Coverage for /usr/local/lib/python3.7/dist-packages/pyrocko/plot/directivity.py : 85%

# http://pyrocko.org - GPLv3 

# 

# The Pyrocko Developers, 21st Century 

# ---|P------/S----------~Lg---------- 

import logging 

import numpy as num 

import matplotlib.pyplot as plt 

from matplotlib.cm import ScalarMappable 

from matplotlib.ticker import FuncFormatter 

from pyrocko.plot import beachball 

from pyrocko.gf.meta import Timing 

from pyrocko.gf import LocalEngine, Target, RectangularSource, map_anchor 

from pyrocko.util import num_full_like 

km = 1e3 

r2d = 180. / num.pi 

d2r = num.pi / 180. 

logger = logging.getLogger(__name__) 

QUANTITY_LABEL = { 

'displacement': 'Displacement [m]', 

'velocity': 'Velocity [m/s]', 

'acceleration': 'Acceleration [m/s²]' 

} 

def get_azimuthal_targets( 

store_id, source, radius, 

azi_begin=0., azi_end=360., dazi=1., 

interpolation='multilinear', 

components='RTZ', quantity='displacement'): 

assert dazi > 0. 

assert azi_begin < azi_end 

nstations = int((azi_end - azi_begin) // dazi) 

assert nstations > 0 

azimuths = num.linspace(azi_begin, azi_end, nstations) 

coords = num.zeros((2, nstations)) 

coords[0, :] = num.cos(azimuths*d2r) 

coords[1, :] = num.sin(azimuths*d2r) 

coords *= radius 

dips = {'R': 0., 'T': 0., 'Z': -90.} 

for comp in components: 

assert comp in dips.keys() 

target_kwargs = dict( 

quantity='displacement', 

interpolation=interpolation, 

store_id=store_id) 

targets = [ 

Target( 

lat=source.lat, 

lon=source.lon, 

north_shift=coords[0, iazi] + source.north_shift, 

east_shift=coords[1, iazi] + source.east_shift, 

azimuth={ 

'R': azi, 

'T': azi+90., 

'Z': 0. 

}[channel], 

dip=dips[channel], 

codes=('', 'S%01d' % iazi, '', channel), 

**target_kwargs) 

for iazi, azi in enumerate(azimuths) 

for channel in components] 

for target, azi in zip(targets, azimuths): 

target.azimuth = azi 

target.dazi = dazi 

return targets, azimuths 

def get_seismogram_array( 

response, fmin=None, fmax=None, 

component='R', envelope=False): 

resp = response 

assert len(resp.request.sources) == 1, 'more than one source in response' 

tmin = None 

tmax = None 

traces = [] 

for _, target, tr in response.iter_results(): 

if target.codes[-1] != component: 

continue 

assert hasattr(target, 'azimuth') 

assert target.dazi 

if fmin and fmax: 

tr.bandpass(2, fmin, fmax) 

elif fmin: 

tr.highpass(4, fmin) 

elif fmax: 

tr.lowpass(4, fmax) 

tmin = min(tmin, tr.tmin) if tmin else tr.tmin 

tmax = max(tmax, tr.tmax) if tmax else tr.tmax 

traces.append(tr) 

for tr in traces: 

tr.extend(tmin, tmax, fillmethod='repeat') 

if envelope: 

tr.abshilbert() 

data = num.array([tr.get_ydata() for tr in traces]) 

data -= data.mean() 

nsamples = data.shape[1] 

return data, num.linspace(tmin, tmax, nsamples) 

def hillshade(array, azimuth, angle_altitude): 

azimuth = 360.0 - azimuth 

azi = azimuth * r2d 

alt = angle_altitude * r2d 

x, y = num.gradient(array) 

slope = num.pi/2. - num.arctan(num.sqrt(x*x + y*y)) 

aspect = num.arctan2(-x, y) 

shaded = num.sin(alt)*num.sin(slope) \ 

+ num.cos(alt)*num.cos(slope)*num.cos((azi - num.pi/2.) - aspect) 

return (shaded + 1.)/2. 

def hillshade_seismogram_array( 

seismogram_array, rgba_map, 

shad_lim=(.4, .98), contrast=1., blend_mode='multiply'): 

assert blend_mode in ('multiply', 'screen'), 'unknown blend mode' 

assert shad_lim[0] < shad_lim[1], 'bad shading limits' 

from scipy.ndimage import convolve as im_conv 

# Light source from somewhere above - psychologically the best choice 

# from upper left 

ramp = num.array([[1., 0.], [0., -1.]]) * contrast 

# convolution of two 2D arrays 

shad = im_conv(seismogram_array, ramp.T).ravel() 

shad *= -1. 

