance(xline,dict): ax.axhline(**xline)
if isinstance(yline,(int,np.integer,float,np.inexact)): ax.axvline(x=yline, color='black')
elif isinstance(xline,dict): ax.axvline(**yline)
# return handle
return plts
if ax is None and ldetrend: ax = np.arange(var.size) # make dummy axis, if necessary
if var.ndim != 1:
shape = var.shape
var = var.ravel() # flatten array, if necessary
else: shape = None
# apply optional detrending
if ldetrend or ltrend:
# fit linear trend
trend = np.polyfit(ax, var, deg=degree, rcond=rcond, w=w, full=False, cov=False)
# evaluate and subtract linear trend
if ldetrend and ltrend: raise ArgumentError, "Can either return trend/polyfit or residuals, not both."
elif ldetrend and not ltrend: var -= np.polyval(trend, ax) # residuals
elif ltrend and not ldetrend: var = np.polyval(trend, ax) # residuals
# apply optional smoothing
if lsmooth and lresidual: raise ArgumentError, "Can either return smoothed array or residuals, not both."
elif lsmooth: var = smooth(var, window_len=window_len, window=window)
elif lresidual: var -= smooth(var, window_len=window_len, window=window)
# return detrended and/or smoothed time-series
if shape is not None: var = var.reshape(shape)
return var
# function to smooth a vector (numpy array): moving mean, nothing fancy
def movingMean(x,i):
''' smooth a vector (x, numpy array) using a moving mean of window width 2*i+1 '''
if x.ndim > 1: raise ValueError
xs = x.copy() # smoothed output vector
i = 2*i
d = i+1 # denominator for later
while i>0:
t = x.copy(); t[i:] = t[:-i]; xs += t
t = x.copy(); t[:-i] = t[i:]; xs += t
if ax is None and ldetrend: ax = np.arange(var.size) # make dummy axis, if necessary
if var.ndim != 1:
shape = var.shape
var = var.ravel() # flatten array, if necessary
else: shape = None
# apply optional detrending
if ldetrend or ltrend:
# fit linear trend
trend = np.polyfit(ax, var, deg=degree, rcond=rcond, w=w, full=False, cov=False)
# evaluate and subtract linear trend
if ldetrend and ltrend: raise ArgumentError, "Can either return trend/polyfit or residuals, not both."
elif ldetrend and not ltrend: var -= np.polyval(trend, ax) # residuals
elif ltrend and not ldetrend: var = np.polyval(trend, ax) # residuals
# apply optional smoothing
if lsmooth and lresidual: raise ArgumentError, "Can either return smoothed array or residuals, not both."
elif lsmooth: var = smooth(var, window_len=window_len, window=window)
elif lresidual: var -= smooth(var, window_len=window_len, window=window)
# return detrended and/or smoothed time-series
if shape is not None: var = var.reshape(shape)
return var
# function to smooth a vector (numpy array): moving mean, nothing fancy
def movingMean(x,i):
''' smooth a vector (x, numpy array) using a moving mean of window width 2*i+1 '''
if x.ndim > 1: raise ValueError
xs = x.copy() # smoothed output vector
i = 2*i
d = i+1 # denominator for later
while i>0:
t = x.copy(); t[i:] = t[:-i]; xs += t
t = x.copy(); t[:-i] = t[i:]; xs += t
