"""

Analyse observed and predicted alert times in California.

"""

def __init__(self):

self.del_coord = []

self.event_excludes = ['NC_71736656']

def background_map(self, ax):

llcrnrlat, urcrnrlat, llcrnrlon, urcrnrlon, lat_ts = (31, 44, -126, -113, 37.5)

m = Basemap(projection='merc', llcrnrlat=llcrnrlat,

urcrnrlat=urcrnrlat, llcrnrlon=llcrnrlon, urcrnrlon=urcrnrlon,

lat_ts=lat_ts, resolution='i', ax=ax)

m.drawmapboundary(fill_color='lightblue', zorder=0)

m.fillcontinents(zorder=0)

etopofn = '/home/behry/uni/data/etopo1_central_europe_gmt.grd'

etopodata = Dataset(etopofn, 'r')

z = etopodata.variables['z'][:]

x_range = etopodata.variables['x_range'][:]

y_range = etopodata.variables['y_range'][:]

spc = etopodata.variables['spacing'][:]

lats = np.arange(y_range[0], y_range[1], spc[1])

lons = np.arange(x_range[0], x_range[1], spc[0])

topoin = z.reshape(lats.size, lons.size, order='C')

# transform to nx x ny regularly spaced 5km native projection grid

nx = int((m.xmax - m.xmin) / 5000.) + 1; ny = int((m.ymax - m.ymin) / 5000.) + 1

topodat, x, y = m.transform_scalar(np.flipud(topoin), lons, lats, nx, ny, returnxy=True)

ls = LightSource(azdeg=300, altdeg=15, hsv_min_sat=0.2, hsv_max_sat=0.3,

hsv_min_val=0.2, hsv_max_val=0.3)

# shade data, creating an rgb array.

rgb = ls.shade(np.ma.masked_less(topodat / 1000.0, 0.0), cm.gist_gray_r)

m.imshow(rgb)

m.drawmeridians(np.arange(6, 12, 2), labels=[0, 0, 0, 1], color='white',

linewidth=0.5, zorder=0)

m.drawparallels(np.arange(44, 50, 2), labels=[1, 0, 0, 0], color='white',

linewidth=0.5, zorder=0)

m.drawcoastlines(zorder=1)

m.drawcountries(linewidth=1.5, zorder=1)

m.drawstates()

m.drawrivers(color='lightblue', zorder=1)

return m

def popup(self, fig, dataX, dataY, values):

# add a pop-up window showing the station and its value

try:

tooltip = wx.ToolTip(tip='')

tooltip.Enable(False)

tooltip.SetDelay(0)

fig.canvas.SetToolTip(tooltip)

def onMotion(event):

# dir(event.artist)

line2d = event.artist

x = line2d.get_xdata()[0]

y = line2d.get_ydata()[0]

found = False

for i in xrange(len(dataX)):

radius = 5

if abs(x - dataX[i]) < radius and abs(y - dataY[i]) < radius:

tip = '%s' % values[i]

tooltip.SetTip(tip)

tooltip.Enable(True)

found = True

break

if not found:

tooltip.Enable(False)

fig.canvas.mpl_connect('pick_event', onMotion)

except Exception, e:

print "Cannot add wx.ToolTip: ", e

def load_event_info(self, fn):

ei = {}

fh = open(fn)

for _l in fh.readlines():

a = _l.split()

# if int(a[1]) != 0

ei[a[0]] = int(a[1])

return ei

def closest_stations(self, stations, lat, lon, nnst=4, vp=6.5, depth=8):

"""

Find the 4 closest stations that are most likely to have detected this

event first.

"""

g = pyproj.Geod(ellps='WGS84')

min_dt = 1.e38

for net in ['bk', 'ca', 'mp']:

if len(stations.networks[net]['lat']) < 1:

continue

stlat = stations.networks[net]['lat']

stlon = stations.networks[net]['lon']

nwcode = np.array(stations.networks[net]['nw'])

names = np.array(stations.networks[net]['nm'])

# Find the <nnst> nearest stations

lats = np.ones((len(stlat),)) * lat

lons = np.ones((len(stlon),)) * lon

az, baz, dist = g.inv(lons, lats, stlon, stlat)

dist_sorted = np.sort(dist)

stat_names = names[np.argsort(dist)]

networks = nwcode[np.argsort(dist)]

dz = np.ones((10,)) * depth

distance = np.sqrt(dz * dz + dist_sorted[0:10] / 1000. * dist_sorted[0:10] / 1000.)

dt = max(distance[0:nnst] / vp)

if dt < min_dt:

min_dt = dt

# stats = ['%s.%s' % (net, st) for st in stat_names[0:nnst]]

stats = stat_names[0:nnst]

return stats

def statistics(self, fns, fn, stationfn, eventinfo=None, latencies=None,

computedelay=False, map=False, interactive=False):

"""

Compare predicted and observed alert times quantitatively.

