def extract_ndates_around_date(self, date, n):
###################################################################
"""
Extract n values before and n values after a given date
If n values are not available, returns all available values
.data is set to None if no value at all is available
:param date: date in decimal year
:param n: number of observations to be extracted
:return: a new Gts
"""
# import
import inspect
import numpy as np
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
# copy
new_Gts = self.copy(data_xyz=None)
new_Gts.data = None
try:
sel_data_after = np.copy(self.data[np.where((self.data[:, 0] > date))])
except:
print("-- time series ", self.code, " does not have data after ", date)
return (new_Gts)
try:
sel_data_before = np.copy(self.data[np.where(
(self.data[:, 0] < date))])
except:
print("-- time series ", self.code, " does not have data before ",
date)
return (new_Gts)
# extract data
if sel_data_after.shape[0] > n:
sel_data_after = sel_data_after[:n, :]
if sel_data_before.shape[0] > n:
sel_data_before = sel_data_before[-n:, :]
sel_data = np.vstack((sel_data_before, sel_data_after))
new_Gts.data = sel_data
return (new_Gts)
def detrend_annual(self,
method='L2',
in_place=False,
periods=None,
exclude_periods=None):
"""
estimates a trend + annual terms in a time series and removes them
velocity and annual attribute are saved in Gts.velocity & Gts.annual
:param periods : periods used for estimation
:param exclude_periods : periods to be excluded from estimation
:param in_place : if True then replace the current time series
:return : the detrended time series
:note : outliers from Gts.outliers are ommitted in the estimation and
offsets given Gts.offsets_dates are estimated simultaneously
"""
###########################################################################
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
###########################################################################
import copy
outliers = copy.deepcopy(self.outliers)
tmp_ts = self.remove_outliers()
if periods:
tmp_ts = tmp_ts.extract_periods(periods)
if exclude_periods:
tmp_ts = tmp_ts.exclude_periods(periods)
detrended = tmp_ts.make_model(option='detrend_annual', method=method)
vel = detrended.velocity
offsets_values = detrended.offsets_values
annual = detrended.annual
new_gts = self.copy()
new_gts.outliers = outliers
new_gts.offsets_values = offsets_values
new_gts.velocity = vel
new_gts.annual = annual
model = new_gts.mmodel()
new_gts.data[:, 1:4] = new_gts.data[:, 1:4] - model.data[:, 1:4]
if in_place:
self.data = new_gts.data
else:
return (new_gts)
def extract_ndates_before_date(self, date, n, verbose=False):
###################################################################
"""
Extract n values before a given date
If n values are not available, returns all available values before date
.data is set to None if no value at all is available
:param date: date in decimal year
:param n: number of observations to be extracted
:return: a new Gts
"""
# import
import inspect
import numpy as np
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
# copy
new_Gts = self.copy(data_xyz=None)
new_Gts.data = None
# extract data
sel_data = np.copy(self.data[np.where(self.data[:, 0] < date)])
if self.data_xyz is not None:
sel_data_xyz = np.copy(self.data_xyz[np.where(self.data[:, 0] < date)])
# extract data
if sel_data.shape[0] > n:
sel_data = sel_data[-n:, :]
if self.data_xyz is not None:
sel_data_xyz = sel_data_xyz[-n:, :]
elif sel_data.shape[0] > 0:
sel_data = sel_data[:, :]
if self.data_xyz is not None:
sel_data_xyz = sel_data_xyz[:, :]
else:
sel_data = None
if verbose:
print("-- time series ", self.code, " does not have data before ",
date)
new_Gts.data = sel_data
if self.data_xyz is not None:
new_Gts.data_xyz = sel_data_xyz
return (new_Gts)
def remove_pole(self, pole, pole_type='euler', in_place=False, verbose=True):
"""
remove velocity predicted by an Euler pole or a rotation rate vector from a time series
pole is a 1D array with 3 values
requires self.lon & self.lat attributes to have been filled before
if in_place = True then replace the current time series
"""
import numpy as np
from pyacs.gts.Gts import Gts
import inspect
# after this method .data and .data_xyz are not consistent so .data_xyz is set to None
self.data_xyz = None
###########################################################################
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
###########################################################################
if (self.lon is None) or (self.lat is None):
print(
"!!! ERROR: lon & lat needs to be properly filled to use this method."
)
return ()
from pyacs.lib.gmtpoint import GMT_Point
M = GMT_Point(code=self.code,
lon=self.lon,
lat=self.lat,
Ve=0.,
Vn=0.,
SVe=0.,
SVn=0.)
# N=M.substract_pole(pole,pole_type)
N = M.pole(W=pole, SW=None, type_euler='euler', option='predict')
vel_neu = np.array([N.Vn, N.Ve, 0.]) * 1E-3
if verbose: print("-- Removing velocity (NEU)", vel_neu * 1.E3)
new_Gts = self.remove_velocity(vel_neu)
if in_place:
self.data = new_Gts.data.copy()
return (new_Gts)
def rotate(self,angle,in_place=False):
###################################################################
"""
rotates the axis by an angle
:param angle: angle in decimal degrees clockwise
if in_place = True then replace the current time series
"""
# import
import inspect
import numpy as np
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3],__name__,self)
except GtsInputDataNone as error:
# print PYACS WARNING
print( error )
return( self )
# computes the rotates components
angle_radian = np.radians( -angle )
new_data=self.data.copy()
new_x = new_data[:,2] * np.cos( angle_radian ) - new_data[:,1] * np.sin( angle_radian )
new_y = new_data[:,2] * np.sin( angle_radian ) + new_data[:,1] * np.cos( angle_radian )
new_data[:,1] = new_y
new_data[:,2] = new_x
new_data[:,4] = new_data[:,5] = 1.E-3
new_data[:,7:] = 0.0
new_Gts=self.copy(data=new_data)
# .data_xyz set to None
new_Gts.data_xyz = None
if in_place:
self.data=new_Gts.data.copy()
return( self )
else:
return(new_Gts)
def remove_velocity(self,vel_neu,in_place=False):
###################################################################
"""
remove velocity from a time series
vel_neu is a 1D array of any arbitrary length, but with the velocities (NEU) to be removed in the first 3 columns
if in_place = True then replace the current time series
"""
# import
import inspect
import numpy as np
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3],__name__,self)
except GtsInputDataNone as error:
# print PYACS WARNING
print( error )
return( self )
if self.t0 != None:
ref_date=self.t0
else:
ref_date=np.mean(self.data[0,0])
v_neu= np.array( vel_neu[0:3] )
new_data=self.data.copy()
new_data[:,1:4]=new_data[:,1:4]-np.dot( (new_data[:,0]-ref_date).reshape(-1,1) , v_neu.reshape(1,3) )
new_Gts=self.copy(data=new_data)
# .data_xyz set to None
new_Gts.data_xyz = None
vel = vel_neu
new_Gts.velocity = vel
if in_place:
self.data=new_Gts.data.copy()
return( self )
else:
return(new_Gts)
def frame(self, frame=None, in_place=False, verbose=False):
"""
Rotates a time series according to an Euler pole
Returns a new Gts instance
"""
import numpy as np
from pyacs.gts.Gts import Gts
import inspect
# after this method .data and .data_xyz are not consistent so .data_xyz is set to None
self.data_xyz = None
###########################################################################
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
###########################################################################
# Euler poles taken from pygvel_pole_info.py
lEuler = {}
lEuler['soam'] = [-132.21, -18.83, 0.121]
lEuler['nas'] = [-97.52, 6.48, 0.359]
lEuler['nazca'] = [-94.4, 61.0, 0.57]
lEuler['nas_wrt_soam'] = [-83.40, 15.21, 0.287]
lEuler['inca_wrt_soam'] = [-63.76, 22.47, 0.092]
lEuler['eura'] = [
-98.83483039304333, 54.22539546556553, 0.25678223107826376
] # from Altamimi et al., 2012, eura_wrt_itrf2008
euler_vector = np.array(lEuler[frame])
new_Gts = self.remove_pole(euler_vector, verbose=verbose)
if in_place:
self.data = new_Gts.data.copy()
return (new_Gts)
def differentiate(self):
###################################################################
"""
differentiate the current time series
:return: the differentiated time series as a new Gts object
:note : differentiation is made on .data. .data_xyz is set to None.
"""
# import
import inspect
import numpy as np
import pyacs.lib.astrotime
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
new_Gts = self.copy()
data = self.data
data_diff = np.diff(data, axis=0)
# for dates
data_mjd = pyacs.lib.astrotime.decyear2mjd(data[:, 0])
new_data_mjd = (data_mjd[0:-1] + data_mjd[1:]) / 2.
# data_diff[:,0]=data[0:len(data[:,0])-1,0]+data_diff[:,0]
data_diff[:, 0] = pyacs.lib.astrotime.mjd2decyear(new_data_mjd)
new_Gts.data = data_diff.copy()
# set .data_xyz to None
new_Gts.data_xyz = None
return (new_Gts)
def decimate(self,time_step=30.,dates=[],method='median',verbose=False):
###################################################################
"""
decimate a time series
:param time_step: time step in days
:param dates: list of dates where point are forced to be written regardless time_step
:param method: method used to be used to calculated the position. choose among ['median','mean','exact']
:param verbose: verbose mode
:return : new Gts
"""
# import
import inspect
import numpy as np
import pyacs.lib.astrotime as at
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3],__name__,self)
except GtsInputDataNone as error:
# print PYACS WARNING
print( error )
return( self )
# start and end date
start_decyear = self.data[0,0]
end_decyear = self.data[-1,0]
start_mjd = at.decyear2mjd(start_decyear)
end_mjd = at.decyear2mjd(end_decyear)
# new_gts to be filled
new_gts=self.copy()
new_gts.data=np.zeros((1,self.data.shape[1]))
# loop on dates by time_step
for date in np.arange(start_mjd ,end_mjd + time_step / 2. , time_step ):
sub_ts=self.\
extract_periods( [at.mjd2decyear(date-time_step/2.),at.mjd2decyear(date+time_step/2.)] ,\
verbose = verbose)
if sub_ts.data is None:
continue
if method == 'median':
new_obs = np.median(sub_ts.data,axis=0)
if method == 'mean':
new_obs = np.mean(sub_ts.data,axis=0)
if method == 'exact':
new_obs = sub_ts.data[int(sub_ts.data.shape[0])-1,:]
new_gts.data = np.vstack( ( new_gts.data, new_obs.reshape(1,-1) ) )
# case dates are provided
for date in dates:
sub_ts=self.extract_periods( [date,end_decyear] , verbose = verbose)
if sub_ts.data is not None:
new_obs = sub_ts.data[0,:]
if verbose:
print("-- adding observation at user requested date: %lf = %s = doy %d" %
( date, '-'.join(map(str, at.decyear2cal(date))), at.decyear2dayno(date)))
if new_gts.data is not None:
new_gts.data = np.vstack( ( new_gts.data, new_obs.reshape(1,-1) ) )
else:
new_gts.data = new_obs.new_obs.reshape(1,-1)
# remove first obs
if new_gts.data.shape[0] == 1:
if verbose:
print('!!! decimated Gts has no date')
new_gts.data = None
new_gts.data_xyz = None
return(new_gts)
else:
new_gts.data = np.delete(new_gts.data,0,axis=0)
new_gts.neu2xyz()
# reorder
new_gts.reorder(verbose=verbose)
if verbose:
print('-- decimated Gts has ',new_gts.data.shape[0],' entries')
return(new_gts)
def spline(self, smoothing=1, degree=5, date=None):
"""
:param smoothing: Positive smoothing factor used to choose the number of knots. Number of knots will be increased
until the smoothing condition is satisfied:
sum((w[i] * (y[i]-spl(x[i])))**2, axis=0) <= s
:param degree: Degree of the smoothing spline. Must be <= 5. Default is k=3, a cubic spline.
