def map_lakepct(var,
lon,
lat,
levels,
title_str,
clrmap,
fig_name,
alone=True,
ax=False):
# figure creation
if alone:
fig, ax = plt.subplots(1,
1,
figsize=(13, 8),
subplot_kw={'projection': ccrs.PlateCarree()})
ax.add_feature(ctp.feature.OCEAN, color='gainsboro')
ax.coastlines(color='darkgrey')
#ax.coastlines(color="grey")
LON, LAT = mpu.infer_interval_breaks(lon, lat)
# add the data to the map (more info on colormaps: https://matplotlib.org/users/colormaps.html)
cmap, norm = mpu.from_levels_and_cmap(levels, clrmap)
# do not show zeros
var[var == 0] = np.nan
cmap.set_over(cmap(0.99))
h = ax.pcolormesh(LON, LAT, var, cmap=cmap, norm=norm)
# set the extent of the cartopy geoAxes to "global"
ax.set_global()
ax.set_title(title_str, pad=10, loc='right')
# remove the frame
ax.outline_patch.set_visible(False)
# plot the colorbar
cbar = mpu.colorbar(h, ax, orientation='horizontal', pad=0.1, extend='max')
cbar.ax.set_xlabel('Natural lake area fraction [%]', size=16)
cbar.set_ticks(np.arange(0, 110, 10)) # horizontal colorbar
#cbar.ax.set_xticklabels(['Low', 'Medium', 'High']) # horizontal colorbar
# if so, save the figure
if fig_name != 0 and alone:
plt.savefig(fig_name + '.jpeg', dpi=1000, bbox_inches='tight')
def plot_region_hc_map(var, region_props, lakes_path, indir_lakedata):
# get region specific info from dictionary
extent = region_props['extent']
continent_extent = region_props['continent_extent']
name = region_props['name']
name_str = region_props['name_str']
ax_location = region_props['ax_location']
levels = region_props['levels']
fig_size = region_props['fig_size']
cb_orientation = region_props['cb_orientation']
path_lakes = lakes_path + name + '.shp'
# settings
clb_label = 'Joule'
title_str = name_str + ' heat content anomaly'
fig_name = 'Heat_content_' + name
cmap = 'YlOrBr'
cmap, norm = mpu.from_levels_and_cmap(levels, cmap, extend='max')
lon, lat = get_lonlat(indir_lakedata)
LON, LAT = mpu.infer_interval_breaks(lon, lat)
lakes = gpd.read_file(path_lakes)
# plotting
fig, ax = plt.subplots(1,
1,
figsize=fig_size,
subplot_kw={'projection': ccrs.PlateCarree()})
ax.add_feature(ctp.feature.OCEAN, color='gainsboro')
ax.coastlines(color="grey")
# add the data to the map (more info on colormaps: https://matplotlib.org/users/colormaps.html)
h = ax.pcolormesh(LON, LAT, var, cmap=cmap, norm=norm)
# load the lake shapefile
lakes.plot(ax=ax, edgecolor='gray', facecolor='none')
# set grid lines
gl = ax.gridlines(crs=ccrs.PlateCarree(),
draw_labels=True,
linewidth=0.5,
color='gainsboro',
alpha=0.5)
gl.xlines = True
gl.xformatter = LONGITUDE_FORMATTER
gl.yformatter = LATITUDE_FORMATTER
gl.xlabels_bottom = None
gl.ylabels_right = None
# set extent (in right way)
extent[1], extent[2] = extent[2], extent[1]
ax.set_extent(extent)
# create effect for map borders:
effect = Stroke(linewidth=1.5, foreground='darkgray')
# set effect for main ax
ax.outline_patch.set_path_effects([effect])
# Create an inset GeoAxes showing the location of the lakes region
#x0 y0 width height
sub_ax = fig.add_axes(ax_location, projection=ccrs.PlateCarree())
