def grideventmask(allpolys, s, lon, lat):
'''Returns a mask for gridded dataset for each day of each event
format: dictionary for each entry'''
keylist = allpolys.keys()
keylist.sort()
gridmasks = {}
masklist = []
maskdtimes = []
maskekeys = []
for k in keylist:
polylist = []
hrsince = s.events[k].trkarrs[s.events[k].refkey][:, 0]
dtimes = my.hrnum2dates(hrsince)
chulltrue = s.events[k].trkarrs[s.events[k].refkey][:, 1] != -1
print chulltrue, dtimes
cnt = 0
for poly in allpolys[k]:
if chulltrue[cnt]:
chmask = my.poly2mask(lon, lat, poly)
polylist.append(chmask)
masklist.append(chmask)
maskdtimes.append(dtimes[cnt])
maskekeys.append(k)
cnt += 1
gridmasks[k] = polylist
return gridmasks, masklist, maskdtimes, maskekeys
def clim_wetdayspercent(erainmasked, rtime, wetness=1):
'''clim_wetdayspercent(erainmasked,rtime)
Usage: erainm is outputfrom projectrainmask_in2_rawdata
rtime is associated time
Returns: wetdayscount (stations x 13) (ndarray), 12 months and annual mean
freq_wet frequency of wewet
scycle_wet & scycle_freqwet (stations x nseasons x nmonths),
which are same as above stats but per season'''
season = [8, 9, 10, 11, 12, 1, 2, 3, 4, 5, 6, 7]
dates = my.hrnum2dates(rtime)
years = np.unique(dates[:, 0])
wetdayscnt = np.zeros((erainmasked.shape[0], len(season)))
freqwet = wetdayscnt.copy()
eventmask = ~erainmasked.mask
wetmask = erainmasked.data > wetness
# LOOK AT THE TOTAL PERCENTAGE IN ALL DATA
cnt = 0
for mn in season:
ix = np.where(dates[:, 1] == mn)[0]
wetmsk = np.int32(wetmask[:, ix])
ewetmsk = np.int32(wetmask[:, ix] & eventmask[:, ix])
wetdayscnt[:, cnt] = wetmsk.sum(1) / np.float32(len(years))
freqwet[:, cnt] = ewetmsk.sum(1) / np.float32(len(years)) / np.float32(
wetdayscnt[:, cnt])
cnt += 1
# SPLIT UP BY ACTUAL SEASONS
scycle_wet = np.zeros((erainmasked.shape[0], len(years[:-1]), len(season)))
scycle_freqwet = scycle_wet.copy()
for iyr in xrange(len(years[:-1])):
yr = years[iyr]
for imn in xrange(len(season)):
mn = season[imn]
if imn < 5:
ix = np.where((dates[:, 0] == yr) & (dates[:, 1] == mn))[0]
wetmsk = np.int32(wetmask[:, ix])
ewetmsk = np.int32(wetmask[:, ix] & eventmask[:, ix])
else:
ix = np.where((dates[:, 0] == yr + 1) & (dates[:, 1] == mn))[0]
wetmsk = np.int32(wetmask[:, ix])
ewetmsk = np.int32(wetmask[:, ix] & eventmask[:, ix])
scycle_wet[:, iyr, imn] = wetmsk.sum(1)
scycle_freqwet[:, iyr,
imn] = ewetmsk.sum(1) / np.float32(wetmsk.sum(1))
return wetdayscnt, freqwet, scycle_wet, scycle_freqwet
def eventanddaycount(s, rainmasks):
'''Simple count of number of events and rain days per month'''
erain, etime, ekeys, flagarr = rainmasks['rain']
uky, ixk = np.unique(ekeys, return_index=True)
erain, etime, ekeys, flagarr = erain[
ixk, :], etime[ixk], ekeys[ixk], flagarr[ixk, :]
ixwet = np.any(~erain.mask).nonzero()
edates = my.hrnum2dates(etime)
yrs = np.unique(edates[:, 0])
yrs = yrs[:-1]
eventcnt = np.zeros((len(yrs), len(season)), dtype=np.float32)
daycnt = np.zeros((len(yrs), len(season)), dtype=np.float32)
iyr = -1
for yr in yrs:
iyr += 1
imn = -1
for mn in season:
imn += 1
ix = np.where((edates[:, 0] == yr) & (edates[:, 1] == mn))
if len(ix) > 0: ix = ix[0]
else:
eventcnt[iyr, imn], daycnt[iyr, imn] = 0, 0
continue
mnkeys = ekeys[ix]
nky = len(mnkeys)
dcnt = 0
for k in mnkeys:
dcnt += len((
~np.isnan(s.events[k].rainfall['wrc'][:, 0])).nonzero()[0])
eventcnt[iyr, imn] = nky
daycnt[iyr, imn] = dcnt
octmarecnt = eventcnt.sum(1)
octmardcnt = daycnt.sum(1)
summarystats = {
'eventcnt_mean': eventcnt.mean(0),
'eventcnt_std': eventcnt.std(0),
'daycnt_mean': daycnt.mean(0),
'daycnt_std': daycnt.std(0),
'ONDJFevent_mean': octmarecnt.mean(),
'ONDJFevent_std': octmarecnt.std(),
'ONDJFday_mean': octmardcnt.mean(),
'ONDJFday_std': octmardcnt.std()
}
return eventcnt, daycnt, octmarecnt, octmardcnt, summarystats
def multiplot(e, m, etime, blobkeylist, trkkeylist, *args):
'''A plotting multitool that can plot all variables used and produced