:param date: 1D array of interpolation dates in decimal year, or 'day' for every day. defualt None will interpolate
at data date only.
:return: new gts instance
"""
import numpy as np
from pyacs.gts.Gts import Gts
import inspect
###########################################################################
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
###########################################################################
# import
from scipy.interpolate import UnivariateSpline
import numpy as np
Spline_ts = self.copy()
s_e = UnivariateSpline(self.data[:, 0],
self.data[:, 1],
s=smoothing * 1E-3,
k=degree)
s_n = UnivariateSpline(self.data[:, 0],
self.data[:, 2],
s=smoothing * 1E-3,
k=degree)
s_u = UnivariateSpline(self.data[:, 0],
self.data[:, 3],
s=smoothing * 1E-3,
k=degree)
if date is None:
Spline_ts.data[:, 1] = s_e(self.data[:, 0])
Spline_ts.data[:, 2] = s_n(self.data[:, 0])
Spline_ts.data[:, 3] = s_u(self.data[:, 0])
if isinstance(date, np.ndarray):
Spline_ts.data[:, 1] = s_e(date)
Spline_ts.data[:, 2] = s_n(date)
Spline_ts.data[:, 3] = s_u(date)
if date == 'day':
import pyacs.lib.astrotime as at
np_date = at.mjd2decyear(
np.arange(at.decyear2mjd(self.data[0, 0]),
at.decyear2mjd(self.data[-1, 0])))
Spline_ts.data = np.zeros((np_date.shape[0], 10))
Spline_ts.data[:, 0] = np_date
Spline_ts.data[:, 1] = s_e(np_date)
Spline_ts.data[:, 2] = s_n(np_date)
Spline_ts.data[:, 3] = s_u(np_date)
return (Spline_ts)
def substract_ts_daily(self,ts,verbose=True):
###################################################################
"""
substract the ts provided as argument to the current time series
:param ts: time series to be substracted as a Gts instance
:param verbose: verbose mode
:return : new Gts
:note: this method assumes daily time series
"""
# import
import inspect
import numpy as np
import pyacs.lib.astrotime as at
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3],__name__,self)
except GtsInputDataNone as error:
# print PYACS WARNING
print( error )
return( self )
# check data is not None
try:
if ts.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3],__name__,ts)
except GtsInputDataNone as error:
# print PYACS WARNING
print( error )
return( self )
# find common dates
np_mjd_1 = at.decyear2mjd( self.data[:,0] ).astype( int )
np_mjd_2 = at.decyear2mjd( ts.data[:,0] ).astype( int )
np_mjd_common_dates = np.intersect1d( np_mjd_1 , np_mjd_2 )
if verbose:
print('-- ', np_mjd_common_dates.shape[0] , ' common dates found.')
# test whether there are common dates
if np_mjd_common_dates.shape[0] == 0:
print('! WARNING No common dates between ',self.code,' from ',self.ifile)
print('!!! and ',ts.code, ' from ',ts.ifile)
new_data = None
else:
# extract
mask1 = np.isin( np_mjd_1, np_mjd_common_dates )
data1 = self.data[ mask1 ]
mask2 = np.isin( np_mjd_2, np_mjd_common_dates )
data2 = ts.data[ mask2 ]
if data1.shape != data2.shape:
print("!!! ERROR. Extracted data have different shapes")
new_data = None
# ENU
new_data = data1 - data2
# dates
new_data[:,0] = data1[:,0]
# uncertainties
new_data[:,4:8] = np.sqrt( data1[:,4:8]**2 + data2[:,4:8]**2 )
new_data[:,8:] = 0.
new_code=self.code+'_'+ts.code
new_Gts=self.copy(data=new_data)
new_Gts.code=new_code
# .data_xyz set to None
new_Gts.data_xyz = None
return(new_Gts)
def substract_ts(self, ts, tol=0.05, verbose=True):
###################################################################
"""
substract the ts provided as argument to the current time series
:param ts: time series to be substracted as a Gts instance
:param tol: date tolerance to decide whether two dates are identical in both time series. default = 1/4 day
:param verbose: verbose mode
:return : new Gts
"""
# import
import inspect
import numpy as np
import pyacs.gts.Gts
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
# check data is not None
try:
if ts.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, ts)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
tol = tol / 365.25 # tolerance in decimal year
# find common dates
#common_dates=np.intersect1d(np.round(self.data[:,0],decimals=7),np.round(ts.data[:,0],decimals=7))
lindex = pyacs.gts.Gts.get_index_from_dates(self.data[:, 0], ts.data, tol)
if verbose:
print('-- ', len(lindex), ' common dates found.')
common_dates = self.data[lindex, 0]
# test whether there are common dates
if common_dates.shape[0] == 0:
print('! WARNING No common dates between ', self.code, ' from ',
self.ifile)
print('!!! and ', ts.code, ' from ', ts.ifile)
new_data = None
else:
# extract
extracted_target = self.extract_dates(common_dates, tol=tol)
extracted_ref = ts.extract_dates(common_dates, tol=tol)
new_data = extracted_target.data - extracted_ref.data
new_data[:, 0] = common_dates
new_data[:, 4] = np.sqrt(extracted_target.data[:, 4]**2 +
extracted_ref.data[:, 4]**2)
new_data[:, 5] = np.sqrt(extracted_target.data[:, 5]**2 +
extracted_ref.data[:, 5]**2)
new_data[:, 6] = np.sqrt(extracted_target.data[:, 6]**2 +
extracted_ref.data[:, 6]**2)
new_code = self.code + '_' + ts.code
new_Gts = self.copy(data=new_data)
new_Gts.code = new_code
# .data_xyz set to None
new_Gts.data_xyz = None
return (new_Gts)
def plot(self,
title=None,
loffset=True,
loutliers=True,
verbose=False,
date=[],
yaxis=None,
min_yaxis=None,
yupaxis=None,
xticks_minor_locator=1,
lcomponent=['N', 'E', 'U'],
error_scale=1.0,
lperiod=[[]],
lvline=[],
save_dir_plots='.',
save=None,
show=True,
unit='mm',
date_unit='cal',
date_ref=0.0,
center=True,
superimposed=None,
lcolor=['r', 'g', 'c', 'm', 'y', 'k', 'b'],
label=None,
legend=False,
set_zero_at_date=None,
grid=True,
plot_size=None,
info=[],
xlabel_fmt=None,
**kwargs):
###############################################################################
"""
Create a plot of a North-East-Up time series and related info (offsets, outliers) using Matplotlib
Coordinates of the time series are assumed to be in meters
default plots units will be mm; Use unit='m' to get meters instead
:param title: string to be added to the site name as a plot title
:param loffset: boolean
print a dash vertical line at offset dates
:param loutliers: boolean
print outliers
:param verbose: boolean
verbose mode
:param date: [sdate,edate]
start and end date for plots
sdate and edate in decimal years if date_unit is either 'decyear' or 'cal', or in days if date_unit is 'days'
:param yaxis: [min_y,max_y]
min and max value for the yaxis
if not provided automatically adjusted
:param yupaxis: same as yaxis but applies to the up component only
:param xticks_minor_locator: where xticks_minor_locator will be placed. Float when date_unit is 'decyear' or 'days', a string '%Y','%m','%d' is date_unit is 'cal'.
:param lcomponent: list of components to be plotted (default =['N','E','U'])
:param error_scale: scaling factor for error bars (default = 1.0, 0 means no error bar)
:param lperiod: list of periods to be drawn in background (color=light salmon)
:param lvline: list of dates where vertical lines will be drawn in background (color=green)
:param save_dir_plots: directory used for saving the plots
:param save: name, save the plot into name, if simply True an automatic name is given
:param show: boolean, is True show the plot
:param unit: 'm','cm','mm', default='mm'
:param date_unit: 'decyear' or 'cal' or 'days', default='decyear'
:param date_ref: reference date, default=0.0
:param center: boolean, if True the y_axis is centered around the mean value for the plotted period
:param superimposed: if a gts is provided, it is superimposed to the master, default=None
:param lcolor: color list used for the superimposed time series, default=['r','g','c','m','y','k','b']
:param label: label for superimposed time series to be displayed in legend, default=None
:param legend: boolean. Set true to display label for superimposed time series, default=False
:param set_zero_at_date: date at which the master and superimposed gts will be equal (default=None). date can also be a list with [date,offset_north,offset_east,offset_up]
:param plot_size: plot size as a tuple. Default, best guess.
:param grid: boolean
:param info: title to appear in time series subplots
:param **kwargs: any argument to be passed to matplotlib.pyplot.errorbar
:note: The list of kwargs are:
{ 'linewidth' : 0,\
'marker' : marker_main_symbol ,\
'markersize' : marker_main_size, \
'markerfacecolor' : marker_main_color,\
'markeredgecolor' : marker_main_color,\
'markeredgewidth' : marker_main_colorlw,\
'ecolor' : error_bar_color,\
'elinewidth' : error_bar_linewidth,\
'capsize' : error_bar_capsize }
"""
###########################################################################
# import
###########################################################################
# system
import inspect
import numpy as np
import os.path
# matplotlib
import matplotlib.pyplot as plt
import matplotlib.ticker as ticker
from matplotlib.ticker import MultipleLocator
import matplotlib.dates as mdates
# pyacs
from pyacs.gts.lib.errors import GtsInputDataNone
import pyacs.gts.lib.plot.init_plot_settings
import pyacs.lib.astrotime
from pyacs.gts.Gts import Gts
import pyacs.lib.utils
import pyacs.gts.lib.plot.make_stitle
import pyacs.gts.lib.plot.gts_to_ts
###########################################################################
# lcolor
###########################################################################
if (superimposed is not None) and (isinstance(
superimposed, list)) and (len(superimposed) > 7):
lcolor = np.random.rand(len(superimposed), 3)
###########################################################################
# check data is not None
###########################################################################
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
###########################################################################
# ensure superimposed is either None or a list of gts
###########################################################################
if superimposed is not None:
if not isinstance(superimposed, list):
superimposed = [superimposed]
for agts in superimposed:
if not isinstance(agts, Gts):
raise TypeError
###########################################################################
# ensure lperiod is a list of list
###########################################################################
lperiod = pyacs.lib.utils.__ensure_list_of_list(lperiod)
###########################################################################
# init default plot settings
###########################################################################
H_plot_settings, H_kwargs_errorbar, H_kwargs_errorbar_superimposed = pyacs.gts.lib.plot.init_plot_settings.__init_kwargs(
)
###########################################################################
# init plot settings with user provided values
###########################################################################
for k, v in kwargs.items():
H_kwargs_errorbar[k] = v
# turn show to off by default
plt.ioff()
# some drawing default parameters
params = {'legend.fontsize': 5}
plt.rcParams.update(params)
# For figure caption
txt_component = {}
txt_component['N'] = 'North'
txt_component['E'] = 'East'
txt_component['U'] = 'Up'
# plot size
if plot_size is None:
if len(lcomponent) == 1:
plot_size = H_plot_settings['plot_size_1_component']
if len(lcomponent) == 2:
plot_size = H_plot_settings['plot_size_2_component']
if len(lcomponent) == 3:
plot_size = H_plot_settings['plot_size_3_component']
lperiod_facecolor = 'r'
lperiod_alpha = 0.4
###########################################################################
# format for printing duration in subplot titles
###########################################################################
if date != []: duration = date[-1] - date[0]
else: duration = self.data[-1, 0] - self.data[0, 0]
###########################################################################
# x-tick label
###########################################################################
# adjust tickers for the decyear case
if date_unit == 'decyear' or date_unit == 'days':
if xlabel_fmt != None:
fmt = ticker.FormatStrFormatter(xlabel_fmt)
else:
if duration >= 1.0: fmt = ticker.FormatStrFormatter('%4.0lf')
if duration < 1.0: fmt = ticker.FormatStrFormatter('%6.2lf')
if duration < 0.2: fmt = ticker.FormatStrFormatter('%6.3lf')
if duration < 0.1: fmt = ticker.FormatStrFormatter('%6.4lf')
plt.rcParams['axes.formatter.limits'] = (-7, 17)
if date_unit == 'cal':
if xlabel_fmt is not None:
fmt = mdates.DateFormatter(xlabel_fmt)
###########################################################################
# title and subtitle
###########################################################################
if title is not None:
stitle = title
else:
stitle = self.code
###########################################################################
# info
###########################################################################
if info == []:
info = len(lcomponent) * ['']
###########################################################################
# to_unit
###########################################################################
if unit == 'cm':
to_unit = 100.
