Emergent constraints on air-sea fluxes
Contents
Emergent constraints on air-sea fluxes¶
%load_ext autoreload
%autoreload 2
from itertools import product
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
import emergent_constraint as ec
import figure_panels
import obs_aircraft
import util
Set up emergent constraint computation¶
air_parms = ec.get_parameters('default')
air_parms
{'ubin': 300.0,
'lbin': 280.0,
'udθ': 10.0,
'lbin_as_upper_bound': True,
'ubin_as_lower_bound': False,
'gradient_lat_range': (-90.0, -45.0),
'flux_lat_range': (-90.0, -45.0),
'flux_memory': 90,
'fit_groups': ('DJF', 'MAMJJASON')}
clobber = False
clobber_deep = False
obj_srf = {}; obj_air = {};
acs = {}; scs = {};
for constraint_type in ['ocean_constraint', 'total_constraint',]:
model_input_lists = ec.get_model_tracer_lists(constraint_type)
obj_srf[constraint_type] = ec.whole_enchilada_srf(**model_input_lists)
obj_air[constraint_type] = ec.whole_enchilada(clobber=clobber_deep, **model_input_lists)
acs[constraint_type] = obj_air[constraint_type].get_ac(**air_parms, clobber=clobber)
scs[constraint_type] = obj_srf[constraint_type].get_sc(clobber=clobber)
computing...
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/CT2017/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/CT2019B/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/s99oc_v2020/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/s99oc_ADJocI40S_v2020/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/s99oc_SOCCOM_v2020/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/CTE2018/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/CTE2020/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/MIROC/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/CAMSv20r1/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/TM5-Flux-m0f/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/TM5-Flux-mmf/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/TM5-Flux-mrf/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/TM5-Flux-mwf/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
done.
writing /glade/work/mclong/so-co2-airborne-obs/cache-emergent-constraint/pickles/c37b8d6f5cf4e06d6d37332ab5b64bca/63a783684bb426f9ddf1f30b6cca551e.pkl
computing...
------------------------------------------------------------
n_stn = 10; n_rep = 2
stn_error = 0.0610 ppm
['DJF: 0.0610', 'MAM: 0.0327', 'JJA: 0.0549', 'SON: 0.0332']
------------------------------------------------------------
SO-SPO seasonal gradient error = 0.0637 ppm
------------------------------------------------------------
------------------------------------------------------------
n_stn = 10; n_rep = 2
stn_error = 0.0610 ppm
['DJF: 0.0610', 'MAM: 0.0327', 'JJA: 0.0549', 'SON: 0.0332']
------------------------------------------------------------
SO-SPO seasonal gradient error = 0.0637 ppm
------------------------------------------------------------
------------------------------------------------------------
n_stn = 10; n_rep = 2
stn_error = 0.0610 ppm
['DJF: 0.0610', 'MAM: 0.0327', 'JJA: 0.0549', 'SON: 0.0332']
------------------------------------------------------------
SO-SPO seasonal gradient error = 0.0637 ppm
------------------------------------------------------------
------------------------------------------------------------
n_stn = 10; n_rep = 2
stn_error = 0.0610 ppm
['DJF: 0.0610', 'MAM: 0.0327', 'JJA: 0.0549', 'SON: 0.0332']
------------------------------------------------------------
SO-SPO seasonal gradient error = 0.0637 ppm
------------------------------------------------------------
------------------------------------------------------------
n_stn = 10; n_rep = 2
stn_error = 0.0610 ppm
['DJF: 0.0610', 'MAM: 0.0327', 'JJA: 0.0549', 'SON: 0.0332']
------------------------------------------------------------
SO-SPO seasonal gradient error = 0.0637 ppm
------------------------------------------------------------
------------------------------------------------------------
n_stn = 10; n_rep = 2
stn_error = 0.0610 ppm
['DJF: 0.0610', 'MAM: 0.0327', 'JJA: 0.0549', 'SON: 0.0332']
------------------------------------------------------------
SO-SPO seasonal gradient error = 0.0637 ppm
------------------------------------------------------------
