def pygeons_strain_view(xdiff_file,ydiff_file,map_resolution='i',**kwargs):
'''
runs the PyGeoNS Interactive Strain Viewer
Parameters
----------
xdiff_file : str
ydiff_file : str
map_resolution : str
basemap resolution
**kwargs :
gets passed to pygeons.strain.view
'''
logger.info('Running pygeons strain-view ...')
data_dx = dict_from_hdf5(xdiff_file)
data_dy = dict_from_hdf5(ydiff_file)
data_dx,data_dy = _common_context([data_dx,data_dy])
if ((data_dx['space_exponent'] != 0) |
(data_dy['space_exponent'] != 0)):
raise ValueError('The input datasets cannot have spatial units')
t = data_dx['time']
id = data_dx['id']
lon = data_dx['longitude']
lat = data_dx['latitude']
dates = [mjd_inv(ti,'%Y-%m-%d') for ti in t]
units = _unit_string(data_dx['space_exponent'],
data_dx['time_exponent'])
# factor that converts units of days and m to the units in *units*
conv = 1.0/unit_conversion(units,time='day',space='m')
exx = conv*data_dx['east']
eyy = conv*data_dy['north']
exy = 0.5*conv*(data_dx['north'] + data_dy['east'])
sxx = conv*data_dx['east_std_dev']
syy = conv*data_dy['north_std_dev']
sxy = 0.5*conv*np.sqrt(data_dx['north_std_dev']**2 + data_dy['east_std_dev']**2)
ts_fig,ts_ax = plt.subplots(3,1,sharex=True,num='Time Series View',
facecolor='white')
_setup_ts_ax(ts_ax)
map_fig,map_ax = plt.subplots(num='Map View',facecolor='white')
bm = make_basemap(lon,lat,resolution=map_resolution)
_setup_map_ax(bm,map_ax)
x,y = bm(lon,lat)
pos = np.array([x,y]).T
interactive_strain_viewer(
t,pos,exx,eyy,exy,sxx=sxx,syy=syy,sxy=sxy,
map_ax=map_ax,ts_ax=ts_ax,time_labels=dates,
station_labels=id,units=units,**kwargs)
return
def pygeons_merge(input_files, output_stem=None):
'''
Merge data files
'''
logger.info('Running pygeons merge ...')
data_list = [dict_from_hdf5(i) for i in input_files]
data_list = _common_context(data_list)
out = data_list[0]
for d in data_list[1:]:
for dir in ['east', 'north', 'vertical']:
# overwrite data in *out* with non-missing data in *d*
missing_in_d = np.isinf(d[dir + '_std_dev'])
missing_in_out = np.isinf(out[dir + '_std_dev'])
if np.any(~missing_in_d & ~missing_in_out):
warnings.warn(
'Data for some stations and times exist in multiple '
'datasets. Precedence is determined by the order the data '
'files were specified in.')
out[dir][~missing_in_d] = d[dir][~missing_in_d]
out[dir + '_std_dev'][~missing_in_d] = d[dir +
'_std_dev'][~missing_in_d]
# set output file name
if output_stem is None:
output_stem = 'merged'
output_file = output_stem + '.h5'
hdf5_from_dict(output_file, out)
logger.info('Merged data written to %s' % output_file)
def pygeons_merge(input_files,output_stem=None):
'''
Merge data files
'''
logger.info('Running pygeons merge ...')
data_list = [dict_from_hdf5(i) for i in input_files]
data_list = _common_context(data_list)
out = data_list[0]
for d in data_list[1:]:
for dir in ['east','north','vertical']:
# overwrite data in *out* with non-missing data in *d*
missing_in_d = np.isinf(d[dir+'_std_dev'])
missing_in_out = np.isinf(out[dir+'_std_dev'])
if np.any(~missing_in_d & ~missing_in_out):
warnings.warn(
'Data for some stations and times exist in multiple '
'datasets. Precedence is determined by the order the data '
'files were specified in.')
out[dir][~missing_in_d] = d[dir][~missing_in_d]
out[dir+'_std_dev'][~missing_in_d] = d[dir+'_std_dev'][~missing_in_d]
# set output file name
if output_stem is None:
output_stem = 'merged'
output_file = output_stem + '.h5'
hdf5_from_dict(output_file,out)
logger.info('Merged data written to %s' % output_file)
def pygeons_autoclean(input_file,
network_model=('spwen12-se',),
network_params=(1.0,0.1,100.0),
station_model=('linear',),
station_params=(),
output_stem=None,
outlier_tol=4.0):
'''
Remove outliers with a data editing algorithm
'''
logger.info('Running pygeons autoclean ...')
