if plotseacycmag: # only for CanESM2 SST
maxfld = np.max(plotflds,axis=0)
maxidx = np.argmax(plotflds,axis=0)
minfld = np.min(plotflds,axis=0)
minfld2 = np.min(ma.masked_where(gctsoneyr==1,plotflds),axis=0)
minidx = np.argmin(plotflds,axis=0)
janfld = plotflds[0,:,:]
julfld = plotflds[6,:,:]
seacycmon = julfld - janfld
seacycmag = maxfld - minfld
# PLOT SEASONAL CYCLE MAG as a map
figa,axs = plt.subplots(1,2)
bm,pc = cplt.kemmap(seacycmon,lat,lon,cmin=cminscm,cmax=cmaxscm,cmap=cmap,
type=mtype,axis=axs[0],suppcb=1,lmask=1,flipmask=flipmask)
axs[0].set_title('Jul - Jan')
bm,pc = cplt.kemmap(seacycmag,lat,lon,cmin=cminscm,cmax=cmaxscm,cmap=cmap,
type=mtype,axis=axs[1],suppcb=1,lmask=1,flipmask=flipmask)
axs[1].set_title('Max - Min')
cbar_ax = figa.add_axes([.91,.25,.02,.5])
figa.colorbar(pc,cax=cbar_ax)
lons, lats = np.meshgrid(lon,lat)
## # PLOT MINIMUM GT: shouldn't get below freezing in the sea (GC=0)
## figb,axs = plt.subplots(1,1)
## bm,pc = cplt.kemmap(minfld2,lat,lon,cmin=238,cmax=308,cmap='blue2red_20',
## type=mtype,axis=axs,lmask=1,flipmask=flipmask)
## bm.contour(lons,lats,minfld2,levels=[273, 273],colors='k',linewidths='2',latlon=True)
# pdf
# standard dev
# # # ################## Make Figures #########################
# Maps
# control ANN climo (whole timeseries)
# GLOBE
plotfld = np.mean(anntsc, axis=0)
plt.figure()
cplt.kemmap(
plotfld, lat, lon, type="sq", cmap=cmapclimo, cmin=cminclimo, cmax=cmaxclimo, title=field + " ANN mean " + casename
)
if printtofile:
plt.savefig(field + "_ANNclimo_" + casename + ".pdf")
plt.savefig(field + "_ANNclimo_" + casename + ".png")
# POLE
plt.figure()
cplt.kemmap(
plotfld, lat, lon, type="nh", cmap=cmapclimo, cmin=cminclimo, cmax=cmaxclimo, title=field + " ANN mean " + casename
)
if printtofile:
plt.savefig(field + "_ANNclimo_" + casename + "_nh.pdf")
plt.savefig(field + "_ANNclimo_" + casename + "_nh.png")
# annual mean diffs, all sims
print fldcgclim.shape lons, lats = np.meshgrid(lon, lat) plotfldg = fldcgclim[2, :, :] # test Jan sst plotflds = fldcsclim[2, :, :] # test Jan sicn plotflds = ma.masked_where(plotflds <= 0, plotflds) # fig = plt.figure() # cplt.kemmap(plotfldg,lat,lon,cmin=cmincg,cmax=cmaxcg,cmap=cmapg) fig2 = plt.figure() bm, pc = cplt.kemmap(plotfldg, lat, lon, cmin=cmincnhg, cmax=cmaxcnhg, cmap=cmapg, type="nh") # cf = bm.contour(lons,lats,plotflds,colors='k',latlon=True) cplt.kemmap(plotflds, lat, lon, cmin=cmincs, cmax=cmaxcs, cmap=cmaps, type="nh", suppcb=1) cf2 = bm.contour(lons, lats, plotflds, colors="k", latlon=True, linewidths="2", levels=[15, 15]) # fig3 = plt.figure() # cplt.kemmap(plotflds,lat,lon,cmin=cmincs,cmax=cmaxcs,cmap=cmaps,type='nh') months = con.get_mon() title = casename + " " + fields + " and " + fieldg fig3, spax = plt.subplots(2, 6) fig3.set_size_inches(12, 6) fig3.subplots_adjust(hspace=0, wspace=0)
lons, lats = np.meshgrid(lon,lat)
cmlen=15.