if unit == 'mm':
to_unit = 1000.
###########################################################################
# START PLOTTING
###########################################################################
plt.ioff()
f, ax = plt.subplots(len(lcomponent), sharex=True, figsize=plot_size)
###########################################################################
# rotate xticks labels
###########################################################################
f.autofmt_xdate(bottom=0.2, rotation=30, ha='right')
# handle case where only one component and ax is not subscriptable
if not isinstance(ax, np.ndarray):
ax = np.array([ax])
np_date , data = pyacs.gts.lib.plot.gts_to_ts.gts_to_ts( self , \
date=date, \
unit=unit , \
date_unit=date_unit , \
date_ref=date_ref, \
set_zero_at_date = set_zero_at_date, \
center = center )
# master time series
###########################################################################
idx_ax = 0
for component in lcomponent:
# idx_component
if component == 'N':
idx = 0
if component == 'E':
idx = 1
if component == 'U':
idx = 2
# subplot title
str_title = pyacs.gts.lib.plot.make_stitle.make_stitle(
component, info=info[idx_ax])
ax[idx_ax].set_title(str_title,
x=0,
horizontalalignment='left',
fontsize='11')
# plot the data
# print( H_kwargs_errorbar )
ax[idx_ax].errorbar(np_date,
data[:, idx],
yerr=data[:, idx + 3] * error_scale,
**H_kwargs_errorbar)
# plot outliers
if self.outliers != []:
ax[idx_ax].plot(np_date[self.outliers],
data[self.outliers, idx],
'ro',
markersize=H_kwargs_errorbar['markersize'])
# Y-label
ax[idx_ax].set_ylabel(unit)
# grid
ax[idx_ax].grid(grid)
# xticks
if date_unit == 'decyear':
ax[idx_ax].xaxis.set_major_formatter(fmt)
if date_unit == 'cal' and xlabel_fmt is not None:
ax[idx_ax].xaxis.set_major_formatter(fmt)
idx_ax = idx_ax + 1
# superimposed time series
###########################################################################
if superimposed is None:
wsuperimposed = []
else:
wsuperimposed = list(superimposed)
# Get label
if label is None:
label = []
for my_ts in wsuperimposed:
label.append(my_ts.code)
# test
np_date = None
data = None
for i in np.arange(len(wsuperimposed)):
if verbose:
print('-- plotting superimposed time series: ', label[i])
np_date , data = pyacs.gts.lib.plot.gts_to_ts.gts_to_ts( wsuperimposed[i] , \
date=date, \
unit=unit , \
date_unit=date_unit , \
date_ref=date_ref, \
set_zero_at_date = set_zero_at_date, \
center = center )
idx_ax = 0
for component in lcomponent:
# idx_component
if component == 'N':
idx = 0
if component == 'E':
idx = 1
if component == 'U':
idx = 2
# plot the data
error_scale = 0.
ax[idx_ax].errorbar(np_date,
data[:, idx],
yerr=data[:, idx + 3] * error_scale,
color=lcolor[i],
label=label[i])
idx_ax = idx_ax + 1
# plot offset_date of the master time series
###########################################################################
if loffset:
if date_unit == 'days':
if (date_ref is None) or (date_ref == 0):
# reference data is year.000
date_ref = float(int(self.data[0, 0]))
ldate_offsets_in_date_unit = pyacs.lib.astrotime.decyear2mjd(
self.offsets_dates) - pyacs.lib.astrotime.decyear2mjd(date_ref)
if date_unit == 'decyear':
ldate_offsets_in_date_unit = np.array(
self.offsets_dates) - date_ref
if date_unit == 'cal':
ldate_offsets_in_date_unit = pyacs.lib.astrotime.decyear2datetime(
self.offsets_dates)
idx_ax = 0
for component in lcomponent:
for odate in ldate_offsets_in_date_unit:
ax[idx_ax].axvline(x=odate,
linewidth=2,
color='k',
linestyle='--')
idx_ax = idx_ax + 1
# lperiod option: periods to be highlighted
###########################################################################
if lperiod != [[]]:
if date_unit == 'days':
if (date_ref is None) or (date_ref == 0):
# reference data is year.000
date_ref = float(int(self.data[0, 0]))
lperiod = pyacs.lib.astrotime.day_since_decyear(
np.array(lperiod).flatten(), date_ref).reshape(-1, 2)
if date_unit == 'decyear':
lperiod = (np.array(lperiod).flatten() - date_ref).reshape(-1, 2)
if date_unit == 'cal':
lperiod = np.array(
pyacs.lib.astrotime.decyear2datetime(
np.array(lperiod).flatten())).reshape(-1, 2)
# plot color background for periods
idx_ax = 0
for component in lcomponent:
for period in lperiod:
(sdate, edate) = period
ax[idx_ax].axvspan(sdate,
edate,
facecolor=lperiod_facecolor,
alpha=lperiod_alpha)
idx_ax = idx_ax + 1
# vertical lines option
###########################################################################
if lvline != []:
if date_unit == 'decyear':
lvline_in_date_unit = np.array(lvline) - date_ref
if date_unit == 'days':
lvline_in_date_unit = np.array(lvline) - date_ref
if date_unit == 'cal':
lvline_in_date_unit = pyacs.lib.astrotime.decyear2datetime(lvline)
idx_ax = 0
for component in lcomponent:
for odate in lvline_in_date_unit:
ax[idx_ax].axvline(x=odate,
linewidth=2,
color='g',
linestyle='--')
idx_ax = idx_ax + 1
# title
###########################################################################
plt.suptitle(stitle)
# yaxis
###########################################################################
# added JMN 13/01/2021 to set a minimum value for y-scale
if (min_yaxis is not None) and (yaxis is None):
idx_ax = 0
for component in lcomponent:
if component in ['N', 'E']:
cyaxis = ax[idx_ax].get_ylim()
ax[idx_ax].set_ylim(np.min([cyaxis[0], -min_yaxis]),
np.max([cyaxis[1], min_yaxis]))
if component == 'U':
cyupaxis = ax[idx_ax].get_ylim()
ax[idx_ax].set_ylim(np.min([cyupaxis[0], -min_yaxis]),
np.max([cyupaxis[1], min_yaxis]))
idx_ax = idx_ax + 1
if yaxis is not None:
idx_ax = 0
for component in lcomponent:
if component in ['N', 'E']:
ax[idx_ax].set_ylim(yaxis[0], yaxis[1])
if component == 'U' and yupaxis is not None:
ax[idx_ax].set_ylim(yupaxis[0], yupaxis[1])
idx_ax = idx_ax + 1
# date
###########################################################################
if date != []:
# case 'days'
if date_unit == 'days':
if (date_ref is None) or (date_ref == 0):
np_date_x = pyacs.lib.astrotime.decyear2mjd(
np.array(date)) - pyacs.lib.astrotime.decyear2mjd(date[0])
else:
np_date_x = pyacs.lib.astrotime.decyear2mjd(
np.array(date)) - pyacs.lib.astrotime.decyear2mjd(date_ref)
# case 'decyear'
if date_unit == 'decyear':
if (date_ref is None) or (date_ref == 0):
np_date_x = np.array(date)
else:
np_date_x = np.array(date) - date_ref
# case 'cal'
if date_unit == 'cal':
np_date_x = pyacs.lib.astrotime.decyear2datetime(np.array(date))
idx_ax = 0
for component in lcomponent:
if component in ['N', 'E']:
ax[idx_ax].set_xlim(np_date_x[0], np_date_x[1])
if component == 'U' and yupaxis is not None:
ax[idx_ax].set_xlim(np_date_x[0], np_date_x[1])
idx_ax = idx_ax + 1
# X Label
###########################################################################
if date_unit == 'decyear':
if (date_ref is None) or (date_ref == 0.0):
str_xlabel = 'year'
else:
str_xlabel = ("decimal year since %s" %
pyacs.lib.astrotime.decyear2datetime(date_ref))
if date_unit == 'days':
if (date_ref is None) or (date_ref == 0):
# reference data is year.000
date_ref = float(int(self.data[0, 0]))
str_xlabel = ("days since %s" %
pyacs.lib.astrotime.decyear2datetime(date_ref))
if date_unit == 'cal':
str_xlabel = ("calendar date")
ax[idx_ax - 1].set_xlabel(str_xlabel)
# xtcicks minor locator
###########################################################################
if date_unit == 'decyear':
idx_ax = 0
for component in lcomponent:
ax[idx_ax].xaxis.set_minor_locator(
MultipleLocator(float(xticks_minor_locator)))
idx_ax = idx_ax + 1
if date_unit != 'days':
idx_ax = 0
if duration > 100:
xticks_minor_locator = 10
if duration > 1000:
xticks_minor_locator = 100
if duration > 10000:
xticks_minor_locator = 1000
for component in lcomponent:
ax[idx_ax].xaxis.set_minor_locator(
MultipleLocator(float(xticks_minor_locator)))
idx_ax = idx_ax + 1
if date_unit == 'cal':
# default
if self.data[-1, 0] - self.data[0, 0] > 5.:
locator = mdates.YearLocator()
else:
locator = mdates.MonthLocator()
# user provided default
if xticks_minor_locator == '%Y':
locator = mdates.YearLocator() # every year
if xticks_minor_locator == '%m':
locator = mdates.MonthLocator() # every month
if xticks_minor_locator == '%d':
locator = mdates.DayLocator() # every day
idx_ax = 0
for component in lcomponent:
ax[idx_ax].xaxis.set_minor_locator(locator)
idx_ax = idx_ax + 1
# legend
###########################################################################
if legend:
idx_ax = 0
for component in lcomponent:
ax[idx_ax].legend()
idx_ax = idx_ax + 1
# tight_layout
###########################################################################
plt.subplots_adjust(left=None,
bottom=None,
right=None,
top=None,
wspace=None,
hspace=H_plot_settings['hspace'])
# end of subplots
###########################################################################
# save file as png if save option is set
###########################################################################
if save is not None:
# case save is a string
if isinstance(save, str):
# check whether save includes a path or simply a file name
if os.path.sep in save:
# save includes a path information
fname = save
else:
# else
fname = os.path.normpath(save_dir_plots + '/' + save)
elif save:
# save is simply set to True for automatic output file naming
fname = os.path.normpath(save_dir_plots + '/' + self.code + '.png')
# saving output file
if verbose:
print("-- Saving file to ", fname)
# creates the directory of it does not exist
if os.path.sep in fname:
os.makedirs(os.path.dirname(fname), exist_ok=True)
# plt.savefig(fname, dpi=150, facecolor='w', edgecolor='w', orientation='portrait', papertype='a4', format='png',transparent=False, pad_inches=0.1)
# change 05/04/2021 papertype appears to be decommissioned for matplotlib 3.3
plt.savefig(fname,
dpi=150,
facecolor='w',
edgecolor='w',
orientation='portrait',
format='png',
transparent=False,
pad_inches=0.1)
# show
###########################################################################
if show:
# do not block program execution when displaying figures
plt.ion()
plt.show()
else:
plt.close(f)
# returns the Gts for pyacs convention compatibility
new_Gts = self.copy()
return (new_Gts)
def get_coseismic(self,
eq_date,
window_days=5,
sample_after=1,
method='median',
in_place=False):
###################################################################
"""
Get coseismic displacement at a given date.