data = dict_from_hdf5(input_file)
if data['time_exponent'] != 0:
raise ValueError('input dataset must have units of displacement')
if data['space_exponent'] != 1:
raise ValueError('input dataset must have units of displacement')
# dictionary which will contain the edited data
out = dict((k,np.copy(v)) for k,v in data.iteritems())
# convert params to a dictionary of hyperparameters for each direction
network_params = _params_dict(network_params)
station_params = _params_dict(station_params)
# make output file name
if output_stem is None:
output_stem = _remove_extension(input_file) + '.autoclean'
output_file = output_stem + '.h5'
# convert geodetic positions to cartesian
bm = make_basemap(data['longitude'],data['latitude'])
x,y = bm(data['longitude'],data['latitude'])
xy = np.array([x,y]).T
_log_autoclean(input_file,
network_model,network_params,
station_model,station_params,
outlier_tol,
output_file)
for dir in ['east','north','vertical']:
de,sde = autoclean(t=data['time'][:,None],
x=xy,
d=data[dir],
sd=data[dir+'_std_dev'],
network_model=network_model,
network_params=network_params[dir],
station_model=station_model,
station_params=station_params[dir],
tol=outlier_tol)
out[dir] = de
out[dir+'_std_dev'] = sde
hdf5_from_dict(output_file,out)
logger.info('Edited data written to %s' % output_file)
return
def pygeons_fit(input_file,
network_model=('spwen12-se',),
network_params=(1.0,0.1,100.0),
station_model=('linear',),
station_params=(),
output_stem=None):
'''
Condition the Gaussian process to the observations and evaluate the
posterior at the observation points.
'''
logger.info('Running pygeons fit ...')
data = dict_from_hdf5(input_file)
if data['time_exponent'] != 0:
raise ValueError('input dataset must have units of displacement')
if data['space_exponent'] != 1:
raise ValueError('input dataset must have units of displacement')
# create output dictionary
out = dict((k,np.copy(v)) for k,v in data.iteritems())
# convert params to a dictionary of hyperparameters for each direction
network_params = _params_dict(network_params)
station_params = _params_dict(station_params)
# make output file name
if output_stem is None:
output_stem = _remove_extension(input_file) + '.fit'
output_file = output_stem + '.h5'
# convert geodetic positions to cartesian
bm = make_basemap(data['longitude'],data['latitude'])
x,y = bm(data['longitude'],data['latitude'])
xy = np.array([x,y]).T
_log_fit(input_file,
network_model,network_params,
station_model,station_params,
output_file)
for dir in ['east','north','vertical']:
u,su = fit(t=data['time'][:,None],
x=xy,
d=data[dir],
sd=data[dir+'_std_dev'],
network_model=network_model,
network_params=network_params[dir],
station_model=station_model,
station_params=station_params[dir])
out[dir] = u
out[dir+'_std_dev'] = su
hdf5_from_dict(output_file,out)
logger.info('Posterior fit written to %s' % output_file)
return
def pygeons_vector_view(input_files,map_resolution='i',**kwargs):
'''
runs the PyGeoNS interactive vector viewer
Parameters
----------
data_list : (N,) list of dicts
list of data dictionaries being plotted
map_resolution : str
basemap resolution
**kwargs :
gets passed to pygeons.plot.view.interactive_view
'''
logger.info('Running pygeons vector-view ...')
data_list = [dict_from_hdf5(i) for i in input_files]
data_list = _common_context(data_list)
# use filenames for dataset labels if none were provided
dataset_labels = kwargs.pop('dataset_labels',input_files)
t = data_list[0]['time']
lon = data_list[0]['longitude']
lat = data_list[0]['latitude']
id = data_list[0]['id']
dates = [mjd_inv(ti,'%Y-%m-%d') for ti in t]
units = _unit_string(data_list[0]['space_exponent'],
data_list[0]['time_exponent'])
# factor that converts units of days and m to the units in *units*
conv = 1.0/unit_conversion(units,time='day',space='m')
u = [conv*d['east'] for d in data_list]
v = [conv*d['north'] for d in data_list]
z = [conv*d['vertical'] for d in data_list]
su = [conv*d['east_std_dev'] for d in data_list]
sv = [conv*d['north_std_dev'] for d in data_list]
sz = [conv*d['vertical_std_dev'] for d in data_list]
ts_fig,ts_ax = plt.subplots(3,1,sharex=True,num='Time Series View',
facecolor='white')
_setup_ts_ax(ts_ax)
map_fig,map_ax = plt.subplots(num='Map View',facecolor='white')
bm = make_basemap(lon,lat,resolution=map_resolution)
_setup_map_ax(bm,map_ax)
x,y = bm(lon,lat)
pos = np.array([x,y]).T
interactive_vector_viewer(
t,pos,u=u,v=v,z=z,su=su,sv=sv,sz=sz,
ts_ax=ts_ax,map_ax=map_ax,
dataset_labels=dataset_labels,
station_labels=id,time_labels=dates,
units=units,**kwargs)
return
def pygeons_totext(input_file, output_stem=None):
'''
converts an hdf5 file to a text file
'''
logger.info('Running pygeons totext ...')