by MetBot package
USAGE: etime can either be a tuple = (hrsbefore, hrsafter)
or a list = [flagday1,flagday2]
*args are gridded dataset and rainfall station dataset lists
[mbskey,data,lat,lon,time,alt_mbskey]
IMPORTANT: if u,v vector pairs are supplied,
they must be supplied in that order
if u,v vector pairs for multiple levels are supplied
they need to be kept paired'''
### UNPACK *ARGS LISTS OF VARIABLES INTO USABLE FORMAT
hgtlist, uvlist, quvlist, olrlist = [], [], [], []
print 'Event:', e.trkkey
for arg in args:
v = arg[0]
dset, varstr, lev, drv = v.split('-')
varlev = varstr + '_' + lev
#print varlev
exec(varlev + 'key,' + varlev + ',' + varlev + 'lat,' + varlev +
'lon,' + varlev + 'time,' + varlev + '_altkey=arg')
if varstr == 'qv': quvlist.append(['qu' + '_' + lev, varlev])
elif varstr == 'hgt': hgtlist.append(varlev)
elif varstr == 'v': uvlist.append(['u' + '_' + lev, varlev])
### PLOT TRKARRS DATA USING INDEX LIST
masktimes = e.trkarrstime[e.refkey]
# this because had some issues with some empty trakarrs which won't have the times6hr needed
for nmbs in e.mbskeys[3:]:
if ~isinstance(e.trkarrs[nmbs], dict): mbs6hrkey = nmbs
times6hr = e.trkarrstime[nmbs] # Used for labeling of figure
dlabels, ixlist, flarr = daylabel(e, etime)
for i in ixlist:
# OLR PLOT
if 'olr_0' in locals():
eolr = maskeddata(e, i, olr_0, olr_0time, olr_0key)
plotolr(m, olr_0lat, olr_0lon, eolr)
# HGT PLOT
for ph in hgtlist:
exec('hgt, hgtlat, hgtlon, hgttime, hgtkey = ' + ph + ', ' + ph +
'lat, ' + ph + 'lon, ' + ph + 'time, ' + ph + 'key')
ehgt = maskeddata(e, i, hgt, hgttime, hgtkey)
plothgt(m, hgtlat, hgtlon, ehgt)
# STATION RAINFALL PLOT
if 'rain_0' in locals():
plotrainstations(m, e, i, rain_0, rain_0lat, rain_0lon, rain_0time)
# BLOB CONTOURS
for blbkey in blobkeylist:
plotblob(m, e, i, blbkey, blobkeylist.index(blbkey))
# BLOB TRACKS
for trkkey in trkkeylist:
plottrack(m, e, i, trkkey, trkkeylist.index(trkkey))
# QUV PLOT
cnt = 0
for pqu, pqv in quvlist:
exec('qu, qv, qlat, qlon, qtime, qu_altkey, qv_altkey = ' + pqu +
', ' + pqv + ', ' + pqu + 'lat, ' + pqu + 'lon, ' + pqu +
'time, ' + pqu + '_altkey, ' + pqv + '_altkey')
equ = maskeddata(e, i, qu, qtime, qu_altkey)
eqv = maskeddata(e, i, qv, qtime, qv_altkey)
plotquv(m, qlat, qlon, equ, eqv, cols[cnt])
cnt += 1
cnt = 0
for pu, pv in uvlist:
exec('u, v, qlat, qlon, qtime, u_altkey, v_altkey = ' + pu + ', ' +
pv + ', ' + pu + 'lat, ' + pu + 'lon, ' + pu + 'time, ' + pu +
'_altkey, ' + pv + '_altkey')
eu = maskeddata(e, i, u, qtime, u_altkey)
ev = maskeddata(e, i, v, qtime, v_altkey)
plotuv(m, qlat, qlon, eu, ev, gcols[cnt])
cnt += 1
# DRAW EVERYTHING
redrawmap(m, lns=True)
# a little thing to handle time labeling relative to event flagging
tm6, tm24 = times6hr[i], masktimes[i]
tdate = my.hrnum2dates([tm6])[0]
eventcode = dlabels[i]
humandate = 'Date: %02d-%02d-%02d %02dh00' % (tdate[2], tdate[1],
tdate[0], tdate[3])
plt.title(str(e.trkkey) + ': ' + eventcode)
plt.text(30, -49, humandate)
plt.draw()
# SAVE FIGURE
dstring = str(e.trkkey) + '_%02d-%02d-%02d_%02dh00' % (
tdate[0], tdate[1], tdate[2], tdate[3])
fname = figdir + dstring + '.png'
plt.savefig(fname, dpi=150)
#sleep(1)
#raw_input()
plt.clf()
def extremerainfall(s,
kchoice,
rain,
rtime,
perc=.95,
areacrit=.1,
selectmonths=False):
'''extremeindices, assocTTT,assocTCOL = extremerainfall(s,kchoice,rain,rtime,perc=.95,areacriteria=.1,selectmonths=False))
Big and ungainly approach to get out extreme rainfall day indices and associations to TTTs
Usage: pretty obvious
Returns: extremeindices, ndarray of extreme days in rain data set
assocTTTs, ndarray of (extremeeventkey, ttteventkey, eventsoverlap_in_days, cutofflow_or_not)
assocTCOL, same as above but where cutofflow_or_not is true'''
rsort = rain.copy()
rsort.sort(axis=1)
### CALCULATE EXTREME VALUE CRITERIA FOR EACH STATION
wet = (rsort > 1.)