Coseismic displacement is estimated as the position difference between the median of window_days before
the earthquake date and the median of sample_after samples after the earthquake date.
note: only median method implemented
"""
# import
import inspect
import numpy as np
import pyacs.lib.astrotime
from datetime import timedelta
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
# eq date as datetime
eq_datetime = pyacs.lib.astrotime.decyear2datetime(eq_date)
# get the eq day at 00:00:00 and then go backward by window_days
wdatetime = eq_datetime.replace(
hour=0, minute=0, second=0,
microsecond=0) - timedelta(days=window_days)
# convert back to decimal year
wdecyear = pyacs.lib.astrotime.datetime2decyear(wdatetime)
# make extraction between eq - window_days and eq_date
lindex = np.where((self.data[:, 0] > wdecyear)
& (self.data[:, 0] < eq_date))[0]
data_before_eq = self.data[lindex, 1:4]
# take sample_after values after eq_date
lindex = np.where(self.data[:, 0] > eq_date)[0][0]
data_after_eq = self.data[lindex:lindex + sample_after, 1:4]
# get offset values
disp = np.median(data_after_eq, axis=0) - np.median(data_before_eq, axis=0)
# calculates std before eq
std_sdate = np.std(data_before_eq, axis=0)
# calculates std after eq
std_edate = np.std(data_after_eq, axis=0)
# if std cannot be calculated or is 0.
if np.max(std_sdate) == 0.0:
lindex = np.where((self.data[:, 0] > wdecyear)
& (self.data[:, 0] < eq_date))[0]
std_sdate = np.mean(self.data[lindex, 4:7], axis=0)
if np.max(std_edate) == 0.0:
lindex = np.where(self.data[:, 0] > eq_date)[0][0]
std_sdate = np.mean(self.data[lindex, 4:7], axis=0)
std_disp = np.sqrt(std_edate**2 + std_sdate**2)
new_Gts = self.copy()
# modify the output time series
new_Gts.data_xyz = None
new_Gts.data[:, 1:4] = new_Gts.data[:, 1:4] - np.median(data_before_eq,
axis=0)
if in_place:
self.offsets_values = np.array([eq_date] + disp.tolist() +
std_disp.tolist()).reshape(1, 7)
return (self)
del new_Gts
else:
new_Gts.offsets_values = np.array([eq_date] + disp.tolist() +
std_disp.tolist()).reshape(1, 7)
return (new_Gts)
def detrend_median(self,
delta_day=None,
in_place=False,
periods=[],
exclude_periods=[],
verbose=False,
auto=False):
###############################################################################
"""
Calculates a velocity using the median of pair of displacements exactly separated by one year, inspired from MIDAS
If the time series has less than a year of data, then the time series is kept untouched.
:param delta_day: if None, it is one year, if 0 then it is the relax mode for campaign data,
any integer is the time delta (in days) used to compute velocity.
:param in_place: boolean, if True, in_place, if False, returns a new Gts instance (default)
:param periods: periods (list of lists) to be included for trend calculation
:param exclude_periods: periods (list of lists) to be excluded for trend calculation
:param verbose: verbose mode
:param auto: if True, then start will delta_day=None, if it fails or found less than 100 pairs then use delta_day=0,
if fails then use regular detrend
:note: returns None if time series is shorter than 1 year
"""
import numpy as np
from pyacs.gts.Gts import Gts
import inspect
# after this method .data and .data_xyz are not consistent so .data_xyz is set to None
self.data_xyz = None
###########################################################################
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
###########################################################################
###########################################################################
# 1-yr threshold
###########################################################################
if (self.data[-1, 0] - self.data[0, 0]) < 1.:
if verbose:
print("!!! WARNING: time series shorter than 1 year for side: %s" %
self.code)
return (None)
###########################################################################
# run the appropriate estimator using autoreference if auto is True
###########################################################################
if auto:
tts = self.detrend_median(delta_day=None, in_place=in_place, \
periods=periods, exclude_periods=exclude_periods, \
verbose=verbose, auto=False)
if tts is None:
tts = self.detrend_median(delta_day=0, in_place=in_place, \
periods=periods, exclude_periods=exclude_periods, \
verbose=verbose, auto=False)
if tts is None:
tts = self.detrend(in_place=in_place, \
periods=periods, exclude_periods=exclude_periods, \
verbose=verbose)
return (tts)
###########################################################################
# start main detrend_median
###########################################################################
import pyacs.lib.astrotime
tmp_ts = self.remove_outliers()
if periods != []:
tmp_ts = tmp_ts.extract_periods(periods)
if exclude_periods != []:
tmp_ts = tmp_ts.exclude_periods(periods)
# minimum number of pair velocity estimates to consider the result
min_vel_estimate = 100
###########################################################################
# MIDAS-like method, using 1-year pair data for velocity estimate
###########################################################################
if delta_day is None:
duration_in_year = tmp_ts.data[-1, 0] - tmp_ts.data[0, 0]
if duration_in_year >= 1.:
H_year = {}
# creates H_year arrays
for year in np.arange(int(tmp_ts.data[0, 0]),
int(tmp_ts.data[-1, 0]) + 1):
H_year[year] = np.zeros((365, 3))
H_year[year][:, :] = np.nan
# deal with dates
np_mjd = pyacs.lib.astrotime.decyear2mjd(tmp_ts.data[:, 0])
(np_doy, np_year, _np_ut) = pyacs.lib.astrotime.mjd2dayno(np_mjd)
np_doy = np_doy.astype(int)
np_year = np_year.astype(int)
# fill H_year arrays
for i in np.arange(np_doy.shape[0]):
if np_doy[i] != 366:
H_year[np_year[i]][np_doy[i] - 1, :] = tmp_ts.data[i, 1:4]
# stack the velocities
np_vel = np.zeros((365 * (len(H_year) - 1), 3))
np_vel[:, :] = np.nan
i = 0
for year in sorted(H_year.keys())[0:-1]:
np_vel[i * 365:(i + 1) * 365] = H_year[year + 1] - H_year[year]
i = i + 1
# test whether there are at least 3 non-Nan numbers
if np.count_nonzero(~np.isnan(np_vel)) < min_vel_estimate:
return (None)
# calculates the median velocity
med_vel = np.nanmedian(np_vel, axis=0)
[med_vel_north, med_vel_east, med_vel_up] = med_vel
# calculate uncertainty and refined value of vel using step 2 from Blewitt et al. (2016) p. 2057
# remove Nan
np_vel_north = np_vel[:, 0][np.logical_not(np.isnan(np_vel[:, 0]))]
np_vel_east = np_vel[:, 1][np.logical_not(np.isnan(np_vel[:, 1]))]
np_vel_up = np_vel[:, 2][np.logical_not(np.isnan(np_vel[:, 2]))]
# calculates sigma
fabs_vn = np.fabs(np_vel_north - med_vel_north)
fabs_ve = np.fabs(np_vel_east - med_vel_east)
fabs_vu = np.fabs(np_vel_up - med_vel_up)
sigma_vn = 1.4826 * np.median(fabs_vn)
sigma_ve = 1.4826 * np.median(fabs_ve)
sigma_vu = 1.4826 * np.median(fabs_vu)
# removes all vel > 1.4826 * med_vel
np_vel_north_cln = np_vel_north[np.where(fabs_vn <= 2 * sigma_vn)]
np_vel_east_cln = np_vel_east[np.where(fabs_ve <= 2 * sigma_ve)]
np_vel_up_cln = np_vel_up[np.where(fabs_vu <= 2 * sigma_vu)]
# get the new vel estimates
med_vn = np.median(np_vel_north_cln)
med_ve = np.median(np_vel_east_cln)
med_vu = np.median(np_vel_up_cln)
# get the new sigma
med_sigma_vn = 1.4826 * np.median(
np.fabs(np_vel_north_cln - med_vn))
med_sigma_ve = 1.4826 * np.median(
np.fabs(np_vel_east_cln - med_ve))
med_sigma_vu = 1.4826 * np.median(np.fabs(np_vel_up_cln - med_vu))
# empirical factor for correlated noise
ef = 3.
sigma_vn = 1.2533 * med_sigma_vn / np.sqrt(fabs_vn.shape[0]) * ef
sigma_ve = 1.2533 * med_sigma_ve / np.sqrt(fabs_ve.shape[0]) * ef
sigma_vu = 1.2533 * med_sigma_vu / np.sqrt(fabs_vu.shape[0]) * ef
# final med_vel
med_vel = np.array(
[med_vn, med_ve, med_vu, sigma_vn, sigma_ve, sigma_vu])
else:
# time series has less than a year of data
# velocity is set to 0.
med_vel = np.array([0., 0., 0.])