data = dict_from_hdf5(input_file)
if output_stem is None:
output_stem = _remove_extension(input_file)
output_file = output_stem + '.csv'
text_from_dict(output_file, data)
logger.info('Data written to %s' % output_file)
return
def pygeons_totext(input_file,output_stem=None):
'''
converts an hdf5 file to a text file
'''
logger.info('Running pygeons totext ...')
data = dict_from_hdf5(input_file)
if output_stem is None:
output_stem = _remove_extension(input_file)
output_file = output_stem + '.csv'
text_from_dict(output_file,data)
logger.info('Data written to %s' % output_file)
return
def pygeons_info(input_file):
'''
prints metadata
'''
logger.info('Running pygeons info ...')
data_dict = dict_from_hdf5(input_file)
# put together info string
units = _unit_string(data_dict['space_exponent'],
data_dict['time_exponent'])
stations = str(len(data_dict['id']))
times = str(len(data_dict['time']))
observations = (np.sum(~np.isinf(data_dict['east_std_dev'])) +
np.sum(~np.isinf(data_dict['north_std_dev'])) +
np.sum(~np.isinf(data_dict['vertical_std_dev'])))
time_range = '%s, %s' % (mjd.mjd_inv(data_dict['time'][0], '%Y-%m-%d'),
mjd.mjd_inv(data_dict['time'][-1], '%Y-%m-%d'))
lon_range = '%s, %s' % (np.min(
data_dict['longitude']), np.max(data_dict['longitude']))
lat_range = '%s, %s' % (np.min(
data_dict['latitude']), np.max(data_dict['latitude']))
# split names into groups of no more than 8
station_name_list = list(data_dict['id'])
station_name_groups = []
while len(station_name_list) > 0:
station_name_groups += [', '.join(station_name_list[:7])]
station_name_list = station_name_list[7:]
msg = '\n'
msg += '------------------ PYGEONS DATA INFORMATION ------------------\n\n'
msg += 'file : %s\n' % input_file
msg += 'units : %s\n' % units
msg += 'stations : %s\n' % stations
msg += 'times : %s\n' % times
msg += 'observations : %s\n' % observations
msg += 'time range : %s\n' % time_range
msg += 'longitude range : %s\n' % lon_range
msg += 'latitude range : %s\n' % lat_range
msg += 'station names : %s\n' % station_name_groups[0]
for g in station_name_groups[1:]:
msg += ' %s\n' % g
msg += '\n'
msg += '--------------------------------------------------------------'
print(msg)
return
def pygeons_info(input_file):
'''
prints metadata
'''
logger.info('Running pygeons info ...')
data_dict = dict_from_hdf5(input_file)
# put together info string
units = _unit_string(data_dict['space_exponent'],
data_dict['time_exponent'])
stations = str(len(data_dict['id']))
times = str(len(data_dict['time']))
observations = (np.sum(~np.isinf(data_dict['east_std_dev'])) +
np.sum(~np.isinf(data_dict['north_std_dev'])) +
np.sum(~np.isinf(data_dict['vertical_std_dev'])))
time_range = '%s, %s' % (mjd.mjd_inv(data_dict['time'][0],'%Y-%m-%d'),
mjd.mjd_inv(data_dict['time'][-1],'%Y-%m-%d'))
lon_range = '%s, %s' % (np.min(data_dict['longitude']),
np.max(data_dict['longitude']))
lat_range = '%s, %s' % (np.min(data_dict['latitude']),
np.max(data_dict['latitude']))
# split names into groups of no more than 8
station_name_list = list(data_dict['id'])
station_name_groups = []
while len(station_name_list) > 0:
station_name_groups += [', '.join(station_name_list[:7])]
station_name_list = station_name_list[7:]
msg = '\n'
msg += '------------------ PYGEONS DATA INFORMATION ------------------\n\n'
msg += 'file : %s\n' % input_file
msg += 'units : %s\n' % units
msg += 'stations : %s\n' % stations
msg += 'times : %s\n' % times
msg += 'observations : %s\n' % observations
msg += 'time range : %s\n' % time_range
msg += 'longitude range : %s\n' % lon_range
msg += 'latitude range : %s\n' % lat_range
msg += 'station names : %s\n' % station_name_groups[0]
for g in station_name_groups[1:]:
msg += ' %s\n' % g
msg += '\n'
msg += '--------------------------------------------------------------'
print(msg)
return
def pygeons_crop(input_file,
start_date=None,
stop_date=None,
min_lat=-np.inf,
max_lat=np.inf,
min_lon=-np.inf,
max_lon=np.inf,
stations=None,
output_stem=None):
'''
Sets the time span of the data set to be between *start_date* and
*stop_date*. Sets the stations to be within the latitude and
longitude bounds.