------------------------------------------------------------
n_stn = 10; n_rep = 2
stn_error = 0.0610 ppm
['DJF: 0.0610', 'MAM: 0.0327', 'JJA: 0.0549', 'SON: 0.0332']
------------------------------------------------------------
SO-SPO seasonal gradient error = 0.0637 ppm
------------------------------------------------------------
------------------------------------------------------------
n_stn = 10; n_rep = 2
stn_error = 0.0610 ppm
['DJF: 0.0610', 'MAM: 0.0327', 'JJA: 0.0549', 'SON: 0.0332']
------------------------------------------------------------
SO-SPO seasonal gradient error = 0.0637 ppm
------------------------------------------------------------
------------------------------------------------------------
n_stn = 10; n_rep = 2
stn_error = 0.0610 ppm
['DJF: 0.0610', 'MAM: 0.0327', 'JJA: 0.0549', 'SON: 0.0332']
------------------------------------------------------------
SO-SPO seasonal gradient error = 0.0637 ppm
------------------------------------------------------------
------------------------------------------------------------
n_stn = 10; n_rep = 2
stn_error = 0.0610 ppm
['DJF: 0.0610', 'MAM: 0.0327', 'JJA: 0.0549', 'SON: 0.0332']
------------------------------------------------------------
SO-SPO seasonal gradient error = 0.0637 ppm
------------------------------------------------------------
------------------------------------------------------------
n_stn = 10; n_rep = 2
stn_error = 0.0610 ppm
['DJF: 0.0610', 'MAM: 0.0327', 'JJA: 0.0549', 'SON: 0.0332']
------------------------------------------------------------
SO-SPO seasonal gradient error = 0.0637 ppm
------------------------------------------------------------
------------------------------------------------------------
n_stn = 10; n_rep = 2
stn_error = 0.0610 ppm
['DJF: 0.0610', 'MAM: 0.0327', 'JJA: 0.0549', 'SON: 0.0332']
------------------------------------------------------------
SO-SPO seasonal gradient error = 0.0637 ppm
------------------------------------------------------------
------------------------------------------------------------
n_stn = 10; n_rep = 2
stn_error = 0.0610 ppm
['DJF: 0.0610', 'MAM: 0.0327', 'JJA: 0.0549', 'SON: 0.0332']
------------------------------------------------------------
SO-SPO seasonal gradient error = 0.0637 ppm
------------------------------------------------------------
------------------------------------------------------------
n_stn = 10; n_rep = 2
stn_error = 0.0610 ppm
['DJF: 0.0610', 'MAM: 0.0327', 'JJA: 0.0549', 'SON: 0.0332']
------------------------------------------------------------
SO-SPO seasonal gradient error = 0.0637 ppm
------------------------------------------------------------
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/CT2017/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/CT2019B/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/s99oc_v2020/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/s99oc_ADJocI40S_v2020/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/s99oc_SOCCOM_v2020/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/CTE2018/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/CTE2020/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/MIROC/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/CAMSv20r1/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/TM5-Flux-m0f/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/TM5-Flux-mmf/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/TM5-Flux-mrf/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/TM5-Flux-mwf/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
done.
writing /glade/work/mclong/so-co2-airborne-obs/cache-emergent-constraint/pickles/af03b30f55a9a2db891b8ab356d03840/67214989ae7b9ac834ff720d5021ed8a.pkl
Fig 3: flux-gradient relationships¶
def plot_flux_v_gradient(constraint_type):
ac = acs[constraint_type]
sc = scs[constraint_type]
fig = plt.figure(figsize=(10, 8)) #dpi=300)
gs = gridspec.GridSpec(
nrows=2, ncols=2,
left=0, right=1,
bottom=0., top=1,
hspace=0.25, wspace=0.2,
)
theta_centers = [np.mean(t) for t in ac.theta_bins]
fit_groups = ac.fit_groups
i_campaign = ac.df_data.xs((fit_groups[0]), level=('campaign_group')).index.names.index('campaign')
assert len(fit_groups) == 2
subplots = ['air_sum', 'air_win', 'srf_sum', 'srf_win',]
nrow = 2