# if there are strong artifical edges in the data, shades get 

# dominated by them. Cutting off the largest and smallest 2% of 

# # shades helps 

percentile2 = num.percentile(shad, 2.0) 

percentile98 = num.percentile(shad, 98.0) 

shad[shad > percentile98] = percentile98 

shad[shad < percentile2] = percentile2 

# # normalize shading 

shad -= num.nanmin(shad) 

shad /= num.nanmax(shad) 

# # reduce range to balance gray color 

shad *= shad_lim[1] - shad_lim[0] 

shad += shad_lim[0] 

if blend_mode == 'screen': 

rgba_map[:, :3] = 1. - ((1. - rgba_map[:, :3])*(shad[:, num.newaxis])) 

elif blend_mode == 'multiply': 

rgba_map[:, :3] *= shad[:, num.newaxis] 

return rgba_map 

def plot_directivity( 

engine, source, store_id, 

distance=300*km, azi_begin=0., azi_end=360., dazi=1., 

phases={'P': 'first{stored:any_P}-10%', 

'S': 'last{stored:any_S}+50'}, 

quantity='displacement', envelope=False, 

component='R', fmin=0.01, fmax=0.1, 

hillshade=True, cmap=None, 

plot_mt='full', show_phases=True, show_description=True, 

reverse_time=False, show_nucleations=True, axes=None, nthreads=0): 

''' 

Plot the directivity and radiation characteristics of source models. 

Synthetic seismic traces (R, T or Z) are forward-modelled at a defined 

radius, covering the full or partial azimuthal range and projected on a 

polar plot. Difference in the amplitude are enhanced by hillshading 

the data. 

:param engine: Forward modelling engine 

:type engine: :py:class:`~pyrocko.gf.seismosizer.Engine` 

:param source: Parametrized source model 

:type source: :py:class:`~pyrocko.gf.seismosizer.Source` 

:param store_id: Store ID used for forward modelling 

:type store_id: str 

:param distance: Distance in [m] 

:type distance: float 

:param azi_begin: Begin azimuth in [deg] 

:type azi_begin: float 

:param azi_end: End azimuth in [deg] 

:type azi_end: float 

:param dazi: Delta azimuth, bin size [deg] 

:type dazi: float 

:param phase_begin: Start time of the window 

:type phase_begin: :py:class:`~pyrocko.gf.meta.Timing` 

:param phase_end: End time of the window 

:type phase_end: :py:class:`~pyrocko.gf.meta.Timing` 

:param quantity: Seismogram quantity, default ``displacement`` 

:type quantity: str 

:param envelope: Plot envelop instead of seismic trace 

:type envelope: bool 

:param component: Forward modelled component, default ``R``. Choose from 

`RTZ` 

:type component: str 

:param fmin: Bandpass lower frequency [Hz], default ``0.01`` 

:type fmin: float 

:param fmax: Bandpass upper frequency [Hz], default ``0.1`` 

:type fmax: float 

:param hillshade: Enable hillshading, default ``True`` 

:type hillshade: bool 

:param cmap: Matplotlit colormap to use, default ``seismic``. 

When ``envelope`` is ``True`` the default colormap will be ``Reds``. 

:type cmap: str 

:param plot_mt: Plot a centered moment tensor, default ``full``. 

Choose from ``full, deviatoric, dc or False`` 

:type plot_mt: str, bool 

:param show_phases: Show annotations, default ``True`` 

:type show_phases: bool 

:param show_description: Show desciption, default ``True`` 

:type show_description: bool 

:param reverse_time: Reverse time axis. First phases arrive at the center, 

default ``False`` 

:type reverse_time: bool 

:param show_nucleations: Show nucleation piercing points on the moment 

tensor, default ``True`` 

:type show_nucleations: bool 

:param axes: Give axes to plot into 

:type axes: :py:class:`matplotlib.axes.Axes` 

:param nthreads: Number of threads used for forward modelling, 

default ``0`` - all available cores 

:type nthreads: int 

''' 

if axes is None: 

fig = plt.figure() 

ax = fig.add_subplot(111, polar=True) 

else: 

fig = axes.figure 

ax = axes 

if envelope and cmap is None: 

cmap = 'Reds' 

elif cmap is None: 

cmap = 'seismic' 

targets, azimuths = get_azimuthal_targets( 

store_id, source, distance, azi_begin, azi_end, dazi, 

components='R', quantity=quantity) 

ref_target = targets[0] 

store = engine.get_store(store_id) 

mt = source.pyrocko_moment_tensor(store=store, target=ref_target) 

resp = engine.process(source, targets, nthreads=nthreads) 

data, times = get_seismogram_array( 

resp, fmin, fmax, 

component=component, envelope=envelope) 

nucl_depth = source.depth 

nucl_distance = distance 

if hasattr(source, 'nucleation_x') and hasattr(source, 'nucleation_y'): 

try: 

iter(source.nucleation_x) 

nx = float(source.nucleation_x[0]) 

ny = float(source.nucleation_y[0]) 

except TypeError: 

nx = source.nucleation_x 

ny = source.nucleation_y 

nucl_distance += nx * source.length/2. 

nucl_depth += ny*num.sin(source.dip*d2r) * source.width/2. 

if hasattr(source, 'anchor'): 

anch_x, anch_y = map_anchor[source.anchor] 

nucl_distance -= anch_x * source.length/2. 

nucl_depth -= anch_y*num.sin(source.dip*d2r) * source.width/2. 