sub_ax.set_extent(continent_extent)
#lakes.plot(ax=sub_ax)
sub_ax.outline_patch.set_path_effects([effect])
extent_box = sgeom.box(extent[0], extent[2], extent[1], extent[3])
sub_ax.add_geometries([extent_box],
ccrs.PlateCarree(),
facecolor='none',
edgecolor='red',
linewidth=2)
# Add the land, coastlines and the extent of the inset axis
sub_ax.add_feature(cfeature.LAND, edgecolor='gray')
sub_ax.coastlines(color='gray')
extent_box = sgeom.box(extent[0], extent[2], extent[1], extent[3])
sub_ax.add_geometries([extent_box],
ccrs.PlateCarree(),
facecolor='none',
edgecolor='black',
linewidth=2)
# plot the colorbar
cbar = mpu.colorbar(h,
ax,
extend='max',
orientation=cb_orientation,
pad=0.05)
if cb_orientation == 'vertical':
cbar.ax.set_ylabel(clb_label, size=16)
elif cb_orientation == 'horizontal':
cbar.ax.set_xlabel(clb_label, size=16)
#ax.set_title(title_str, pad=10)
plotdir = '/home/inne/documents/phd/data/processed/isimip_lakeheat/plots/'
plt.savefig(plotdir + fig_name + '.png', dpi=500)
def plot_global_hc_map(name_str, var, lakes_path, indir_lakedata):
# get region specific info from dictionary
if name_str == 'global_absolute':
levels = np.arange(-1e17, 1.1e17, 0.1e17)
elif name_str == 'global':
levels = np.arange(-1e19, 1.1e19, 0.1e19)
cb_orientation = 'horizontal'
path_lakes = lakes_path + name_str + '.shp'
# settings
clb_label = 'Joule'
title_str = name_str + ' heat content anomaly'
fig_name = 'Heat_content_' + name_str
cmap = 'RdBu_r' #, 'YlOrBr'
cmap, norm = mpu.from_levels_and_cmap(levels, cmap, extend='max')
lon, lat = get_lonlat(indir_lakedata)
LON, LAT = mpu.infer_interval_breaks(lon, lat)
# plotting
fig, ax = plt.subplots(1,
1,
figsize=(13, 8),
subplot_kw={'projection': ccrs.PlateCarree()})
ax.add_feature(ctp.feature.OCEAN, color='gainsboro')
ax.coastlines(color="grey")
# add the data to the map (more info on colormaps: https://matplotlib.org/users/colormaps.html)
h = ax.pcolormesh(LON, LAT, var, cmap=cmap, norm=norm)
# set grid lines
gl = ax.gridlines(crs=ccrs.PlateCarree(),
draw_labels=True,
linewidth=0.5,
color='gainsboro',
alpha=0.5)
gl.xlines = True
gl.xformatter = LONGITUDE_FORMATTER
gl.yformatter = LATITUDE_FORMATTER
gl.xlabels_bottom = None
gl.ylabels_right = None
effect = Stroke(linewidth=1.5, foreground='darkgray')
# set effect for main ax
ax.outline_patch.set_path_effects([effect])
# plot the colorbar
cbar = mpu.colorbar(h,
ax,
extend='max',
orientation=cb_orientation,
pad=0.05)
if cb_orientation == 'vertical':
cbar.ax.set_ylabel(clb_label, size=16)
elif cb_orientation == 'horizontal':
cbar.ax.set_xlabel(clb_label, size=16)
#ax.set_title(title_str, pad=10)
plotdir = '/home/inne/documents/phd/data/processed/isimip_lakeheat/plots/'
plt.savefig(plotdir + fig_name + '.png')
def load_data_single_mod(gen,model,scenario,Tanglob_idx=False,Tref_all=True,Tref_start='1870-01-01',Tref_end='1900-01-01',usr_time_res="ann",var='tas'):
""" Load the all initial-condition members of a single model in cmip5 or cmip6 for given scenario plus associated historical period.