incr = (cmaxsp2-cminsp2) / (cmlen)
conts = np.arange(cminsp2,cmaxsp2+incr,incr)
ttl1=seasp + ' regress on ' + sear + ' BKS Z500'
ttl2=seasp + ' regress on ' + sear + ' Eur SAT'
#ttl1=ttl2=''
fig,axs=plt.subplots(1,2)
fig.set_size_inches(10,5)
fig.subplots_adjust(wspace=0.05)
ax=axs[0]
bm,pc=cplt.kemmap(fldsponfldr,lat,lon,ptype='nheur',axis=ax,cmin=cminsp,cmax=cmaxsp,
title=ttl1,suppcb=True,
panellab='a',lcol='0.2')
bm.contour(lons,lats,fldsp2onfldr,levels=conts,
colors='0.5',linewidths=1,latlon=True)
ax=axs[1]
bm,pc=cplt.kemmap(fldsponfldr2,lat,lon,ptype='nheur',axis=ax,cmin=cminsp,cmax=cmaxsp,
title=ttl2,suppcb=True,
panellab='b',lcol='0.2')
bm.contour(lons,lats,fldsp2onfldr2,levels=conts,
colors='0.5',linewidths=1,latlon=True)
cplt.add_colorbar(fig,pc,orientation='horizontal')
if printtofile:
def plot_nc_agreement_map(ncdt, ax, lin='one',magtype='abs',cmin='',cmax='',cmind='',cmaxd='',cmin2='',cmax2='',
conv=1, ptype='nh',suppcb=False,cmap='blue2red_w20',subtime=None,
screen=False,vert=False,levlim=None,suppttl=False,addsig=False,sigtype='cont',
latlim=None,vertptype=None, fsz=None,plab=None):
"""
cmin/cmax: for ice
lin = 'one' is ice and co2 (ICEcold, CO2hi)
= 'two' is ice2 and co22 (ICEwarm, CO2lo)
= 'sub' subtract lin 'one' from lin 'two' (shows how warm clim/low ice different from cold/high)
fsz: fontsize for title
returns: plot handle (ph)
"""
mag,magsign = calc_nc_agreement_map(ncdt, axs, lin=lin,magtype=magtype,
conv=1, subtime=None)
# @@ subtime not implemented yet. assume average over full time
#tmplen = ncdt['2xco2preiice'].shape[0]-1
fmt='%2.1f' # clabel format
#if ctlconts!=None:
# if (ctlconts < 1.).any() and (ctlconts > -1.).any():
# #if np.logical_and((ctlconts<1).any(),(ctlconts>-1).any()):
# print ctlconts
# fmt='%2.1f'
if lin=='one':
suff=''
suff1='cold'; suff2='hi'
elif lin=='two':
suff='2'
suff1='warm'; suff2='lo'