timings = [Timing(p) for p in phases.values()] 

phase_times = [store.t(t, source, ref_target) for t in timings] 

tbegin = min(phase_times) 

tend = max(phase_times) 

tsel = num.logical_and(times >= tbegin, times <= tend) 

data = data[:, tsel].T 

times = times[tsel] 

duration = times[-1] - times[0] 

vmax = num.abs(data).max() 

cmw = ScalarMappable(cmap=cmap) 

cmw.set_array(data) 

cmw.set_clim(-vmax, vmax) 

if envelope: 

cmw.set_clim(0., vmax) 

ax.set_theta_zero_location('N') 

ax.set_theta_direction(-1) 

strike_label = mt.strike1 

if hasattr(source, 'strike'): 

strike_label = source.strike 

try: 

ax.set_rlabel_position(strike_label % 180.) 

except AttributeError: 

logger.warn('Old matplotlib version: cannot set label positions') 

def r_fmt(v, p): 

if v < tbegin or v > tend: 

return '' 

return '%g s' % v 

ax.yaxis.set_major_formatter(FuncFormatter(r_fmt)) 

if reverse_time: 

ax.set_rlim(times[0] - .3*duration, times[-1]) 

else: 

ax.set_rlim(times[-1] + .3*duration, times[0]) 

ax.grid(zorder=20) 

if isinstance(plot_mt, str): 

mt_size = .15 

beachball.plot_beachball_mpl( 

mt, ax, 

beachball_type=plot_mt, size=mt_size, 

size_units='axes', color_t=(0.7, 0.4, 0.4), 

position=(.5, .5), linewidth=1.) 

if hasattr(source, 'nucleation_x') and hasattr(source, 'nucleation_y')\ 

and show_nucleations: 

try: 

iter(source.nucleation_x) 

nucleation_x = source.nucleation_x 

nucleation_y = source.nucleation_y 

except TypeError: 

nucleation_x = [source.nucleation_x] 

nucleation_y = [source.nucleation_y] 

for nx, ny in zip(nucleation_x, nucleation_y): 

angle = float(num.arctan2(ny, nx)) 

rtp = num.array([[1., angle, (90.-source.strike)*d2r]]) 

points = beachball.numpy_rtp2xyz(rtp) 

x, y = beachball.project(points, projection='lambert').T 

norm = num.sqrt(x**2 + y**2) 

x = x / norm * mt_size/2. 

y = y / norm * mt_size/2. 

ax.plot(x+.5, y+.5, 'x', ms=6, mew=2, mec='darkred', mfc='red', 

transform=ax.transAxes, zorder=10) 

mesh = ax.pcolormesh( 

azimuths * d2r, times, data, 

cmap=cmw.cmap, shading='gouraud', zorder=0) 

if hillshade: 

mesh.update_scalarmappable() 

color = mesh.get_facecolor() 

color = hillshade_seismogram_array( 

data, color, shad_lim=(.85, 1.), blend_mode='multiply') 

mesh.set_facecolor(color) 

if show_phases: 

label_theta = 270. 

theta = num.linspace(0, 2*num.pi, 360) 

for label, phase_str in phases.items(): 

phase = Timing(phase_str) 

phase.offset = 0. 

phase.offset_is_slowness = False 

phase.offset_is_percent = False 

time = store.t(phase, source, ref_target) 

times = num_full_like(theta, time) 

ax.plot(theta, times, color='k', alpha=.3, lw=1., ls='--') 

ax.text( 

label_theta*d2r, time, label, 

ha='left', color='k', fontsize='small') 

label_theta += 30. 

if show_description: 

description = ( 

'Component {component:s}\n' 

'Distance {distance:g} km').format( 

component=component, distance=distance / km) 

if fmin and fmax: 

description += '\nBandpass {fmin:g} - {fmax:g} Hz'.format( 

fmin=fmin, fmax=fmax) 

elif fmin: 

description += '\nHighpass {fmin:g} Hz'.format(fmin=fmin) 

elif fmax: 

description += '\nLowpass {fmax:g} Hz'.format(fmax=fmax) 

ax.text( 

-.05, -.05, description, 

fontsize='small', 

ha='left', va='bottom', transform=ax.transAxes) 

cbar_label = QUANTITY_LABEL[quantity] 

if envelope: 

cbar_label = 'Envelope ' + cbar_label 

cb = fig.colorbar( 

cmw, ax=ax, 

orientation='vertical', shrink=.8, pad=0.11) 

cb.set_label(cbar_label) 

if axes is None: 

plt.show() 

return resp 

__all__ = ['plot_directivity'] 

if __name__ == '__main__': 

engine = LocalEngine(store_superdirs=['.'], use_config=True) 

rect_source = RectangularSource( 

depth=2.6*km, 

strike=240., 

dip=76.6, 

rake=-.4, 

anchor='top', 

nucleation_x=-.57, 

nucleation_y=-.59, 

velocity=2070., 

length=27*km, 

width=9.4*km, 

slip=1.4) 

resp = plot_directivity( 

engine, rect_source, 'crust2_ib', 

dazi=5, component='R', quantity='displacement', envelope=True)