# end case 1-yr
elif (isinstance(delta_day, int) and delta_day == 0):
###########################################################################
# case campaign, close to relaxed method in MIDAS
###########################################################################
H_year = {}
lvel = []
# creates H_year arrays
for year in np.arange(int(tmp_ts.data[0, 0]),
int(tmp_ts.data[-1, 0]) + 1):
lindex = np.where(tmp_ts.data.astype(int)[:, 0] == year)
# tt = tmp_ts.extract_periods([float(year),float(year)+0.999999])
# if tt.data is not None:
# H_year[year] = tt.data[:,:4]
H_year[year] = tmp_ts.data[lindex[0], :4]
# removes years with no available data
# H_year = {key:val for key, val in H_year.items() if val is not None}
# calculates velocity values for all data pairs separated by more than one year
# while H_year != {}:
for year_start in sorted(list(H_year.keys())):
# np_start = H_year[year_start]
for year_end in sorted(list(H_year.keys())):
ok = False
for i in np.arange(H_year[year_end].shape[0]):
for j in np.arange(H_year[year_start].shape[0]):
# check wether the pair date includes a given offset date
include_offset_date = False
for odate in self.offsets_dates:
if (odate >= H_year[year_start][j, 0]) and (
odate <= H_year[year_end][i, 0]):
include_offset_date = True
if not include_offset_date:
# displacement
v = (H_year[year_end][i] -
H_year[year_start][j])[1:]
# delta_year
delta_year = H_year[year_end][
i, 0] - H_year[year_start][j, 0]
# append to lvel
if delta_year > 0.90:
lvel.append(v / delta_year)
ok = True
if verbose:
print("-- adding %.5lf - %.5lf " %
(H_year[year_start][j, 0],
H_year[year_end][i, 0]))
print(i, j, v / delta_year)
print(H_year[year_end][i],
H_year[year_start][j])
# if some pairs have already been found, stop here to mitigate the effect of offsets
# if ok:
# break
del H_year[year_start]
# calculates the median velocity
np_vel = np.array(lvel)
np_vel_north = np_vel[:, 0]
np_vel_east = np_vel[:, 1]
np_vel_up = np_vel[:, 2]
med_vel = np.median(np_vel, axis=0)
[med_vel_north, med_vel_east, med_vel_up] = med_vel
# calculates sigma
fabs_vn = np.fabs(np_vel_north - med_vel_north)
fabs_ve = np.fabs(np_vel_east - med_vel_east)
fabs_vu = np.fabs(np_vel_up - med_vel_up)
sigma_vn = 1.4826 * np.median(fabs_vn)
sigma_ve = 1.4826 * np.median(fabs_ve)
sigma_vu = 1.4826 * np.median(fabs_vu)
# removes all vel > 1.4826 * med_vel
np_vel_north_cln = np_vel_north[np.where(fabs_vn <= 2 * sigma_vn)]
np_vel_east_cln = np_vel_east[np.where(fabs_ve <= 2 * sigma_ve)]
np_vel_up_cln = np_vel_up[np.where(fabs_vu <= 2 * sigma_vu)]
# get the new vel estimates
med_vn = np.median(np_vel_north_cln)
med_ve = np.median(np_vel_east_cln)
med_vu = np.median(np_vel_up_cln)
# get the new sigma
med_sigma_vn = 1.4826 * np.median(np.fabs(np_vel_north_cln - med_vn))
med_sigma_ve = 1.4826 * np.median(np.fabs(np_vel_east_cln - med_ve))
med_sigma_vu = 1.4826 * np.median(np.fabs(np_vel_up_cln - med_vu))
# empirical factor for correlated noise
ef = 3.
sigma_vn = 1.2533 * med_sigma_vn / np.sqrt(fabs_vn.shape[0]) * ef
sigma_ve = 1.2533 * med_sigma_ve / np.sqrt(fabs_ve.shape[0]) * ef
sigma_vu = 1.2533 * med_sigma_vu / np.sqrt(fabs_vu.shape[0]) * ef
# final med_vel
med_vel = np.array(
[med_vn, med_ve, med_vu, sigma_vn, sigma_ve, sigma_vu])
else:
###########################################################################
# case delta_day with integer value
###########################################################################
def delta(n):
"""
Simple delta function for integer
"""
if not n == 0:
return (1)
else:
return (0)
# first form the day vector
np_mjd_data = list(
map(int, pyacs.lib.astrotime.decyear2mjd(tmp_ts.data[:, 0])))
# so the index are
i_np_mjd_data = np_mjd_data - np.min(np_mjd_data)
# the void array filled with np.nan
void = np.zeros(
(np.max(np_mjd_data) - np.min(np_mjd_data) + 1, 3)) * np.nan
# I fill void with the time series at existing dates
void[i_np_mjd_data] = tmp_ts.data[:, 1:4]
# now reshape the vector for easy pair differentiation
# if TS = void[ : , (0,1,2) ] easy diff would be achieved by working on a array W = TS.reshape(-1,delta_day)
# however, this requires to complete the number of lines of TS with np.nan to be able to reshape it
CTS = np.zeros((delta_day - np.mod(void.shape[0], delta_day)) *
delta(np.mod(void.shape[0], delta_day))) * np.nan
# init med_vel
med_vel = np.array([0., 0., 0.])
to_year = 365.25 / float(delta_day)
for i in [0, 1, 2]:
med_vel[i] = np.nanmedian(
np.diff(np.append(void[:, i], CTS).reshape(-1, delta_day).T,
axis=1)) * to_year
###########################################################################
# return detrended time series
###########################################################################
new_gts = self.remove_velocity(med_vel)
new_gts.outliers = self.outliers
new_gts.offsets_values = self.offsets_values
new_gts.velocity = med_vel
# if in_place:
# self.velocity = new_gts.velocity
# return( self )
# else:
# return( new_gts )
return (self.remove_velocity(med_vel, in_place=in_place))
def plot(self,\
title=None,\
loffset=True,\
loutliers=True,\
verbose=False,\
date=[],\
yaxis=None, \
yupaxis=None, \
lcomponent=['N','E','U'],\
error_scale=1.0, \
lperiod=[[]],\
lvline=[],\
save_dir_plots='.', \
save=None,\
show=True,\
unit='mm',\
date_unit='decyear', \
date_ref=0.0, center=True,\
superimposed=None, \
superimposed_equal_date=None, \
plot_size=None, \
info = [], \
**kwargs):
###############################################################################
"""
Create a plot of a North-East-Up time series and related info (offsets, outliers) using Matplotlib
Coordinates of the time series are assumed to be in meters
default plots units will be mm; Use unit='m' to get meters instead
:param title: string to be added to the site name as a plot title
:param loffset: boolean
print a dash vertical line at offset dates
:param loutliers: boolean
print outliers
:param verbose: boolean
verbose mode
:param date: [sdate,edate]
start and end date for plots
sdate and edate in decimal years if date_unit is either 'decyear' or 'cal', or in days if date_unit is 'days'
:param yaxis: [min_y,max_y]
min and max value for the yaxis
if not provided automatically adjusted
:param yupaxis: same as yaxis but applies to the up component only
:param lcomponent: list of components to be plotted (default =['N','E','U'])
:param error_scale: scaling factor for error bars (default = 1.0, 0 means no error bar)
:param lperiod: list of periods to be drawn in background (color=light salmon)
:param lvline: list of dates where vertical lines will be drawn in background (color=green)
:param save_dir_plots: directory used for saving the plots
:param save: name, save the plot into name, if simply True an automatic name is given
:param show: boolean, is True show the plot
:param unit: 'm','cm','mm', default='mm'
:param date_unit: 'decyear' or 'cal' or 'days', default='decyear'
:param date_ref: reference date, default=0.0
:param center: boolean, if True the y_axis is centered around the mean value for the plotted period
:param superimposed: if a gts is provided, it is superimposed to the master, default=None
:param superimposed_equal_date: date at which the master and superimposed gts will be equal (default=None). date can also be a list with [date,offset_north,offset_east,offset_up]
:param plot_size: plot size as a tuple. Default, best guess.
:param info: info to appear in time series subtitles. A list of keywords among: 'wrms','bias','vel','period'
:param **kwargs: any argument to be passed to plot_date (unit_date='cal') or errorbar (unit_date='decyear' or 'days')
:note: The list of kwargs are:
H_kwargs_plot_date={\\
'marker' : marker_main_symbol,\\
'markersize' : marker_main_size, \\
'markerfacecolor' : marker_main_color,\\
'markeredgecolor' : marker_main_color,\\
'markeredgewidth' : marker_main_colorlw \\
}
H_kwargs_plot_date_superimposed={ \\
'color' : line_2_color,\\
'linewidth' : line_2_width,\\
'linestyle' : line_2_style,\\
'marker' : marker_2_symbol,\\
'markersize' : marker_2_size, \\
'markerfacecolor' : marker_2_color,\\
'markeredgecolor' : marker_2_color,\\
'markeredgewidth' : marker_2_colorlw \\
}
H_kwargs_errorbar={'linewidth' : 0,\
'marker' : marker_main_symbol ,\
'markersize' : marker_main_size, \
'markerfacecolor' : marker_main_color,\
'markeredgecolor' : marker_main_color,\
'markeredgewidth' : marker_main_colorlw,\
'ecolor' : error_bar_color,\
'elinewidth' : error_bar_linewidth,\
'capsize' : error_bar_capsize }
H_kwargs_errorbar_superimposed={ \
'color' : line_2_color,\
'linewidth' : line_2_width,\
'linestyle' : line_2_style,\
'marker' : marker_2_symbol,\
'markersize' : marker_2_size, \
'markerfacecolor' : marker_2_color,\
'markeredgecolor' : marker_2_color,\
'markeredgewidth' : marker_2_colorlw, \
'ecolor' : error_bar_color,\
'elinewidth' : error_bar_linewidth,\
'capsize' : error_bar_capsize }
"""
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3],__name__,self)
except GtsInputDataNone as error:
# print PYACS WARNING
print( error )
return( self )
# cdata
#if not self.cdata(data=True):