# pparams for ice!
# pparams2 for co2![sum|full]
if vert:
pparams = {'cmin': cmin, 'cmax': cmax,
'suppcb': suppcb, 'screen': screen, 'cmap':cmap,'levlim':levlim,
'latlim':latlim,'ptype':vertptype}
else:
pparams = {'ptype': ptype, 'cmin': cmin, 'cmax': cmax,
'suppcb': suppcb, 'cmap':cmap}
if suppttl: ttl=''
else: ttl='ICE'+suff1 +' & CO2'+suff2
if vert:
ph = cplt.vert_plot(magsign,lev,lat,axis=ax,title=ttl,**pparams)
#if addsig:
# cplt.addtsig(ax,icepv,lat,lev/100.,sigtype=sigtype,reverse=sigreverse)
#if addclimcont:
# cplt.add_contoursvert(ax,ctl,lat,lev,verb=True,clab=True,conts=ctlconts,fmt=fmt)
else:
bm,ph = cplt.kemmap(magsign,lat,lon,axis=ax,title=ttl,**pparams)
#if addsig:
# cplt.addtsigm(bm,icepv,lat,lon,sigtype=sigtype,reverse=sigreverse)
if fsz!=None:
ax.set_title(ttl,fontsize=fsz)
if plab!=None:
ax.annotate(plab, xy=(0.02,1.02),xycoords='axes fraction',
fontsize=fsz-1,fontweight='bold')
return ph
def plot_nc_linearity_maps(ncdt, axs, lin='one',cmin='',cmax='',cmind='',cmaxd='',cmin2='',cmax2='',
conv=1, ptype='nh',suppcb=False,cmap='blue2red_w20',subtime=None,
screen=False,vert=False,levlim=None,suppttl=False,addsig=False,sigtype='cont',
latlim=None,vertptype=None,addclimcont=False,ctlconts=None,
nosum=False,nofull=False,nolin=False,scaledlin=False,ncicedt=None,sigreverse=False,
northof=0,fsz=None):
"""
cmin/cmax: for ice
cmin2/cmax2: for co2 & sum & full, if given. otherwise same as ice
cmind/cmaxd: for lin combo subtraction from full
lin = 'one' is ice and co2 (ICEcold, CO2hi)
= 'two' is ice2 and co22 (ICEwarm, CO2lo)
= 'sub' subtract lin 'one' from lin 'two' (shows how warm clim/low ice different from cold/high)
nosum: include ice+co2 panel or not?
nofull: include Full panel or not?
nolin: include non-linearity panel or not?
scaledlin: is the non-linearity panel to be scaled or not?
ncicedt: must be set if scaledlin=True!
sigreverse: if True, plot hatching/contours where NOT significant on
every map but non-lin. (only if addsig=True)
northof: compute pattern correlation b/w SUM and FULL north of this latitude.
fsz: fontsize for title
returns: plot handles (ph,ph2[,phd]) (ice,co2|sum|full[,difference/nonlin])
"""
# @@ subtime not implemented yet. assume average over full time
#tmplen = ncdt['2xco2preiice'].shape[0]-1
fmt='%2.1f' # clabel format
#if ctlconts!=None:
# if (ctlconts < 1.).any() and (ctlconts > -1.).any():
# #if np.logical_and((ctlconts<1).any(),(ctlconts>-1).any()):
# print ctlconts
# fmt='%2.1f'
pico = ncdt['prei2xco2iceb']
pipi = ncdt['preipreiice']
copi = ncdt['2xco2preiice']
coco = ncdt['2xco22xco2ice']
lat = ncdt['lat']
if lin=='one':
_,icepv = cutl.ttest_ind(pico, pipi,axis=0,effdof=False)
ice = (np.mean(pico,axis=0) - np.mean(pipi,axis=0))*conv
#icep = ncfldzmdt['pi2xco2ipulse'] - ncfldzmdt['preipreiice']
_,co2pv = cutl.ttest_ind(copi, pipi,axis=0,effdof=False)
co2 = (np.mean(copi,axis=0) - np.mean(pipi,axis=0))*conv
_,combopv = cutl.ttest_ind(pico-pipi+(copi-pipi),
coco-pipi)
ctl = np.mean(pipi,axis=0)*conv
suff=''
suff1='cold'; suff2='hi'
elif lin=='two':
_,icepv = cutl.ttest_ind(coco, copi,axis=0,effdof=False)
ice = (np.mean(coco,axis=0) - np.mean(copi,axis=0))*conv
_,co2pv = cutl.ttest_ind(coco, pico,axis=0,effdof=False)
co2 = (np.mean(coco,axis=0) - np.mean(pico,axis=0))*conv
_,combopv = cutl.ttest_ind(coco-copi+(coco-pico),
coco-pipi)
ctl = np.mean(coco,axis=0)*conv
suff='2'
suff1='warm'; suff2='lo'
elif lin=='sub':
#two - one: (coco-copi) - (pico-pipi)
_,icepv = cutl.ttest_ind(coco-copi, pico-pipi,axis=0,effdof=False)
ice = (np.mean(coco-copi,axis=0) - np.mean(pico-pipi,axis=0))*conv
#two - one: (coco-pico) - (copi-pipi)
_,co2pv = cutl.ttest_ind(coco-pico, copi-pipi,axis=0,effdof=False)
co2 = (np.mean(coco-pico,axis=0) - np.mean(copi-pipi,axis=0))*conv
suff='2-1'
suff1='warm-cold'; suff2='lo-hi'
_,fullpv = cutl.ttest_ind(coco, pipi,axis=0,effdof=False)
full = (np.mean(coco,axis=0) - np.mean(pipi,axis=0))*conv
# compute pattern correlation b/w SUM and FULL
prval,ppval = calc_linpatterncorr(ncdt,lin=lin,northof=northof,vert=vert)