"""

a = np.load(fn)

lats_tt = a['lat'][:, :, 0]

lons_tt = a['lon'][:, :, 0]

times = np.median(a['ttP'], axis=-1)

tree = spatial.KDTree(zip(lats_tt.ravel(), lons_tt.ravel()))

vals = []

perc_max = 84

perc_min = 16

rp = ReportsParser(dmin=UTCDateTime(2012, 1, 1, 0, 0, 0),

dmax=UTCDateTime(2013, 11, 1, 0, 0, 0))

# t = EventCA()

t = EventSoCal()

rp.sfilter = t.point_in_polygon

for _f in fns:

rp.read_reports(_f)

correct = rp.get_correct(mmin=3.5, mmax=10.0)

pid = correct[:, 0]

ot = correct[:, 2].astype('float')

lats = correct[:, 3].astype('float')

lons = correct[:, 4].astype('float')

deps = correct[:, 5].astype('float')

mags = correct[:, 6].astype('float')

ts1 = correct[:, 7].astype('float')

lats1 = correct[:, 9].astype('float')

lons1 = correct[:, 10].astype('float')

mags1 = correct[:, 12].astype('float')

rfns = correct[:, 21]

diff = ts1 - ot

magdiff = mags - mags1

cnt = 0

allcnt = 0

allm = []

dataX = []

dataY = []

popup_values = []

fig = plt.figure()

ax = fig.add_axes([0.1, 0.1, 0.8, 0.8])

m = self.background_map(ax)

cmap = cm.ScalarMappable(norm=Normalize(vmin=0, vmax=2), cmap='RdBu_r')

stats_used = []

for lon, lat, dep, delay, evid, lat1, lon1, dmag, time, mag, rfn in \

zip(lons, lats, deps, diff, pid, lats1, lons1, magdiff, ot, mags, rfns):

allcnt += 1

try:

if eventinfo is not None and len(eventinfo[evid]) != 4:

# print "Event %s does not have 4 initial picks." % evid

continue

except KeyError:

print "No event information available for: %s (%s)" % (evid, UTCDateTime(time))

continue

if evid in self.event_excludes:

print "Event %s was set to be excluded." % evid

continue

if computedelay:

# Compute the expected alert time for the actual epicenter and

# the first stations that detected the event

class NetworkInfo:

def __init__(self):

self.networks = {'ca':{'lat': [], 'lon': [], 'chn': [],

'nw': [], 'nm': [], 'lc': [],

'color':'black',

'label':'UC Berkeley'}}

def get_networks(self):

return self.networks

# read in SCEDC master station list

fh = open(stationfn)

scedc_stations = {}

for _l in fh.readlines():

if _l.startswith('#'):

continue

net, sta, chan, loc, lt, ln, elev, ondate, offdate = _l.split()

ns = '.'.join((net, sta))

if ns not in scedc_stations:

scedc_stations[ns] = (float(lt), float(ln))

ni = NetworkInfo()

for _st in eventinfo[evid]:

ni.networks['ca']['lat'].append(scedc_stations[_st][0])

ni.networks['ca']['lon'].append(scedc_stations[_st][1])

ni.networks['ca']['nm'].append(_st)

if _st not in stats_used:

stats_used.append(_st)

de = DelayEEW()

elat, elon, edep, ttP, tstarget = \

de.compute(ni, np.array([float(lon)]), np.array([float(lat)]),

np.array([float(dep)]),

vp=6.5, vs=3.5, nnst=4, procdelay=True, nmaps=500,

resultsfn=None, latencies=latencies)

med = np.median(ttP)

lb = scoreatpercentile(ttP, perc_min)

ub = scoreatpercentile(ttP, perc_max)

else:

distance, index = tree.query(np.array([[lat, lon]]))

irow, icol = divmod(index[0], lats_tt.shape[1])

med = np.median(times[:, irow, icol])

lb = scoreatpercentile(times[:, irow, icol], perc_min)

ub = scoreatpercentile(times[:, irow, icol], perc_max)

cnt += 1

allm.append(mag)

val = (delay - lb) / (ub - lb)

print med, lb, ub, delay, val, med - delay

vals.append(val)

cl = cmap.to_rgba(val)

x, y = m(lon, lat)

dataX.append(x)

dataY.append(y)

info = '%s: %.2f %s\n' % (UTCDateTime(time), mag, evid)

info += '%.2f %.2f %.2f\n' % (delay, med, val)

for _st in eventinfo[evid]:

info += ' %s' % _st

popup_values.append(info)

m.plot(x, y, ms=8, c=cl, marker='o', picker=5.)