Keyword argument:
- gen: generation (cmip5 = 5 and cmip6 = 6 are implemented)
- model: model str
- scenario: scenario str
- Tanglob_idx: decides if wgt Tanglob is computed (and returned) or not, default is not returned
- Tref_all: decides if the Tref at each grid point is dervied based on all available runs or not, default is yes
- Tref_start: starting point for the reference period with default 1870
- Tref_end: first year to no longer be included in reference period with default 1900
Output:
- y: the land grid points of the anomalies of the variable on grid centered over 0 longitude (like the srexgrid)
- time: the time slots
- srex: the gridded srex regions
- df_srex: data frame containing the shape files of the srex regions
- lon_pc: longitudes for pcolormesh (needs 1 more than on grid)
- lat_pc: latitudes for pcolormesh (needs 1 more than on grid)
- idx_l: array with 0 where sea, 1 where land (assumption: land if frac land > 0)
- wgt_l: land grid point weights to derive area weighted mean vars
- Tan_wgt_globmean = area weighted global mean temperature
"""
# the dictionaries are NOT ordered properly + some other adjustments -> will need to be careful with my old scripts
# see e-mail from Verena on 20191112 for additional infos how could read in several files at once with xarr
# additionally: she transforms dataset into dataarray to make indexing easier -> for consistency reason with earlier
# version of emulator (& thus to be able to reuse my scripts), I do not do this (fow now).
# right now I keep reloading constants fields for each run I add -> does not really make sense.
# Maybe add boolean to decide instead. however they are small & I have to read them in at some point anyways
# -> maybe path of least resistence is to not care about it
print('start with model',model)
# vars which used to be part of the inputs but did not really make sense as I employ the same ones all the time anyways (could be changed later if needed)
#var='tas'
temp_res = usr_time_res # if not, reading the var file needs to be changed as time var is not named in the same way anymore
spatial_res = 'g05'
# load in the constants files
dir_data = '/net/so4/landclim/snath/data/'
file_ls = 'interim_invariant_lsmask_regrid_g05.nc' # ERA-interim mask regridded by Richard from 73x144 to 72x144
file_srex = 'srex-region-masks_20120709.srex_mask_SREX_masks_all.50deg.time-invariant.nc'
file_srex_shape = 'referenceRegions.shp'
#df_all_regs = geopandas.read_file(dir_data+file_srex_shape)
srex_names = ['ALA','CGI','WNA','CNA','ENA','CAM','AMZ','NEB','WSA','SSA','NEU','CEU','MED','SAH','WAF','EAF','SAF',
'NAS','WAS','CAS','TIB','EAS','SAS','SEA','NAU','SAU'] # SREX names ordered according to SREX mask I am
# employing
#df_srex = df_all_regs.loc[df_all_regs['LAB'].isin(srex_names)] # alternative indexing: search in column LAB for names
srex_raw = xr.open_mfdataset(dir_data+file_srex, combine='by_coords',decode_times=False) # srex_raw nrs from 1-26
#df_srex=srex_raw
#srex_raw["time"]=pd.to_datetime(srex_raw.time.values)
frac_l = xr.open_mfdataset(dir_data+file_ls, combine='by_coords',decode_times=False) #land-sea mask of ERA-interim bilinearily interpolated
lons, lats = np.meshgrid(frac_l.longitude.values,frac_l.latitude.values) # the lon, lat grid (just to derive weights)
#frac_1["time"]=pd.to_datetime(frac_1.time.values)
frac_l = frac_l.where(frac_l.latitude>-60,0) # remove Antarctica from frac_l field (ie set frac l to 0)
idx_l=np.squeeze(frac_l.lsm.values)>0.0 # idex land #-> everything >0 I consider land
wgt = norm_cos_wgt(lats) # area weights of each grid point
wgt_l = wgt[idx_l] # area weights for land grid points
lon_pc, lat_pc = mpu.infer_interval_breaks(frac_l.longitude, frac_l.latitude) # the lon / lat for the plotting with pcolormesh