# print '! Cannot make plot. Problem with .data attribute.'
# return(self)
# deal with **kwargs for plot settings
__init_kwargs()
if date_unit == 'cal':
# **kwarg for matplotlib plot_date
for key,value in list(kwargs.items()):
H_kwargs_plot_date[key]=value
if (date_unit == 'days') or (date_unit == 'decyear'):
# **kwarg for matplotlib errorbar
for key,value in list(kwargs.items()):
H_kwargs_errorbar[key]=value
# we usually want the plot in mm
if unit=='mm':to_mm=1000.0
if unit=='cm':to_mm=100.0
if unit=='m':to_mm=1.0
# import new modules
import matplotlib.pyplot as plt
import matplotlib.ticker as ticker
import pyacs.lib.astrotime
# turn show to off by default
plt.ioff()
# some drawing default parameters
params = {'legend.fontsize': 4}
plt.rcParams.update(params)
# For figure caption
txt_component={}
txt_component['N']='North'
txt_component['E']='East'
txt_component['U']='Up'
# format for printing duration in subplot titles
if date != []:duration=date[-1]-date[0]
else:duration=self.data[-1,0]-self.data[0,0]
# adjust tickers for the decila years case
if date_unit=='decyear':
fmt=ticker.FormatStrFormatter('%6.1lf')
if duration < 1.0:fmt=ticker.FormatStrFormatter('%6.2lf')
if duration < 0.2:fmt=ticker.FormatStrFormatter('%6.3lf')
if duration < 0.1:fmt=ticker.FormatStrFormatter('%6.4lf')
plt.rcParams['axes.formatter.limits']=(-7,17)
# ensure superimposed is either None or a list of gts
if superimposed is not None:
if not isinstance(superimposed,list):
superimposed=[superimposed]
for agts in superimposed:
if not isinstance(agts,Gts):
raise TypeError
#for subplot title information
if self.code:stitle=self.code
else:stitle=''
# list of color for superimposed ts
lcolor = ['r','g','c','m','y','k','b']
if superimposed is not None:
i=0
for agts in superimposed:
stitle+=' / '
stitle+=agts.code+("-%s" % lcolor[i])
i = i + 1
if title is not None:
stitle = stitle + ' ' + title
# ensure lperiod is a list of list
lperiod = __ensure_list_of_list(lperiod)
# plot size
if plot_size is None:
if len(lcomponent)==1:plot_size=plot_size_1_component
if len(lcomponent)==2:plot_size=plot_size_2_component
if len(lcomponent)==3:plot_size=plot_size_3_component
plt.figure(num=None, figsize=plot_size)
# a small routine to select the data to be plotted depending on the component
def __data_4_plot(gts, component, unit='mm'):
if component=='N':data=gts.data[:,(0,1,4)]
if component=='E':data=gts.data[:,(0,2,5)]
if component=='U':data=gts.data[:,(0,3,6)]
data[:,1:]=data[:,1:]*to_mm
return(data)
# handle superimposed_equal_date option
if superimposed_equal_date is not None:
center = False
wgts=self.set_zero_at_date(superimposed_equal_date,offset=0.0)
for i in np.arange( len(superimposed)):
superimposed[i] = superimposed[i].set_zero_at_date(superimposed_equal_date,offset=0.0)
wsuperimposed=superimposed
else:
wgts=self
wsuperimposed=superimposed
# Create frames
for component in lcomponent:
if component == 'N' and len(lcomponent)==3:
nsubplot=311
if component == 'E' and len(lcomponent)==3:
nsubplot=312
if component == 'U' and len(lcomponent)==3:
nsubplot=313
if component == 'N' and len(lcomponent)==2:
nsubplot=211
if component == 'E' and len(lcomponent)==2:
nsubplot=212
if len(lcomponent)==1:
nsubplot=111
# setup_save current subplot
plt.subplot(nsubplot)
# if date_unit=='decyear':
# plt.subplot(nsubplot).xaxis.set_major_formatter(fmt)
if wgts.velocity is None:
vel=None
else:
if component == 'N':vel=wgts.velocity[0]*to_mm
if component == 'E':vel=wgts.velocity[1]*to_mm
if component == 'U':vel=wgts.velocity[2]*to_mm
str_title=__make_stitle(component,vel,__data_4_plot(wgts,component),unit, info=info)
# plot the data
if date_unit == 'cal':
(xmin_subplot,xmax_subplot) = __plot_ts(plt,__data_4_plot(wgts,component),\
loutliers=wgts.outliers,\
date=date,\
date_unit=date_unit,\
ref_date=date_ref,\
yaxis=yaxis,\
center=center,\
superimposed=None,\
verbose=verbose,\
error_scale=error_scale,
**H_kwargs_plot_date)
if (date_unit == 'days') or (date_unit == 'decyear'):
(xmin_subplot,xmax_subplot) = __plot_ts(plt,__data_4_plot(wgts,component),\
loutliers=wgts.outliers,\
date=date,\
date_unit=date_unit,\
ref_date=date_ref,\
yaxis=yaxis,\
center=center,\
superimposed=None,\
verbose=verbose,\
error_scale=error_scale,
**H_kwargs_errorbar)
# plot offset dates as vertical lines
if loffset:
if date_unit == 'days':
if (date_ref is None) or (date_ref == 0):
# reference data is year.000
date_ref=float(int(wgts.data[0,0]))
ldate_offsets_in_date_unit=pyacs.lib.astrotime.day_since_decyear(wgts.offsets_dates,date_ref)
if date_unit == 'decyear':
ldate_offsets_in_date_unit=np.array(wgts.offsets_dates)-date_ref
if date_unit == 'cal':
ldate_offsets_in_date_unit = __decyear2num(wgts.offsets_dates)
for odate in ldate_offsets_in_date_unit:
if odate < xmax_subplot and odate > xmin_subplot:
plt.axvline(x= odate ,linewidth=2, color='g', linestyle='--')
# subplot title
plt.title(str_title, x=0, horizontalalignment='left', fontsize= '11')
# superimposed option
if superimposed is not None:
for i in np.arange(len(wsuperimposed)):
agts = wsuperimposed[i]
# color for agts
H_kwargs_plot_date_superimposed['color'] = lcolor[i]
H_kwargs_errorbar_superimposed['color'] = lcolor[i]
# calendar date
if date_unit == 'cal':
(xmin_subplot,xmax_subplot) = __plot_ts(plt,__data_4_plot( agts , component),\
loutliers = agts.outliers,\
date=date,\
date_unit=date_unit,\
ref_date=date_ref,\
yaxis=yaxis,\
center=center,\
superimposed=None,\
verbose=verbose,\
error_scale=error_scale,
**H_kwargs_plot_date_superimposed)
# date_init is 'days' or 'decyear'
if (date_unit == 'days') or (date_unit == 'decyear'):
(xmin_subplot,xmax_subplot) = __plot_ts(plt,__data_4_plot( agts ,component),\
loutliers = agts.outliers,\
date=date,\
date_unit=date_unit,\
ref_date=date_ref,\
yaxis=yaxis,\
center=center,\
superimposed=None,\
verbose=verbose,\
error_scale=error_scale,
**H_kwargs_errorbar_superimposed)
# lperiod option: periods to be highlighted
if lperiod != [[]]:
if date_unit == 'days':
if (date_ref is None) or (date_ref == 0):
# reference data is year.000
date_ref=float(int(wgts.data[0,0]))
__plot_lperiod(plt,lperiod,date_unit,date_ref)
# vertical lines option
if lvline != []:
if date_unit == 'days':
if (date_ref is None) or (date_ref == 0):
# reference data is year.000
date_ref=float(int(wgts.data[0,0]))
ldate_vline_in_date_unit=pyacs.lib.astrotime.day_since_decyear(lvline,date_ref)
if date_unit == 'decyear':
ldate_vline_in_date_unit=np.array(lvline)-date_ref
if date_unit == 'cal':
ldate_vline_in_date_unit = __decyear2num(lvline)
for odate in ldate_vline_in_date_unit:
if odate < xmax_subplot and odate > xmin_subplot:
plt.axvline(x= odate ,linewidth=1, color='k', linestyle='-')
# Y-label
plt.ylabel(unit)
# Xticks
if date_unit == 'decyear':
plt.gca().xaxis.set_major_formatter(fmt)
# X Label
if date_unit=='decyear':
if (date_ref is None) or (date_ref == 0.0):
str_xlabel='year'
else:
str_xlabel=("decimal year since %s" %pyacs.lib.astrotime.decyear2datetime(date_ref))
if date_unit=='days':
if (date_ref is None) or (date_ref == 0):
# reference data is year.000
date_ref=float(int(wgts.data[0,0]))
str_xlabel=("days since %s" %pyacs.lib.astrotime.decyear2datetime(date_ref))
if date_unit=='cal':
str_xlabel=("calendar date")
plt.xlabel(str_xlabel)
plt.suptitle(stitle)
# tight_layout
plt.subplots_adjust(left=None, bottom=None, right=None, top=None, wspace=None, hspace=hspace)
#plt.tight_layout()
# end of subplots
# save file as png if save option is set
if save is not None:
# import
import os
import os.path
# case save is a string
if isinstance(save, str):
# check whether save includes a path or simply a file name
if os.path.sep in save:
# save includes a path information
fname = save
else:
# else
fname = os.path.normpath(save_dir_plots+'/'+save)
elif save:
# save is simply set to True for automatic output file naming
fname = os.path.normpath( save_dir_plots + '/' + self.code+'.png' )
# saving output file
if verbose:
print("-- Saving file to ",fname)
# creates the directory of it does not exist
if os.path.sep in fname:
os.makedirs(os.path.dirname(fname), exist_ok=True)
plt.savefig(fname, dpi=150, facecolor='w', edgecolor='w', orientation='portrait', papertype='a4', format='png',transparent=False, pad_inches=0.1)
# show
if show:
# do not block program execution when displaying figures
plt.ion()
plt.show()
# returns the Gts for pyacs convention compatibility
new_Gts=self.copy()
return(new_Gts)
def extract_dates(self, dates, tol=0.05, in_place=False, verbose=True):
###################################################################
"""
Returns a time series extracted for a given list of dates
:param dates: dates either as a list or 1D numpy array of decimal dates
:param tol: date tolerance in days to assert that two dates are equal (default 0.05 day)
:param in_place: if True, will make change in place, if False, return s a new time series
:param verbose: boolean, verbose mode
"""
# import
import inspect
import numpy as np
import pyacs.gts
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
# working gts
new_gts = self.copy()
# case .data_xyz is None
if new_gts.data_xyz is None:
new_gts.neu2xyz(corr=True)
else:
# check data/data_xyz consistency
try:
if not new_gts.cdata(data=True):
# raise exception
from pyacs.gts.lib.errors import GtsCDataError
raise GtsCDataError(inspect.stack()[0][3], __name__, self)
except GtsCDataError as error:
print(error)
return (self)
new_data = None
# extract dates
index = np.array(
pyacs.gts.Gts.get_index_from_dates(dates, self.data, tol=tol))
if verbose:
print('-- Extracting ', index.shape[0], ' entries from Gts or code: ',
self.code)
if index.shape[0] > 0:
new_data_xyz = self.data_xyz[index, :]
new_sigma = self.data[index, 4:]
else:
new_data = None
if verbose:
print("-- time series ", self.code,
" does not have dates at the requested dates ")
# handles outliers
if new_data is not None:
ldate_outliers = self.data[:, 0][self.outliers]
lupdated_outliers = pyacs.gts.Gts.get_index_from_dates(ldate_outliers,
new_data,
tol=tol)
else:
lupdated_outliers = []
if verbose:
print('-- Transmitting ', len(lupdated_outliers),
' outliers to the extracted Gts ')
# case observations
new_gts.data_xyz = new_data_xyz
# handle outliers
ldate_outliers = self.data[:, 0][self.outliers]
lupdated_outliers = pyacs.gts.Gts.get_index_from_dates(ldate_outliers,
new_data_xyz,
tol=0.05)
# handles offsets_date
upd_offsets = []
for offset_date in self.offsets_dates:
if offset_date >= new_data_xyz[0,
0] and offset_date <= new_data_xyz[-1,
0]:
upd_offsets.append(offset_date)
# handles X0,Y0,Z0
new_gts.X0 = new_data_xyz[0, 1]
new_gts.Y0 = new_data_xyz[0, 2]
new_gts.Z0 = new_data_xyz[0, 3]
# re-generate NEU time series
new_gts.xyz2neu(corr=False)
# re-populate the uncertainties columns
new_gts.data[:, 4:] = new_sigma
# offsets & outliers
new_gts.offsets_dates = upd_offsets
new_gts.outliers = lupdated_outliers
if in_place:
self = new_gts
return (self)
else:
return (new_gts)
def make_model(self,
option='detrend',
method='L2',
loutlier=None,
in_place=False):
"""
Estimate linear model parameters using least squares
input: data: Gts format
option are: 'detrend'/'detrend_annual'/'detrend_seasonal'
output: new Gts object: time series is now the residuals wrt to the model and its associated values (vel, annual, semi-annual etc)
"""
import numpy as np
from pyacs.gts.Gts import Gts
import inspect
# after this method .data and .data_xyz are not consistent so .data_xyz is set to None
self.data_xyz = None
###########################################################################
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
###########################################################################
import pyacs.lib.glinalg as glinalg
import pyacs.lib.robustestimators as RobEst
# from gts_estimators import least_square
data = np.copy(self.data)
### REMOVE OUTLIERS FOR LINEAR ESTIMATION
if loutlier: data = np.delete(data, loutlier, axis=0)
### REMOVE OFFSET DATES OUTSIDE THE TIME SPAN OF THE TIME SERIES
if self.offsets_dates is not None and len(self.offsets_dates) > 0:
# keep offsets_dates only within the time series
sel_offsets_dates = \
[self.offsets_dates[i] for i in range(len(self.offsets_dates)) if
(self.offsets_dates[i] > self.data[0, 0] and self.offsets_dates[i] < self.data[-1, 0])]
noffset = len(sel_offsets_dates)
### offsets_values = [time_offsets offsets_NEU s_offsets_NEU]
offsets_values = np.zeros((noffset, 7))
offsets_values[:, 0] = self.offsets_dates
else:
noffset = 0
offsets_values = None
# REF DATES
t_ref = self.data[0, 0]
t_ref_seasonal = 2010.0
# INDEX
if option == 'detrend_seasonal':
del_index = []
n_default_unknown = 6
elif option == 'detrend_annual':
del_index = [4, 5]
n_default_unknown = 4
elif option == 'detrend':
del_index = [2, 3, 4, 5]
n_default_unknown = 2
else:
print(
' ERROR!!! check the option of estimation: detrend/detrend_seasonal/detrend_annual'
)
# INIT VEL, ANNUAL, SEMI_ANNUAL, RESIDUALS ARRAYS
### vel = [vel_N vel_E vel_U svel_N s_vel_E svel_U]
vel = np.zeros(6)
### annual = [amplitude_NEU phase_NEU]
annual = []
### semi_annual = [amplitude_NEU phase_NEU]
semi_annual = []
residuals = np.zeros(data.shape)
residuals[:, 0] = data[:, 0]
ndate = len(data[:, 0])
# BUILD LINEAR SYSTEM
for k in range(1, 4):
## write matrix A in general case
A = np.zeros([ndate, (6 + noffset)], float)
for i in range(ndate):
ti = data[i, 0]
A[i, 0], A[i, 1], A[i, 2], A[i, 3], A[i, 4], A[i, 5] = 1., (ti - t_ref), \
np.cos(2. * np.pi * (ti - t_ref_seasonal)), np.sin(
2. * np.pi * (ti - t_ref_seasonal)), \
np.cos(4. * np.pi * (ti - t_ref_seasonal)), np.sin(
4. * np.pi * (ti - t_ref_seasonal))
## for offsets
for j in range(noffset):
if ti > self.offsets_dates[j]: A[i, (6 + j)] = 1.