pvarexp = (prval**2)*100
# pparams for ice!
# pparams2 for co2![sum|full]
if vert:
lev = ncdt['lev']
pparams = {'cmin': cmin, 'cmax': cmax,
'suppcb': suppcb, 'screen': screen, 'cmap':cmap,'levlim':levlim,
'latlim':latlim,'ptype':vertptype}
if cmin2 != '':
pparams2 = {'cmin': cmin2, 'cmax': cmax2,
'suppcb': suppcb, 'screen': screen, 'cmap':cmap,'levlim':levlim,
'latlim':latlim,'ptype':vertptype}
else:
pparams2 = {'cmin': cmin, 'cmax': cmax,
'suppcb': suppcb, 'screen': screen, 'cmap':cmap,'levlim':levlim,
'latlim':latlim,'ptype':vertptype}
else:
lon = ncdt['lon']
pparams = {'ptype': ptype, 'cmin': cmin, 'cmax': cmax,
'suppcb': suppcb, 'cmap':cmap}
if cmin2 != '':
pparams2 = {'ptype': ptype, 'cmin': cmin2, 'cmax': cmax2,
'suppcb': suppcb, 'cmap':cmap}
else:
pparams2 = {'ptype': ptype, 'cmin': cmin, 'cmax': cmax,
'suppcb': suppcb, 'cmap':cmap}
aii=0
ax=axs[aii]
if suppttl: ttl=''
else: ttl='ICE'+suff1
if vert:
ph = cplt.vert_plot(ice,lev,lat,axis=ax,title=ttl,**pparams)
if addsig:
cplt.addtsig(ax,icepv,lat,lev/100.,sigtype=sigtype,reverse=sigreverse)
if addclimcont:
cplt.add_contoursvert(ax,ctl,lat,lev,verb=True,clab=True,conts=ctlconts,fmt=fmt)
else:
bm,ph = cplt.kemmap(ice,lat,lon,axis=ax,title=ttl,**pparams)
if addsig:
cplt.addtsigm(bm,icepv,lat,lon,sigtype=sigtype,reverse=sigreverse)
if fsz!=None:
ax.set_title(ttl,fontsize=fsz)
aii+=1
ax=axs[aii]
if suppttl: ttl=''
else: ttl='CO2'+suff2
if vert:
ph2 = cplt.vert_plot(co2,lev,lat,axis=ax,title=ttl,suppylab=True,**pparams2)
if addsig:
cplt.addtsig(ax,co2pv,lat,lev/100.,sigtype=sigtype,reverse=sigreverse)
if addclimcont:
cplt.add_contoursvert(ax,ctl,lat,lev,verb=True,clab=True,conts=ctlconts,fmt=fmt)
else:
bm,ph2 = cplt.kemmap(co2,lat,lon,axis=ax,title=ttl,**pparams2)
if addsig:
cplt.addtsigm(bm,co2pv,lat,lon,sigtype=sigtype,reverse=sigreverse)
if fsz!=None:
ax.set_title(ttl,fontsize=fsz)
aii+=1
if not nosum:
ax=axs[aii]
if suppttl: ttl=''
else: ttl='Sum' #ice'+suff1+'+co2'+suff2
if vert:
ph2 = cplt.vert_plot(ice+co2,lev,lat,axis=ax,title=ttl,suppylab=True,**pparams2)
if addclimcont:
cplt.add_contoursvert(ax,ctl,lat,lev,verb=True,clab=True,conts=ctlconts,fmt=fmt)
else:
bm,ph2 = cplt.kemmap(ice+co2,lat,lon,axis=ax,title=ttl,**pparams2)
if fsz!=None:
ax.set_title(ttl,fontsize=fsz)
aii+=1
if not nofull:
ax=axs[aii]
if suppttl: ttl=''
else: ttl='Full'
if vert:
ph2 = cplt.vert_plot(full,lev,lat,axis=ax,title=ttl,suppylab=True,**pparams2)
if addsig:
cplt.addtsig(ax,fullpv,lat,lev/100.,sigtype=sigtype,reverse=sigreverse)
if addclimcont:
cplt.add_contoursvert(ax,ctl,lat,lev,verb=True,clab=True,conts=ctlconts,fmt=fmt)
else:
bm,ph2 = cplt.kemmap(full, lat,lon,axis=ax,title=ttl,**pparams2)
if addsig:
cplt.addtsigm(bm,fullpv,lat,lon,sigtype=sigtype,reverse=sigreverse)
ax.annotate('$%.0f$'%(pvarexp) + '%', xy=(0.78,0.98),xycoords='axes fraction',
fontsize=16)
if fsz!=None:
ax.set_title(ttl,fontsize=fsz)
aii+=1
if not nolin:
ax=axs[aii] # the linearity test
# @@@ add scaledlin here @@@
# e.g. c1*ICEcold + c2*CO2hi - Full
# c1*(pico-pipi) + c2*(copi-pipi) - (coco-pipi)
# solve c1, c2 with sea ice s.t. above eqn would give 0
# @@@ Question: how to scale? how to get c1,c2? Use difference from target sea ice?
if scaledlin:
# compute target offset
offs = compute_targoffset(ncicedt,dosia=True)
# want to be a fraction so div by 100.
# and for ice comparisons, we don't reach target so we need to bump it up (add 1)
c1 = offs['ice'+suff1]/100. + 1
# for co2 comparison, not sure how to scale, b/c want to remove ice effect, which isn't a scaling?
c2 = offs['co2'+suff2]/100.
if suppttl: ttl=''
else: ttl='Full-Sum' #ttl='(ice'+suff +'+co2'+suff+')-full'
if cmind=='': # just divide the other clims by 2
if vert:
phd=cplt.vert_plot(full-(ice+co2),lev,lat,axis=ax,title=ttl,
cmin=cmin/2,cmax=cmax/2,suppcb=suppcb,screen=screen,cmap=cmap,
suppylab=True,levlim=levlim,latlim=latlim,ptype=vertptype)
if addsig:
cplt.addtsig(ax,combopv,lat,lev,sigtype=sigtype)
if addclimcont: # want in the linearity fig?? @@@@
cplt.add_contoursvert(ax,ctl,lat,lev,verb=True,clab=True,conts=ctlconts,fmt=fmt)
else:
bm,phd=cplt.kemmap(full-(ice+co2),lat,lon,axis=ax,title=ttl,ptype=ptype,
cmin=cmin/2,cmax=cmax/2,suppcb=suppcb,cmap=cmap)
if addsig:
cplt.addtsigm(bm,combopv,lat,lon,sigtype=sigtype)
else:
if vert:
phd=cplt.vert_plot(full-(ice+co2),lev,lat,axis=ax,title=ttl,
cmin=cmind,cmax=cmaxd,suppcb=suppcb,screen=screen,cmap=cmap,
suppylab=True,levlim=levlim,latlim=latlim,ptype=vertptype)
if addsig:
cplt.addtsig(ax,combopv,lat,lev,sigtype=sigtype)
if addclimcont: # want in the linearity fig?? @@@@
cplt.add_contoursvert(ax,ctl,lat,lev,verb=True,clab=True,conts=ctlconts,fmt=fmt)
else:
bm,phd=cplt.kemmap(full-(ice+co2),lat,lon,axis=ax,title=ttl,ptype=ptype,
cmin=cmind,cmax=cmaxd,suppcb=suppcb,cmap=cmap)
if addsig:
cplt.addtsigm(bm,combopv,lat,lon,sigtype=sigtype)
if fsz!=None:
ax.set_title(ttl,fontsize=fsz)
return ph,ph2,phd
else:
return ph,ph2