Parameters
----------
data : dict
data dictionary
start_date : str, optional
start date of output data set in YYYY-MM-DD. Uses the start date
of *data* if not provided. Defaults to the earliest date.
stop_date : str, optional
Stop date of output data set in YYYY-MM-DD. Uses the stop date
of *data* if not provided. Defaults to the latest date.
min_lon, max_lon, min_lat, max_lat : float, optional
Spatial bounds on the output data set
stations : str list, optional
List of stations to be removed from the dataset. This is in
addition to the station removed by the lon/lat bounds.
Returns
-------
out_dict : dict
output data dictionary
'''
logger.info('Running pygeons crop ...')
data = dict_from_hdf5(input_file)
out = dict((k, np.copy(v)) for k, v in data.iteritems())
if start_date is None:
start_date = mjd.mjd_inv(data['time'].min(), '%Y-%m-%d')
if stop_date is None:
stop_date = mjd.mjd_inv(data['time'].max(), '%Y-%m-%d')
if stations is None:
stations = []
# remove times that are not within the bounds of *start_date* and
# *stop_date*
start_time = int(mjd.mjd(start_date, '%Y-%m-%d'))
stop_time = int(mjd.mjd(stop_date, '%Y-%m-%d'))
idx = ((data['time'] >= start_time) & (data['time'] <= stop_time))
out['time'] = out['time'][idx]
for dir in ['east', 'north', 'vertical']:
out[dir] = out[dir][idx, :]
out[dir + '_std_dev'] = out[dir + '_std_dev'][idx, :]
# find stations that are within the bounds
in_bounds = ((data['longitude'] > min_lon) & (data['longitude'] < max_lon)
& (data['latitude'] > min_lat) & (data['latitude'] < max_lat))
# find stations that are in the list of stations to be removed
in_list = np.array([i in stations for i in data['id']])
# keep stations that are in bounds and not in the list
idx, = (in_bounds & ~in_list).nonzero()
out['id'] = out['id'][idx]
out['longitude'] = out['longitude'][idx]
out['latitude'] = out['latitude'][idx]
for dir in ['east', 'north', 'vertical']:
out[dir] = out[dir][:, idx]
out[dir + '_std_dev'] = out[dir + '_std_dev'][:, idx]
# set output file name
if output_stem is None:
output_stem = _remove_extension(input_file) + '.crop'
output_file = output_stem + '.h5'
hdf5_from_dict(output_file, out)
logger.info('Cropped data written to %s' % output_file)
return
def pygeons_strain(input_file,
network_prior_model=('spwen12-se',),
network_prior_params=(1.0,0.1,100.0),
network_noise_model=(),
network_noise_params=(),
station_noise_model=('linear',),
station_noise_params=(),
start_date=None,stop_date=None,
positions=None,positions_file=None,
rate=True,vertical=True,covariance=False,
output_stem=None):
'''
calculates strain
'''
logger.info('Running pygeons strain ...')