srex=(np.squeeze(srex_raw.srex_mask.values)-1)[idx_l] # srex indices on land
y={}
T_ref = np.zeros([36,72])
run_nrs={}
if Tanglob_idx == True:
Tan_wgt_globmean = {}
if gen == 5:
dir_var='/net/so4/landclim/snath/DA/data/%s/regridded/'%model
run_names_list=sorted(glob.glob(dir_var+var+'_'+temp_res+'_'+model+'_'+scenario+'_'+'r*i1p1'+'_'+spatial_res+'.nc'))
# ATTENTION: are ordered but does not work for models with runs above digit 9
#index_tr = [i for i, s in enumerate(run_names_list) if 'r1i1p1' in s][0] # find training run
#print(run_names_list)
#run_names_list.insert(0, run_names_list.pop(index_tr)) # move training run to begin of list
# exception for cmip5 GISS-E2-H_rcp85_r2i1p1 ie the strange run I excluded from ESD paper
if '/net/atmos/data/cmip5-ng/tas/tas_%s_GISS-E2-H_rcp85_r2i1p1_g025.nc'%usr_time_res in run_names_list:
run_names_list.remove('/net/atmos/data/cmip5-ng/tas/tas_%s_GISS-E2-H_rcp85_r2i1p1_g025.nc'%usr_time_res )
# loop over all runs to obtain the absolute values
print(run_names_list)
for run_name in run_names_list:
data = xr.open_mfdataset(run_name,decode_times=False)
if usr_time_res=="ann":
data=data.rename({'year':'time'})
data=data.rename({'longitude':'lon'})
data=data.rename({'latitude':'lat'})
data["time"]=cf_units.num2date(data.time.values, 'days since 1800-01-01 00:00:00', cf_units.CALENDAR_STANDARD)
data=data.sel(time=slice('1870-01-01', '2101-01-01'))#.roll(lon=72)
# rename so it is consisten with cmip6
# roll so that it is on same grid as others (no longer Pacific centered)
#print(data.time.values)
data = data.assign_coords(lon= (((data.lon + 180) % 360) - 180)) # assign_coords so same labels as others
run=int(data.attrs['source_ensemble'].split('r')[1].split('i')[0]) # extract ens member
run_nrs[run_name]=run
y[run] = data.tas.values # still absolute values + still contains sea pixels
T_ref += data.tas.sel(time=slice(Tref_start, Tref_end)).mean(dim='time').values*1.0/len(run_names_list) # sum up all ref climates
if run==1 and Tref_all != True:
T_ref_1=data.tas.sel(time=slice(Tref_start, Tref_end)).mean(dim='time').values
print('create ref for ',run_name)
if Tanglob_idx == True:
Tan_wgt_globmean[run] = np.asarray([np.average(y[run][t],weights=wgt) for t in np.arange(y[run].shape[0])]) #area weighted but abs values
if gen == 6:
dir_var = '/net/so4/landclim/snath/DA/data/%s/regridded/'%model
run_names_list=sorted(glob.glob(dir_var+var+'_'+temp_res+'_'+model+'_'+scenario+'_'+'r*i1p1f*'+'_'+spatial_res+'.nc'))
# ATTENTION: are ordered but does not work for models with runs above digit 9
# idea is: every ssp one needs a corresponding hist one (vice versa not the case)
#index_tr = [i for i, s in enumerate(run_names_list) if 'r1i1p1' in s][0] # find training run
#run_names_list.insert(0, run_names_list.pop(index_tr)) # move training run to begin of list
if model == 'EC-Earth3': # temporary fix added on 20191118 because these runs are not available as historical ones
run_names_list.remove('/net/cfc/cmip6/Next_Generation/tas/%s/g025/tas_%s_EC-Earth3_ssp585_r1i1p1f1_g025.nc'%(usr_time_res,usr_time_res))
#run_names_list.remove('/net/cfc/cmip6/Next_Generation/tas/ann/g025/tas_ann_EC-Earth3_ssp585_r6i1p1f1_g025.nc') #6 removed because historical seems faulty (ie only ghg driven) on 20191119
#run_names_list.remove('/net/cfc/cmip6/Next_Generation/tas/ann/g025/tas_ann_EC-Earth3_ssp585_r11i1p1f1_g025.nc')
#run_names_list.remove('/net/cfc/cmip6/Next_Generation/tas/ann/g025/tas_ann_EC-Earth3_ssp585_r13i1p1f1_g025.nc')
#run_names_list.remove('/net/cfc/cmip6/Next_Generation/tas/ann/g025/tas_ann_EC-Earth3_ssp585_r15i1p1f1_g025.nc')
print('ATTENTION: ssp realizations 1 because no historical available on 20191122 (open issue on ESGF).')