### take the design matrix
A = np.delete(A, del_index, axis=1)
# solve
(X, COV, V) = glinalg.lsw_full(A, data[:, k], data[:, k + 3])
if method == 'L1':
(X, V) = RobEst.Dikin(A, data[:, k], data[:, k + 3], eps=1.0E-4)
s_X = np.sqrt(np.diag(COV))
## calculate residuals, predicted mb_files
residuals[:, k] = V
residuals[:, k + 3] = data[:, k + 3]
## velocity
vel[k - 1] = X[1]
vel[k + 2] = s_X[1]
## calculate the offset amplitudes
if noffset > 0:
offsets_values[:, k] = X[n_default_unknown:(n_default_unknown +
noffset)]
offsets_values[:,
k + 3] = s_X[n_default_unknown:(n_default_unknown +
noffset)]
if option == 'detrend_annual':
## calculate the annual motion: amplitude & phase
annual.append((X[2], X[3]))
if option == 'detrend_seasonal':
## calculate the annual motion: amplitude & phase
annual.append((X[2], X[3]))
## calculate the annual motion: amplitude & phase
semi_annual.append((X[4], X[5]))
if len(annual) > 0:
annual = np.hstack(annual)
else:
annual = None
if len(semi_annual) > 0:
semi_annual = np.hstack(semi_annual)
else:
semi_annual = None
new_Gts = self.copy()
new_Gts.offsets_values = offsets_values
new_Gts.annual = annual
new_Gts.semi_annual = semi_annual
new_Gts.velocity = vel
new_Gts.data = residuals
if in_place:
self = new_Gts.copy()
del new_Gts
return (self)
else:
return (new_Gts)
def detrend_seasonal_median(self, wl=11, in_place=False, verbose=False):
###############################################################################
"""
Calculates a velocity using the median of pair of displacements exactly separated by one year, inspired from MIDAS and then removes repeating yearly signal
If the time series has less than three years of data, then the time series is kept untouched.
"""
import numpy as np
from pyacs.gts.Gts import Gts
import inspect
# after this method .data and .data_xyz are not consistent so .data_xyz is set to None
self.data_xyz = None
###########################################################################
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
###########################################################################
import pyacs.lib.astrotime
duration_in_year = self.data[-1, 0] - self.data[0, 0]
if duration_in_year >= 3.:
H_year = {}
# creates H_year arrays
for year in np.arange(int(self.data[0, 0]), int(self.data[-1, 0]) + 1):
H_year[year] = np.zeros((365, 3))
H_year[year][:, :] = np.nan
# deal with dates
np_mjd = pyacs.lib.astrotime.decyear2mjd(self.data[:, 0])
(np_doy, np_year, _np_ut) = pyacs.lib.astrotime.mjd2dayno(np_mjd)
np_doy = np_doy.astype(int)
np_year = np_year.astype(int)
# fill H_year arrays
for i in np.arange(np_doy.shape[0]):
if np_doy[i] != 366:
H_year[np_year[i]][np_doy[i] - 1, :] = self.data[i, 1:4]
# stack the velocities
np_vel = np.zeros((365 * (len(H_year) - 1), 3))
np_vel[:, :] = np.nan
i = 0
for year in sorted(H_year.keys())[0:-1]:
np_vel[i * 365:(i + 1) * 365] = H_year[year + 1] - H_year[year]
i = i + 1
# calculates the median velocity
med_vel = np.nanmedian(np_vel, axis=0)
# return detrended time series
detrended = self.remove_velocity(med_vel)
H_year = {}
# creates H_year arrays
for year in np.arange(int(detrended.data[0, 0]),
int(detrended.data[-1, 0]) + 1):
H_year[year] = np.zeros((365, 3))
H_year[year][:, :] = np.nan
# deal with dates
np_mjd = pyacs.lib.astrotime.decyear2mjd(detrended.data[:, 0])
(np_doy, np_year, _np_ut) = pyacs.lib.astrotime.mjd2dayno(np_mjd)
np_doy = np_doy.astype(int)
np_year = np_year.astype(int)
# fill H_year arrays
for i in np.arange(np_doy.shape[0]):
if np_doy[i] != 366:
H_year[np_year[i]][np_doy[i] - 1, :] = detrended.data[i, 1:4]
# center all H_year arrays
for year in sorted(H_year.keys()):
H_year[year] = H_year[year] - np.nanmedian(H_year[year], axis=0)
# plt.plot(H_year[year][:,1])
# create the median daily signal
A = np.array(list(H_year.values()))
np_doy_median_signal = np.nanmedian(A[:, :, :], axis=0)
# plt.plot(np_doy_median_signal[:,1],'ro')
# run a median filter on it
np_doy_median_signal_3 = np.vstack(
(np_doy_median_signal, np_doy_median_signal, np_doy_median_signal))
import scipy.signal
np_doy_median_signal[:, 0] = scipy.signal.medfilt(
np_doy_median_signal_3[:, 0], wl)[365:2 * 365]
np_doy_median_signal[:, 1] = scipy.signal.medfilt(
np_doy_median_signal_3[:, 1], wl)[365:2 * 365]
np_doy_median_signal[:, 2] = scipy.signal.medfilt(
np_doy_median_signal_3[:, 2], wl)[365:2 * 365]
# plt.plot(np_doy_median_signal[:,1],'bo')
# remove it from the detrended time series
detrended_seasonal = detrended.copy()
# loop on np_doy
for i in np.arange(detrended_seasonal.data.shape[0]):
if np_doy[i] != 366:
detrended_seasonal.data[i, 1:4] = detrended_seasonal.data[
i, 1:4] - np_doy_median_signal[np_doy[i] - 1, :]
else:
# time series is shorter than minimum
detrended_seasonal = self.copy()
return (detrended_seasonal)
def cdata(self, data=False, data_xyz=False, tol=0.001, verbose=False):
###############################################################################
"""
Check data/data_xyz attributes
:param data: boolean, if True, data attribute will be checked
:param data_xyz: boolean, if True, data_xyz attribute will be checked
:param tol: tolerance in days for two dates to be considered as the same (default 0.001 of day)
:param verbose: boolean, verbose mode
:return : boolean, True if everything is OK, False otherwise
:note : in future, this routine should also whether .data and .data_xyz value are consistent
"""
# import
import inspect
import numpy as np
from pyacs.gts.lib.errors import GtsInputDataNone, GtsInputData_xyzNone, GtsInputData_allNone
from pyacs.gts.lib.errors import GtsInputDataTypeError, GtsInputDataBadDim, GtsInputDataBadNcolumns, GtsInputDataBadNrows
from pyacs.gts.lib.errors import GtsInputData_xyzTypeError, GtsInputData_xyzBadDim, GtsInputData_xyzBadNcolumns, GtsInputData_xyzBadNrows
from pyacs.gts.lib.errors import GtsInputDataDiffShape, GtsInputDataDiffDate
# CASE EITHER data OR data_xyz where asked to be checked
# is there the required data
try:
if (self.data is None) and (data):
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (False)
# is there the required data_xyz
try:
if (self.data_xyz is None) and (data_xyz):
# raise exception
raise GtsInputData_xyzNone(inspect.stack()[0][3], __name__, self)
except GtsInputData_xyzNone as error:
# print PYACS WARNING
print(error)
return (False)
# GENERAL CASE
# is there some data?
try:
if (self.data is None) and (self.data_xyz is None):
# raise exception
raise GtsInputData_allNone(inspect.stack()[0][3], __name__, self)
except GtsInputData_allNone as error:
# print PYACS WARNING
print(error)
return (False)
# .data case
if self.data is not None:
# .data numpy array?
try:
if not isinstance(self.data, np.ndarray):
# raise exception
raise GtsInputDataTypeError(inspect.stack()[0][3], __name__,
self)
except GtsInputDataTypeError as error:
# print PYACS WARNING
print(error)
return (False)
# .data with right dimensions?
try:
if self.data.ndim != 2:
# raise exception
raise GtsInputDataBadDim(inspect.stack()[0][3], __name__, self)
except GtsInputDataBadDim as error:
# print PYACS WARNING
print(error)
return (False)
# .data with the right shape?
try:
if self.data.shape[1] not in [7, 10]:
# raise exception
raise GtsInputDataBadNcolumns(inspect.stack()[0][3], __name__,
self)
except GtsInputDataBadNcolumns as error:
# print PYACS WARNING
print(error)
return (False)