data = dict_from_hdf5(input_file)
if data['time_exponent'] != 0:
raise ValueError('input dataset must have units of displacement')
if data['space_exponent'] != 1:
raise ValueError('input dataset must have units of displacement')
out_dx = dict((k,np.copy(v)) for k,v in data.iteritems())
out_dy = dict((k,np.copy(v)) for k,v in data.iteritems())
# convert params to a dictionary of hyperparameters for each direction
network_prior_params = _params_dict(network_prior_params)
network_noise_params = _params_dict(network_noise_params)
station_noise_params = _params_dict(station_noise_params)
# convert geodetic input positions to cartesian
bm = make_basemap(data['longitude'],data['latitude'])
x,y = bm(data['longitude'],data['latitude'])
xy = np.array([x,y]).T
# set output positions
if (positions is None) & (positions_file is None):
# no output positions were specified so return the solution at the
# input data positions
output_id = np.array(data['id'],copy=True)
output_lon = np.array(data['longitude'],copy=True)
output_lat = np.array(data['latitude'],copy=True)
else:
output_id = np.zeros((0,),dtype=str)
output_lon = np.zeros((0,),dtype=float)
output_lat = np.zeros((0,),dtype=float)
if positions_file is not None:
# if positions file was specified
pos = np.loadtxt(positions_file,dtype=str,ndmin=2)
if pos.shape[1] != 3:
raise ValueError(
'positions file must contain a column for IDs, longitudes, '
'and latitudes')
output_id = np.hstack((output_id,pos[:,0]))
output_lon = np.hstack((output_lon,pos[:,1].astype(float)))
output_lat = np.hstack((output_lat,pos[:,2].astype(float)))
if positions is not None:
# if positions were specified via the command line
pos = np.array(positions,dtype=str).reshape((-1,3))
output_id = np.hstack((output_id,pos[:,0]))
output_lon = np.hstack((output_lon,pos[:,1].astype(float)))
output_lat = np.hstack((output_lat,pos[:,2].astype(float)))
# convert geodetic output positions to cartesian
output_x,output_y = bm(output_lon,output_lat)
output_xy = np.array([output_x,output_y]).T
# set output times
if start_date is None:
start_date = mjd_inv(np.min(data['time']),'%Y-%m-%d')
if stop_date is None:
stop_date = mjd_inv(np.max(data['time']),'%Y-%m-%d')
start_time = mjd(start_date,'%Y-%m-%d')
stop_time = mjd(stop_date,'%Y-%m-%d')
output_time = np.arange(start_time,stop_time+1)
# set output file names
if output_stem is None:
output_stem = _remove_extension(input_file) + '.strain'
output_dx_file = output_stem + '.dudx.h5'
output_dy_file = output_stem + '.dudy.h5'
_log_strain(input_file,
network_prior_model,network_prior_params,
network_noise_model,network_noise_params,
station_noise_model,station_noise_params,
start_date,stop_date,output_id,rate,vertical,
covariance,output_dx_file,output_dy_file)
for dir in ['east','north','vertical']:
if (dir == 'vertical') & (not vertical):
logger.debug('Not computing vertical deformation gradients')
# do not compute the deformation gradients for vertical. Just
# return zeros.
dx = np.zeros((output_time.shape[0],output_xy.shape[0]))
sdx = np.zeros((output_time.shape[0],output_xy.shape[0]))
dy = np.zeros((output_time.shape[0],output_xy.shape[0]))
sdy = np.zeros((output_time.shape[0],output_xy.shape[0]))
if covariance:
# if covariance is True then create an empty array of
# covariances
cdx = np.zeros((output_time.shape[0],output_xy.shape[0],
output_time.shape[0],output_xy.shape[0]))
cdy = np.zeros((output_time.shape[0],output_xy.shape[0],
output_time.shape[0],output_xy.shape[0]))
soln = (dx,sdx,cdx,dy,sdy,cdy)
else:
soln = (dx,sdx,dy,sdy)
else:
soln = strain(t=data['time'][:,None],
x=xy,
d=data[dir],
sd=data[dir+'_std_dev'],
network_prior_model=network_prior_model,
network_prior_params=network_prior_params[dir],
network_noise_model=network_noise_model,
network_noise_params=network_noise_params[dir],
station_noise_model=station_noise_model,
station_noise_params=station_noise_params[dir],
out_t=output_time[:,None],
out_x=output_xy,
rate=rate,
covariance=covariance)
if covariance:
# soln contains six entries when covariance is True
dx,sdx,cdx,dy,sdy,cdy = soln
out_dx[dir] = dx
out_dx[dir+'_std_dev'] = sdx
out_dx[dir+'_covariance'] = cdx
out_dy[dir] = dy
out_dy[dir+'_std_dev'] = sdy
out_dy[dir+'_covariance'] = cdy
else:
# soln contains four entries when covariance is False
dx,sdx,dy,sdy = soln
out_dx[dir] = dx
out_dx[dir+'_std_dev'] = sdx
out_dy[dir] = dy
out_dy[dir+'_std_dev'] = sdy
out_dx['time'] = output_time
out_dx['longitude'] = output_lon
out_dx['latitude'] = output_lat
out_dx['id'] = output_id
out_dx['time_exponent'] = -int(rate)
out_dx['space_exponent'] = 0
out_dy['time'] = output_time
out_dy['longitude'] = output_lon
out_dy['latitude'] = output_lat
out_dy['id'] = output_id
out_dy['time_exponent'] = -int(rate)
out_dy['space_exponent'] = 0
hdf5_from_dict(output_dx_file,out_dx)
hdf5_from_dict(output_dy_file,out_dy)
if rate:
logger.info('Posterior velocity gradients written to %s and %s' % (output_dx_file,output_dy_file))
else:
logger.info('Posterior displacement gradients written to %s and %s' % (output_dx_file,output_dy_file))
return
def pygeons_reml(input_file,
network_model=('spwen12-se',),
network_params=(1.0,0.1,100.0),
network_fix=(),
station_model=('linear',),
station_params=(),
station_fix=(),
output_stem=None):
'''
Restricted maximum likelihood estimation
'''
logger.info('Running pygeons reml ...')