# plt.figure()

# plt.hist(times[ilon, ilat, :], bins=np.arange(0, 30), normed=True, histtype='step')

# plt.show()

print "Stations used in detections:"

print stats_used

idx = np.where((np.array(vals) <= 1.0) & (np.array(vals) >= 0))

print "%.1f lie within the %d and %d percentile" % ((idx[0].size / float(len(vals))) * 100, perc_min, perc_max)

# plt.plot(allm, vals, 'bo')

if interactive:

self.popup(fig, dataX, dataY, popup_values)

cax = fig.add_axes([0.87, 0.1, 0.05, 0.8])

cb = ColorbarBase(cax, cmap='RdBu_r',

norm=Normalize(vmin=0, vmax=2))

cb.set_label('Alert accuracy')

plt.figure()

plt.hist(vals, bins=20)

plt.show()

def alert_times_map(self, fns, m=None, fig=None, ax=None, scale=10000.,

cb=True, disterr=False, interactive=False,

eventinfo=None, msscale=1, cmapname='jet'):

"""

Plot a map of observed alert times.

"""

cmap = cm.ScalarMappable(norm=Normalize(vmin=6, vmax=25), cmap=cmapname)

rp = ReportsParser(dmin=UTCDateTime(2012, 1, 1, 0, 0, 0),

dmax=UTCDateTime(2013, 11, 1, 0, 0, 0))

t = EventCA()

rp.sfilter = t.point_in_polygon

for _f in fns:

rp.read_reports(_f)

correct = rp.get_correct(mmin=3.5, mmax=10.0)

pid = correct[:, 0]

ot = correct[:, 2].astype('float')

lats = correct[:, 3].astype('float')

lons = correct[:, 4].astype('float')

mags = correct[:, 6].astype('float')

ts1 = correct[:, 7].astype('float')

lats1 = correct[:, 9].astype('float')

lons1 = correct[:, 10].astype('float')

mags1 = correct[:, 12].astype('float')

rfns = correct[:, 21]

diff = ts1 - ot

magdiff = mags - mags1

if m is None and fig is None and ax is None:

fig = plt.figure()

ax = fig.add_axes([0.1, 0.1, 0.8, 0.8])

m = self.background_map(ax)

dataX = []

dataY = []

values = []

# load event info

cnt = 0

allcnt = 0

for lon, lat, delay, evid, lat1, lon1, dmag, time, mag, rfn in \

zip(lons, lats, diff, pid, lats1, lons1, magdiff, ot, mags, rfns):

allcnt += 1

try:

if eventinfo is not None and len(eventinfo[evid]) != 4:

print "Event %s does not have 4 initial picks." % evid

continue

except KeyError:

print "No event information available for: %s (%s)" % (evid, UTCDateTime(time))

continue

if evid in self.event_excludes:

print "Event %s was set to be excluded." % evid

continue

cnt += 1

ddist, az, baz = gps2DistAzimuth(lat, lon, lat1, lon1)

ddist /= 1000.

x, y = m(lon, lat)

dataX.append(x)

dataY.append(y)

info = '%s: %.2f %.2f %s' % (UTCDateTime(time), delay, mag, evid)

for _st in eventinfo[evid]:

info += ' %s' % _st

values.append(info)

cl = cmap.to_rgba(delay)

if disterr:

factor = math.sqrt(abs(float(ddist)))

sl2 = scale * factor

p2 = Wedge((x, y), sl2, 0, 360, facecolor=cl,

edgecolor='black', picker=5, lw=1.0)

ax.add_patch(p2)

else:

m.plot(x, y, ms=8 * msscale, c=cl, marker='o', picker=5.)

print "Plotted %d out of %d events." % (cnt, allcnt)

if interactive:

self.popup(fig, dataX, dataY, values)

if cb:

# Colorbar

cax = fig.add_axes([0.87, 0.1, 0.05, 0.8])

cb = ColorbarBase(cax, cmap=cmapname,

norm=Normalize(vmin=6., vmax=25.))