# .data with at least one observation?
try:
if self.data.shape[0] < 1:
# raise exception
raise GtsInputDataBadNrows(inspect.stack()[0][3], __name__,
self)
except GtsInputDataBadNrows as error:
# print PYACS WARNING
print(error)
return (False)
# .data_xyz case
if self.data_xyz is not None:
# .data numpy array?
try:
if not isinstance(self.data_xyz, np.ndarray):
# raise exception
raise GtsInputData_xyzTypeError(inspect.stack()[0][3],
__name__, self)
except GtsInputData_xyzTypeError as error:
# print PYACS WARNING
print(error)
return (False)
# .data with right dimensions?
try:
if self.data_xyz.ndim != 2:
# raise exception
raise GtsInputData_xyzBadDim(inspect.stack()[0][3], __name__,
self)
except GtsInputData_xyzBadDim as error:
# print PYACS WARNING
print(error)
return (False)
try:
if self.data_xyz.shape[1] not in [7, 10]:
# raise exception
raise GtsInputData_xyzBadNcolumns(inspect.stack()[0][3],
__name__, self)
except GtsInputData_xyzBadNcolumns as error:
# print PYACS WARNING
print(error)
return (False)
try:
if self.data_xyz.shape[0] < 1:
# raise exception
raise GtsInputData_xyzBadNrows(inspect.stack()[0][3], __name__,
self)
except GtsInputData_xyzBadNrows as error:
# print PYACS WARNING
print(error)
return (False)
# check .data and .data_xyz consistency
if (self.data is not None) and (self.data_xyz is not None):
# check both have the same length
try:
if self.data.shape[0] != self.data_xyz.shape[0]:
raise GtsInputDataDiffShape(inspect.stack()[0][3], __name__,
self)
except GtsInputDataDiffShape as error:
# print PYACS WARNING
print(error)
return (False)
# check both have the same dates
try:
if np.max(
(self.data[:, 0] - self.data_xyz[:, 0])**2) > tol / 365.25:
raise GtsInputDataDiffDate(inspect.stack()[0][3], __name__,
self)
except GtsInputDataDiffDate as error:
# print PYACS WARNING
print(error)
return (False)
return (True)
def set_zero_at_date(self, date, offset=None, in_place=False):
###################################################################
"""
make a translation of a time series, setting to 0 at a given date
if the provided date does not exist, uses the next date available
:param date: date in decimal year
:param offset: an offset (in mm) to be added. Could be a float, a list or 1D numpy array with 3 elements
"""
# import
import inspect
import numpy as np
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
if offset is None:
[cn, ce, cu] = [0.0, 0.0, 0.0]
else:
if isinstance(offset, float):
[cn, ce, cu] = [offset / 1000., offset / 1000., offset / 1000.]
elif isinstance(offset, list):
[cn, ce,
cu] = [offset[0] / 1000., offset[1] / 1000., offset[2] / 1000.]
elif isinstance(offset, np.ndarray):
[cn, ce,
cu] = [offset[0] / 1000., offset[1] / 1000., offset[2] / 1000.]
new_data = self.data.copy()
lindex = np.where(new_data[:, 0] >= date)
try:
index = lindex[0][0]
except:
print("!!! bad date provided")
return ()
print("-- Removing ", new_data[index, 1], new_data[index, 2],
new_data[index, 3])
new_data[:, 1] = new_data[:, 1] - new_data[index, 1] + cn
new_data[:, 2] = new_data[:, 2] - new_data[index, 2] + ce
new_data[:, 3] = new_data[:, 3] - new_data[index, 3] + cu
new_Gts = self.copy(data=new_data)
# .data_xyz set to None
new_Gts.data_xyz = None
# origin XYZ would need to be changed
if in_place:
self.data = new_Gts.data.copy()
return (new_Gts)
def detrend(self, method='L2', in_place=False, periods=[], exclude_periods=[]):
###################################################################
"""
detrends a time series and save velocity estimates in velocity attribute
:param periods : periods used to estimate the velocity
:param exclude_periods : periods to be excluded for the velocity estimate
:param in_place : if True then replace the current time series
:return : the detrended time series
:note : outliers from Gts.outliers are omitted in the estimation and
offsets given Gts.offsets_dates are estimated simultaneously
"""
import numpy as np
from pyacs.gts.Gts import Gts
import inspect
# after this method .data and .data_xyz are not consistent so .data_xyz is set to None
self.data_xyz = None
###########################################################################
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
###########################################################################
import copy
outliers = copy.deepcopy(self.outliers)
tmp_ts = self.remove_outliers()
if periods != []:
tmp_ts = tmp_ts.extract_periods(periods)
if exclude_periods != []:
tmp_ts = tmp_ts.exclude_periods(periods)
detrended = tmp_ts.make_model(option='detrend', method=method)
vel = detrended.velocity
offsets_values = detrended.offsets_values
new_gts = self.copy()
new_gts.outliers = outliers
new_gts.offsets_values = offsets_values
new_gts.velocity = vel
model = new_gts.mmodel()
new_gts.data[:, 1:4] = new_gts.data[:, 1:4] - model.data[:, 1:4]
if in_place:
self.data = new_gts.data
return (self)
else:
return (new_gts)
def smooth(self,
window_len=11,
window='hanning',
in_place=False,
verbose=False,
component='NEU'):
"""
smooth a time series
"""
import numpy as np
from pyacs.gts.Gts import Gts
import inspect
###########################################################################
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return (self)
###########################################################################
###################################################################
## smoothing routines from http://wiki.scipy.org/Cookbook/SignalSmooth
# changes numpy to np # JMN 18/07/2014
###################################################################
def smooth_scipy(x, window_len=11, window='hanning'):
"""smooth the data using a window with requested size.
This method is based on the convolution of a scaled window with the signal.
The signal is prepared by introducing reflected copies of the signal
(with the window size) in both ends so that transient parts are minimized
in the begining and end part of the output signal.
input:
x: the input signal
window_len: the dimension of the smoothing window; should be an odd integer
window: the type of window from 'flat', 'hanning', 'hamming', 'bartlett', 'blackman'
flat window will produce a moving average smoothing.
output:
the smoothed signal
example:
t=linspace(-2,2,0.1)
x=sin(t)+randn(len(t))*0.1
y=smooth(x)
see also:
numpy.hanning, numpy.hamming, numpy.bartlett, numpy.blackman, numpy.convolve
scipy.signal.lfilter
TODO: the window parameter could be the window itself if an array instead of a string
NOTE: length(output) != length(input), to correct this: return y[(window_len/2-1):-(window_len/2)] instead of just y.
"""
if x.ndim != 1:
raise ValueError("smooth only accepts 1 dimension arrays.")
if x.size < window_len:
raise ValueError(
"Input vector needs to be bigger than window size.")
if window_len < 3:
return x
if not window in [
'flat', 'hanning', 'hamming', 'bartlett', 'blackman'
]:
raise ValueError(
"Window is on of 'flat', 'hanning', 'hamming', 'bartlett', 'blackman'"
)
s = np.r_[x[window_len - 1:0:-1], x, x[-1:-window_len:-1]]
# print(len(s))
if window == 'flat': # moving average
w = np.ones(window_len, 'd')
else:
w = eval('np.' + window + '(window_len)')
y = np.convolve(w / w.sum(), s, mode='valid')
return y
new_east = smooth_scipy(self.data[:, 1],
window_len=window_len,
window=window)
new_north = smooth_scipy(self.data[:, 2],
window_len=window_len,
window=window)
new_up = smooth_scipy(self.data[:, 3],
window_len=window_len,
window=window)
new_Gts = self.copy()
if in_place:
return (self)
del new_Gts
else:
new_Gts.data[:, 1] = new_east[window_len // 2 -
1:new_Gts.data[:, 1].shape[0] +
window_len // 2 - 1]
new_Gts.data[:, 2] = new_north[window_len // 2 -
1:new_Gts.data[:, 1].shape[0] +
window_len // 2 - 1]
new_Gts.data[:, 3] = new_up[window_len // 2 -
1:new_Gts.data[:, 1].shape[0] +
window_len // 2 - 1]
return (new_Gts)
def displacement(self,
sdate=None,
edate=None,
window=None,
method='median',
speriod=[],
eperiod=[],
rounding='day',
verbose=True):
###################################################################
"""
Calculates displacements between two dates or two periods
:param sdate: start date in decimal year
:param edate: start date in decimal year
:param window: time window in days for searching available dates
:param method: method to calculate the position. 'median' or 'mean'. default is 'median'.
:param speriod: period for calculating the start position
:param eperiod: period for calculating the end position
:param rounding: rounding for dates. Choose among 'day','hour','minute' or 'second'. default is 'day'.
:param verbose: verbose mode
:return: displacement as np.array([dn,de,du,sdn,sde,sdu])
"""
# import
import inspect
import numpy as np
import pyacs.lib.astrotime as at
from colors import red
# check consistency of passed arguments
if (sdate is not None) and (speriod != []):
print(red("[ERROR] provide sdate or speriod not both"))
if (edate is not None) and (eperiod != []):
print(red("[ERROR] provide edate or eperiod not both"))
# check data is not None
from pyacs.gts.lib.errors import GtsInputDataNone
try:
if self.data is None:
# raise exception
raise GtsInputDataNone(inspect.stack()[0][3], __name__, self)
except GtsInputDataNone as error:
# print PYACS WARNING
print(error)
return ()
# convert dates to seconds
np_seconds = at.decyear2seconds(self.data[:, 0], rounding=rounding)
# convert window to seconds
if window is not None:
window_s = int(window * 86400.)
else:
window_s = 1
# fills speriod and eperiod
if speriod == []:
speriod = [
at.decyear2seconds(sdate, rounding=rounding) - window_s,
at.decyear2seconds(sdate, rounding=rounding) + window_s
]
else:
speriod = [
at.decyear2seconds(speriod[0], rounding=rounding) - window_s,
at.decyear2seconds(speriod[1], rounding=rounding) + window_s
]
if eperiod == []:
eperiod = [
at.decyear2seconds(edate, rounding=rounding) - window_s,
at.decyear2seconds(edate, rounding=rounding) + window_s
]
else:
eperiod = [
at.decyear2seconds(eperiod[0], rounding=rounding) - window_s,
at.decyear2seconds(eperiod[1], rounding=rounding) + window_s
]
# get values
lindex = np.argwhere((np_seconds >= speriod[0])
& (np_seconds <= speriod[1]))
np_start = self.data[lindex, 1:7].reshape(-1, 6)
lindex = np.argwhere((np_seconds >= eperiod[0])
& (np_seconds <= eperiod[1]))
np_end = self.data[lindex, 1:7].reshape(-1, 6)
if method == 'median':
pos_sdate = np.median(np_start, axis=0)
pos_edate = np.median(np_end, axis=0)
else:
pos_sdate = np.mean(np_start, axis=0)
pos_edate = np.mean(np_end, axis=0)
disp = pos_edate[:3] - pos_sdate[:3]
std_disp = np.sqrt(pos_sdate**2 + pos_edate**2)[3:7]
displacement = [
disp[0], disp[1], disp[2], std_disp[0], std_disp[1], std_disp[2]
]
return (np.array(displacement))