data = dict_from_hdf5(input_file)
if data['time_exponent'] != 0:
raise ValueError('input dataset must have units of displacement')
if data['space_exponent'] != 1:
raise ValueError('input dataset must have units of displacement')
# convert params to a dictionary of hyperparameters for each direction
network_params = _params_dict(network_params)
network_fix = np.asarray(network_fix,dtype=int)
station_params = _params_dict(station_params)
station_fix = np.asarray(station_fix,dtype=int)
# make output file name
if output_stem is None:
output_stem = _remove_extension(input_file) + '.reml'
output_file = output_stem + '.txt'
# convert geodetic positions to cartesian
bm = make_basemap(data['longitude'],data['latitude'])
x,y = bm(data['longitude'],data['latitude'])
xy = np.array([x,y]).T
# call "pygeons info" on the input data file. pipe the results to
# the output file
sp.call('pygeons info %s > %s' % (input_file,output_file),shell=True)
msg = _log_reml(input_file,
network_model,network_params,network_fix,
station_model,station_params,station_fix,
output_file)
# write log entry to file
with open(output_file,'a') as fout:
fout.write(msg)
# make a dictionary storing likelihoods
likelihood = {}
for dir in ['east','north','vertical']:
net_opt,sta_opt,like = reml(t=data['time'][:,None],
x=xy,
d=data[dir],
sd=data[dir+'_std_dev'],
network_model=network_model,
network_params=network_params[dir],
network_fix=network_fix,
station_model=station_model,
station_params=station_params[dir],
station_fix=station_fix)
# update the parameter dict with the optimal values
network_params[dir] = net_opt
station_params[dir] = sta_opt
likelihood[dir] = like
msg = _log_reml_results(input_file,
network_model,network_params,network_fix,
station_model,station_params,station_fix,
likelihood,output_file)
# write log entry to file
with open(output_file,'a') as fout:
fout.write(msg)
logger.info('Optimal parameters written to %s' % output_file)
return
def pygeons_clean(input_file,resolution='i',
input_edits_file=None,
break_lons=None,break_lats=None,
break_conn=None,no_display=False,
output_stem=None,**kwargs):
'''
runs the PyGeoNS Interactive Cleaner
Parameters
----------
data : dict
data dictionary
resolution : str
basemap resolution
input_edits_file : str
Name of the file containing edits which will automatically be
applied before opening up the interactive viewer.
output_edits_file : str
Name of the file where all edits will be recorded.
**kwargs :
gets passed to pygeons.clean.clean
Returns
-------
out : dict
output data dictionary
'''
logger.info('Running pygeons clean ...')
data = dict_from_hdf5(input_file)
out = dict((k,np.copy(v)) for k,v in data.iteritems())
ts_fig,ts_ax = plt.subplots(3,1,sharex=True,num='Time Series View',facecolor='white')
_setup_ts_ax(ts_ax)
map_fig,map_ax = plt.subplots(num='Map View',facecolor='white')
bm = make_basemap(data['longitude'],data['latitude'],resolution=resolution)
_setup_map_ax(bm,map_ax)
x,y = bm(data['longitude'],data['latitude'])
pos = np.array([x,y]).T
t = data['time']
dates = [mjd_inv(ti,'%Y-%m-%d') for ti in t]
units = _unit_string(data['space_exponent'],data['time_exponent'])
conv = 1.0/unit_conversion(units,time='day',space='m')
u = conv*data['east']
v = conv*data['north']
z = conv*data['vertical']
su = conv*data['east_std_dev']
sv = conv*data['north_std_dev']
sz = conv*data['vertical_std_dev']
ic = InteractiveCleaner(
t,pos,u=u,v=v,z=z,su=su,sv=sv,sz=sz,
map_ax=map_ax,ts_ax=ts_ax,
time_labels=dates,
units=units,
station_labels=data['id'],
**kwargs)
# make edits to the data set prior to displaying it
if input_edits_file is not None:
with open(input_edits_file,'r') as fin:
for line in fin:
# ignore blank lines
if line.isspace():
continue
type,sta,a,b = line.strip().split()
# set the current station in *ic* to the station for this edit
xidx, = (data['id'] == sta).nonzero()
if len(xidx) == 0:
# continue because the station does not exist in this
# dataset
continue
ic.xidx = xidx[0]
if type == 'outliers':
start_time = mjd(a,'%Y-%m-%d')