print(run_names_list)
for run_name in run_names_list:
#run_name_ssp = run_name
#run_name_hist = run_name.replace(scenario,'historical')
run_name_hist = run_name
run_name_ssp = dir_var+var+'_%s_'%(usr_time_res)+model+'_ssp370_r1i1p1f1_g05.nc'
if model=='HadGEM3-GC31-LL':
run_name_ssp = dir_var+var+'_%s_'%(usr_time_res)+model+'_ssp245_r1i1p1f3_g05.nc'
if model=='CNRM-CM6-1':
run_name_ssp = dir_var+var+'_%s_'%(usr_time_res) +model+'_ssp370_r1i1p1f2_g05.nc'
if model=='GISS-E2-1-G':
run_name_ssp = dir_var+var+'_%s_'%(usr_time_res) +model+'_ssp370_r6i1p1f2_g05.nc'
data = xr.open_mfdataset([run_name_hist,run_name_ssp],concat_dim='time').sel(time=slice('1944-01-01', '2019-01-01'))
#data = data.assign_coords(lon= (((data.lon + 180) % 360) - 180)) # assign_coords so same labels as others
run = int(run_name_hist.split('_')[4].split('r')[1].split('i')[0])#data.attrs['realization_index']
run_nrs[run_name]=run
y[run] = data[var].values # still absolute values + still contains sea pixels
T_ref += data[var].sel(time=slice(Tref_start, Tref_end)).mean(dim='time').values*1.0/len(run_names_list) # sum up all ref climates
if run==1 and Tref_all != True:
T_ref_1=data[var].sel(time=slice(Tref_start, Tref_end)).mean(dim='time').values
print('create ref for ',run_name)
if model=='CAMS-CSM1-0': # temporary (?) fix added on 20191119 because CAMS-CSM1-0 data are currently only available till 2099 instead of 2100
y[run]=y[run][:-1]
if Tanglob_idx == True:
Tan_wgt_globmean[run] = np.asarray([np.average(y[run][t],weights=wgt) for t in np.arange(y[run].shape[0])]) #area weighted but abs values
# obtain the anomalies
for run_name in run_names_list:
run = run_nrs[run_name]
if Tref_all == True:
y[run]=(y[run]-T_ref)
if Tanglob_idx == True:
Tan_wgt_globmean[run]=Tan_wgt_globmean[run]-np.average(T_ref,weights=wgt)
else:
y[run]=y[run]
if Tanglob_idx == True:
Tan_wgt_globmean[run]=Tan_wgt_globmean[run]-np.average(T_ref_1,weights=wgt)
time=data["time"]
if Tanglob_idx == False:
return y,time,srex,srex_names,lon_pc,lat_pc,idx_l,wgt_l
else:
return y,time,srex,srex_names,lon_pc,lat_pc,idx_l,wgt_l,Tan_wgt_globmean