Coverage for /usr/local/lib/python3.11/dist-packages/pyrocko/weeding.py: 0%
77 statements
« prev ^ index » next coverage.py v6.5.0, created at 2023-10-25 15:33 +0000
« prev ^ index » next coverage.py v6.5.0, created at 2023-10-25 15:33 +0000
1# http://pyrocko.org - GPLv3
2#
3# The Pyrocko Developers, 21st Century
4# ---|P------/S----------~Lg----------
6'''
7An algorithm to select station subsets with good coverage in azimuth and
8distance.
9'''
11import numpy as num
14def neighborhood_density(dists, neighborhood=1):
15 sdists = dists.copy()
16 sdists.sort(axis=1)
17 meandists = num.mean(sdists[:, 1:1+neighborhood], axis=1)
18 return meandists
21def _weed(dists, badnesses, neighborhood=1, interaction_radius=3.,
22 del_frac=4, max_del=100, max_depth=100, depth=0):
24 if depth > max_depth:
25 assert False, 'max recursion depth reached'
27 meandists = neighborhood_density(dists, neighborhood)
29 order = meandists.argsort()
30 candidates = order[:order.size//del_frac+1]
31 badness_candidates = badnesses[candidates]
32 order_badness = (-badness_candidates).argsort()
34 order[:order.size//del_frac+1] = \
35 order[:order.size//del_frac+1][order_badness]
37 deleted = num.zeros(order.size, dtype=bool)
38 ndeleted = 0
39 for i, ind in enumerate(order):
40 if (i < order.size // del_frac // 2 + 1
41 and ndeleted < max_del
42 and num.all(
43 dists[ind, deleted] > interaction_radius*meandists[ind])):
45 deleted[ind] = True
46 ndeleted += 1
48 if ndeleted == 0:
49 return deleted
51 kept = num.logical_not(deleted)
52 xdists = dists[kept][:, kept]
53 xbadnesses = badnesses[kept]
54 xdeleted = _weed(xdists, xbadnesses, neighborhood, interaction_radius,
55 del_frac, max_del-ndeleted, max_depth, depth+1)
57 deleted[kept] = xdeleted
58 return deleted
61def weed(x, y, badnesses, neighborhood=1, nwanted=None, interaction_radius=3.):
62 assert x.size == y.size
63 n = x.size
64 NEW = num.newaxis
65 ax = x[NEW, :]
66 ay = y[NEW, :]
67 bx = x[:, NEW]
68 by = y[:, NEW]
70 if nwanted is None:
71 nwanted = n // 2
73 dx = num.abs(ax-bx)
74 dx = num.where(dx > 180., 360.-dx, dx)
76 dists = num.sqrt(dx**2 + (ay-by)**2)
77 deleted = _weed(
78 dists, badnesses, neighborhood, interaction_radius, del_frac=4,
79 max_del=n-nwanted, max_depth=500, depth=0)
81 kept = num.logical_not(deleted)
82 dists_kept = dists[kept][:, kept]
83 meandists_kept = neighborhood_density(dists_kept, neighborhood)
84 return deleted, meandists_kept
87def badnesses_c_mean(badnesses_nslc):
88 # convert stream badnesses to station badnesses by averaging
89 badnesses_nsl = {}
90 for nslc, b in badnesses_nslc.items():
91 nsl = nslc[:3]
92 badnesses_nsl.setdefault(nsl, []).append(b)
94 for nsl in list(badnesses_nsl.keys()):
95 this_station_badness = badnesses_nsl[nsl]
96 badnesses_nsl[nsl] = sum(this_station_badness) \
97 / len(this_station_badness)
99 return badnesses_nsl
102def weed_stations(stations, nwanted, neighborhood=3, default_badness=1.0,
103 badnesses=None,
104 badnesses_ns={}, badnesses_nsl={}, badnesses_nslc={}):
106 azimuths = num.zeros(len(stations), dtype=float)
107 dists = num.zeros(len(stations), dtype=float)
108 for ista, sta in enumerate(stations):
109 azimuths[ista] = sta.azimuth
110 dists[ista] = sta.dist_deg
112 badnesses_nslc_mean_c = badnesses_c_mean(badnesses_nslc)
113 badnesses2 = num.ones(len(stations), dtype=float)
114 for ista, sta in enumerate(stations):
115 nsl = sta.network, sta.station, sta.location
116 if badnesses is not None:
117 b = badnesses[ista]
118 else:
119 b = 1.0
120 b *= badnesses_ns.get(nsl, 1.0)
121 b *= badnesses_nsl.get(nsl, 1.0)
122 b *= badnesses_nslc_mean_c.get(nsl, 1.0)
123 badnesses2[ista] = b
125 deleted, meandists_kept = weed(
126 azimuths, dists, badnesses2,
127 nwanted=nwanted,
128 neighborhood=neighborhood)
130 stations_weeded = [station for (delete, station) in zip(deleted, stations)
131 if not delete]
133 return stations_weeded, meandists_kept, deleted