stop_time = mjd(b,'%Y-%m-%d')
ic.remove_outliers(start_time,stop_time)
elif type == 'jump':
jump_time = mjd(a,'%Y-%m-%d')
delta = int(b)
ic.remove_jump(jump_time,delta)
else:
raise ValueError('edit type must be either "outliers" or "jump"')
if not no_display:
ic.update()
ic.connect()
# set output file name
if output_stem is None:
output_stem = _remove_extension(input_file) + '.clean'
output_file = output_stem + '.h5'
output_edits_file = output_stem + '.txt'
with open(output_edits_file,'w') as fout:
for i in ic.log:
type,xidx,a,b = i
if type == 'outliers':
station = data['id'][xidx]
start_date = mjd_inv(a,'%Y-%m-%d')
stop_date = mjd_inv(b,'%Y-%m-%d')
fout.write('outliers %s %s %s\n' % (station,start_date,stop_date))
elif type == 'jump':
station = data['id'][xidx]
jump_date = mjd_inv(a,'%Y-%m-%d')
fout.write('jump %s %s %s\n' % (station,jump_date,b))
else:
raise ValueError('edit type must be either "outliers" or "jump"')
logger.info('Edits saved to %s' % output_edits_file)
clean_data = ic.get_data()
out['east'] = clean_data[0]/conv
out['north'] = clean_data[1]/conv
out['vertical'] = clean_data[2]/conv
out['east_std_dev'] = clean_data[3]/conv
out['north_std_dev'] = clean_data[4]/conv
out['vertical_std_dev'] = clean_data[5]/conv
hdf5_from_dict(output_file,out)
logger.info('Cleaned data written to %s' % output_file)
logger.info('Edits written to %s' % output_edits_file)
return
#!/usr/bin/env python
# this script removes high sigma values from an HDF5 file
import numpy as np
from pygeons.io.convert import dict_from_hdf5, hdf5_from_dict
import sys
fname = sys.argv[1]
cutoff = 0.02 # 2 cm cutoff
data = dict_from_hdf5(fname)
mask = ((data['north_std_dev'] > cutoff) | (data['east_std_dev'] > cutoff) |
(data['vertical_std_dev'] > cutoff))
data['north'][mask] = np.nan
data['north_std_dev'][mask] = np.inf
data['east'][mask] = np.nan
data['east_std_dev'][mask] = np.inf
data['vertical'][mask] = np.nan
data['vertical_std_dev'][mask] = np.inf
hdf5_from_dict(fname, data)
def pygeons_clean(input_file,
resolution='i',
input_edits_file=None,
break_lons=None,
break_lats=None,
break_conn=None,
no_display=False,
output_stem=None,
**kwargs):
'''
runs the PyGeoNS Interactive Cleaner
Parameters
----------
data : dict
data dictionary
resolution : str
basemap resolution
input_edits_file : str
Name of the file containing edits which will automatically be
applied before opening up the interactive viewer.
output_edits_file : str
Name of the file where all edits will be recorded.
**kwargs :
gets passed to pygeons.clean.clean
Returns
-------
out : dict
output data dictionary
'''
logger.info('Running pygeons clean ...')
data = dict_from_hdf5(input_file)
out = dict((k, np.copy(v)) for k, v in data.iteritems())
ts_fig, ts_ax = plt.subplots(3,
1,
sharex=True,
num='Time Series View',
facecolor='white')
_setup_ts_ax(ts_ax)
map_fig, map_ax = plt.subplots(num='Map View', facecolor='white')
bm = make_basemap(data['longitude'],
data['latitude'],
resolution=resolution)
_setup_map_ax(bm, map_ax)
x, y = bm(data['longitude'], data['latitude'])
pos = np.array([x, y]).T
t = data['time']
dates = [mjd_inv(ti, '%Y-%m-%d') for ti in t]
units = _unit_string(data['space_exponent'], data['time_exponent'])
conv = 1.0 / unit_conversion(units, time='day', space='m')
u = conv * data['east']
v = conv * data['north']
z = conv * data['vertical']
su = conv * data['east_std_dev']
sv = conv * data['north_std_dev']
sz = conv * data['vertical_std_dev']
ic = InteractiveCleaner(t,
pos,
u=u,
v=v,
z=z,
su=su,
sv=sv,
sz=sz,
map_ax=map_ax,
ts_ax=ts_ax,
time_labels=dates,
units=units,
station_labels=data['id'],
**kwargs)
# make edits to the data set prior to displaying it
if input_edits_file is not None:
with open(input_edits_file, 'r') as fin:
for line in fin:
# ignore blank lines
if line.isspace():
continue
type, sta, a, b = line.strip().split()
# set the current station in *ic* to the station for this edit
xidx, = (data['id'] == sta).nonzero()
if len(xidx) == 0:
# continue because the station does not exist in this
# dataset
continue