extrmvals = np.zeros((rsort.shape[0]))
for istn in xrange(rsort.shape[0]):
iw = wet[istn, :].nonzero()[0]
iex = int(len(iw) * perc)
iexwet = iw[iex]
extrmvals[istn] = rsort[istn, iexwet]
### PRODUCE TIME INDICES OF EXTREME DATES AND A EXTREME ARRAY MASK OF DATES
extrm_ixt = []
exmsk = np.ones(rain.shape, dtype=np.bool)
nstns = rain.shape[0]
areacriteria = int(nstns * areacrit)
for ixt in xrange(rain.shape[1]):
exmsk[:, ixt] = (rain[:, ixt] >= extrmvals)
if len(exmsk[:, ixt].nonzero()[0]) >= areacriteria:
extrm_ixt.append(ixt)
### GET EXTREME RAINFALL DATES
extrm_ixt = np.asarray(extrm_ixt)
### IF ONLY SELECTING EXTREMES FROM PARTICULAR MONTHS
if selectmonths:
ixselect = []
exdates = my.hrnum2dates(rtime[extrm_ixt])
for mn in selectmonths:
ixselmn = np.where(exdates[:, 1] == mn)
if len(ixselmn) > 0: ixselect.extend(ixselmn[0])
else: continue
extrm_ixt = extrm_ixt[ixselect]
exrtime = rtime[extrm_ixt]
consecutive = np.array([0], dtype=np.int32)
consecutive = np.append(consecutive, extrm_ixt[1:] - extrm_ixt[:-1])
extrmevents = {}
ix = 0
while ix < len(extrm_ixt):
ixlist = [extrm_ixt[ix]]
if ix + 1 >= len(extrm_ixt):
extrmevents[int(rtime[ixlist[0]])] = ixlist
ix += 1
continue
diff = consecutive[ix + 1]
if diff == 1:
ix += 1
while diff == 1:
ixlist.append(extrm_ixt[ix])
ix += 1
if ix >= len(extrm_ixt): break
diff = consecutive[ix]
else:
ix += 1
extrmevents[int(rtime[ixlist[0]])] = ixlist
ttttime, tttkeys = [], []
for k in kchoice:
ktimes = list(s.events[k].trktimes)
ttttime.extend(list(ktimes))
for dum in xrange(len(ktimes)):
tttkeys.append(k)
ttttime, tttkeys = np.asarray(ttttime), np.asarray(tttkeys)
extrmkeys = extrmevents.keys()
extrmkeys.sort()
assoc = []
for k in extrmkeys:
tttk = []
exhrs = rtime[extrmevents[k]]
#print exhrs
for hr in exhrs:
imatch = np.where(ttttime == hr)[0]
#print int(hr),imatch,len(imatch)
if len(imatch) == 0:
continue
else:
imatch = list(imatch)
#print imatch
tttk.extend(tttkeys[imatch])
tkey = list(np.unique(np.asarray(tttk)))
col = 0
if 'COL' in s.events[kchoice[0]].trkarrs.keys():
for tk in tkey:
if len(s.events[tk].trkarrs['COL']) > 0: col += 1
#print tkey
### USE THIS SETUP TO CAPTURE THE SOMETIMES 2 OR MORE CLOUD BAND EVENTS ASSOCIATED TO EXTREME RAINFALL
#if len(tttk)>0:assoc.append((k,tkey,len(tttk)))
#else:assoc.append((k,[0],0))
if len(tttk) > 0: assoc.append((k, tkey[-1], len(tttk), col))
else: assoc.append((k, 0, 0, 0))
assoc = np.asarray(assoc, dtype=np.int32)
assocTTT = assoc[assoc[:, 1].nonzero()[0], :]
assocTCOL = assoc[assoc[:, 3].nonzero()[0], :]
print assoc.shape, assocTTT.shape, assocTCOL.shape
return extrm_ixt, extrmevents, assocTTT, assocTCOL