ic.xidx = xidx[0]
if type == 'outliers':
start_time = mjd(a, '%Y-%m-%d')
stop_time = mjd(b, '%Y-%m-%d')
ic.remove_outliers(start_time, stop_time)
elif type == 'jump':
jump_time = mjd(a, '%Y-%m-%d')
delta = int(b)
ic.remove_jump(jump_time, delta)
else:
raise ValueError(
'edit type must be either "outliers" or "jump"')
if not no_display:
ic.update()
ic.connect()
# set output file name
if output_stem is None:
output_stem = _remove_extension(input_file) + '.clean'
output_file = output_stem + '.h5'
output_edits_file = output_stem + '.txt'
with open(output_edits_file, 'w') as fout:
for i in ic.log:
type, xidx, a, b = i
if type == 'outliers':
station = data['id'][xidx]
start_date = mjd_inv(a, '%Y-%m-%d')
stop_date = mjd_inv(b, '%Y-%m-%d')
fout.write('outliers %s %s %s\n' %
(station, start_date, stop_date))
elif type == 'jump':
station = data['id'][xidx]
jump_date = mjd_inv(a, '%Y-%m-%d')
fout.write('jump %s %s %s\n' % (station, jump_date, b))
else:
raise ValueError(
'edit type must be either "outliers" or "jump"')
logger.info('Edits saved to %s' % output_edits_file)
clean_data = ic.get_data()
out['east'] = clean_data[0] / conv
out['north'] = clean_data[1] / conv
out['vertical'] = clean_data[2] / conv
out['east_std_dev'] = clean_data[3] / conv
out['north_std_dev'] = clean_data[4] / conv
out['vertical_std_dev'] = clean_data[5] / conv
hdf5_from_dict(output_file, out)
logger.info('Cleaned data written to %s' % output_file)
logger.info('Edits written to %s' % output_edits_file)
return
def pygeons_crop(input_file,start_date=None,stop_date=None,
min_lat=-np.inf,max_lat=np.inf,
min_lon=-np.inf,max_lon=np.inf,
stations=None,output_stem=None):
'''
Sets the time span of the data set to be between *start_date* and
*stop_date*. Sets the stations to be within the latitude and
longitude bounds.
Parameters
----------
data : dict
data dictionary
start_date : str, optional
start date of output data set in YYYY-MM-DD. Uses the start date
of *data* if not provided. Defaults to the earliest date.
stop_date : str, optional
Stop date of output data set in YYYY-MM-DD. Uses the stop date
of *data* if not provided. Defaults to the latest date.
min_lon, max_lon, min_lat, max_lat : float, optional
Spatial bounds on the output data set
stations : str list, optional
List of stations to be removed from the dataset. This is in
addition to the station removed by the lon/lat bounds.
Returns
-------
out_dict : dict
output data dictionary
'''
logger.info('Running pygeons crop ...')
data = dict_from_hdf5(input_file)
out = dict((k,np.copy(v)) for k,v in data.iteritems())
if start_date is None:
start_date = mjd.mjd_inv(data['time'].min(),'%Y-%m-%d')
if stop_date is None:
stop_date = mjd.mjd_inv(data['time'].max(),'%Y-%m-%d')
if stations is None:
stations = []
# remove times that are not within the bounds of *start_date* and
# *stop_date*
start_time = int(mjd.mjd(start_date,'%Y-%m-%d'))
stop_time = int(mjd.mjd(stop_date,'%Y-%m-%d'))
idx = ((data['time'] >= start_time) &
(data['time'] <= stop_time))
out['time'] = out['time'][idx]
for dir in ['east','north','vertical']:
out[dir] = out[dir][idx,:]
out[dir + '_std_dev'] = out[dir + '_std_dev'][idx,:]
# find stations that are within the bounds
in_bounds = ((data['longitude'] > min_lon) &
(data['longitude'] < max_lon) &
(data['latitude'] > min_lat) &
(data['latitude'] < max_lat))
# find stations that are in the list of stations to be removed
in_list = np.array([i in stations for i in data['id']])
# keep stations that are in bounds and not in the list
idx, = (in_bounds & ~in_list).nonzero()
out['id'] = out['id'][idx]
out['longitude'] = out['longitude'][idx]
out['latitude'] = out['latitude'][idx]
for dir in ['east','north','vertical']:
out[dir] = out[dir][:,idx]
out[dir + '_std_dev'] = out[dir + '_std_dev'][:,idx]
# set output file name
if output_stem is None:
output_stem = _remove_extension(input_file) + '.crop'
output_file = output_stem + '.h5'
hdf5_from_dict(output_file,out)
logger.info('Cropped data written to %s' % output_file)
return