ncol = 2
axs = {
subplots[i*2+j]: plt.subplot(gs[i, j])
for i, j in product(range(nrow), range(ncol))
}
theta_str = figure_panels.theta_bin_def(ac.theta_bins)
if constraint_type == 'ocean_constraint':
flux_type = 'Air-sea'
else:
flux_type = 'Surface'
str_co2 = 'CO$_2^{ocn}$' if constraint_type == 'ocean_constraint' else 'CO$_2$'
kwargs_dict = dict(
srf_sum=dict(
title=f'DJF {flux_type} flux vs. SO – SPO CO$_2$',
xlabel='$\Delta_{ y}$' + str_co2 + ' [ppm]',
df_data=sc.df_data.xs(('1999-2020', 'DJF'), level=['period', 'season']),
fit_dict=ec.get_fit_dict(sc.df_fits.loc[('1999-2020', 'DJF')]),
),
air_sum=dict(
title=f'Dec-Feb: 90-day {flux_type} flux vs. {theta_str} CO$_2$ diff',
xlabel='$\Delta_{ θ}$' + str_co2 + ' [ppm]',
highlight_campaign=['ORCAS-F'],
df_data=ac.df_data.xs('DJF', level='campaign_group'),
fit_dict=ec.get_fit_dict(ac.df_fits.loc['DJF']),
),
srf_win=dict(
title=f'JJA {flux_type} flux vs. SO – SPO CO$_2$',
xlabel='$\Delta_{ y}$' + str_co2 + ' [ppm]',
df_data=sc.df_data.xs(('1999-2020', 'JJA'), level=['period', 'season']),
fit_dict=ec.get_fit_dict(sc.df_fits.loc[('1999-2020', 'JJA')]),
),
air_win=dict(
title=f'Mar-Nov: 90-day {flux_type} flux vs. {theta_str} CO$_2$ diff',
xlabel='$\Delta_{ θ}$' + str_co2 + ' [ppm]',
highlight_campaign=['ATom-1'],
df_data=ac.df_data.xs('MAMJJASON', level='campaign_group'),
fit_dict=ec.get_fit_dict(ac.df_fits.loc['MAMJJASON']),
),
)
xlm = {}; ylm = {}; eq_text = {}
plotted_elements = []; legend_elements = [];
for n, k in enumerate(subplots):
kwargs = {key: val for key, val in kwargs_dict[k].items()}
i, j = np.unravel_index(n, (nrow, ncol))
if j == 0:
kwargs['ylabel'] = f'{flux_type} flux south of 45°S [Pg C yr$^{{-1}}$]'
kwargs['circle_one'] = ['sEXTocNEET_SOCCOM_v2020',] #'CarboScope', 'CarboScopeAdjSO', 'sEXTocNEET_SOCCOM_v2020',]
le, pe, eq_text[k] = ec.plot_constraint(axs[k], plotted_elements=plotted_elements, **kwargs)
legend_elements.extend(le); plotted_elements.extend(pe)
xlm[k] = np.array(axs[k].get_xlim())
ylm[k] = np.array(axs[k].get_ylim())
print(f'xlim {k}: {xlm[k]}')
def get_ax_lim(keys):
bnd = lambda lm: (np.min([lm[k][0] for k in keys]), np.max([lm[k][1] for k in keys]))
return bnd(xlm), bnd(ylm)
xlm_summer, ylm_summer = get_ax_lim([k for k in subplots if '_sum' in k])
xlm_winter, ylm_winter = get_ax_lim([k for k in subplots if '_win' in k])
set_ylm = dict(
srf_sum=ylm_summer,
air_sum=ylm_summer,
srf_win=ylm_winter,
air_win=ylm_winter,
)
set_xlm = {}
if constraint_type == 'ocean_constraint':
set_xlm = dict(
srf_sum=(-0.405, 0.105), #(-0.405, 0.095),
srf_win=(-0.205, 0.305), #(-0.405, 0.095),
air_sum=(-1.7, 0.36),
air_win=(-0.705, 0.365),
)
elif constraint_type == 'total_constraint':
set_xlm = dict(
srf_sum=(-0.45, 0.05), #(-0.405, 0.095),
srf_win=(-0.25, 0.25), #(-0.405, 0.095),
air_sum=(-1.48, 0.51),
air_win=(-1.04, 0.22),
)\
for k in subplots:
if k in set_xlm:
if not xlm[k][0] >= set_xlm[k][0]:
print(f'WARNING: {k}: xbound[0] truncated: def: {xlm[k][0]} set: {set_xlm[k][0]}')
if not xlm[k][1] <= set_xlm[k][1]:
print(f'WARNING: {k}: xbound[1] truncated: def: {xlm[k][1]} set: {set_xlm[k][1]}')
print(f'xlim set span {k}: {np.diff(set_xlm[k])}')
axs[k].set_xlim(set_xlm[k])
if k in set_ylm:
if not ylm[k][0] >= set_ylm[k][0]:
print(f'WARNING: {k}: ybound[0] truncated: def: {ylm[k][0]} set: {set_ylm[k][0]}')
if not ylm[k][1] <= set_ylm[k][1]:
print(f'WARNING: {k}: ybound[1] truncated: def: {ylm[k][1]} set: {set_ylm[k][1]}')
axs[k].set_ylim(set_ylm[k])
if eq_text[k] is not None:
xlim = np.array(axs[k].get_xlim())
ylim = np.array(axs[k].get_ylim())
xtxt = xlim[0] + np.diff(xlim) * 0.025
ytxt = ylim[1] - np.diff(ylim) * 0.25
eq_text[k].set_position((xtxt, ytxt))
if constraint_type in ['ocean_constraint', 'total_constraint']:
ec.add_obs_constraint(
ax=axs['srf_sum'],
df=sc.surface_flux,
indexes=[('1999-2020', 'DJF')],
marker_spec={},
labels=['1999-2020',],
label_dx={'1999-2020': 0.05},
#label_dy={'1999-2020': 0.87},
rotation={'1999-2020': 0.0},
)
ec.add_obs_constraint(
ax=axs['air_sum'],
df=ac.campaign_flux,
indexes=['ORCAS-J',],
marker_spec=figure_panels.marker_spec_campaigns(),
labels=['ORCAS-J',],
label_dx={'ORCAS-J': 0.05},
label_dy={'ORCAS-J': 0.45},
)
ec.add_obs_constraint(
ax=axs['srf_win'],
df=sc.surface_flux,
indexes=[('1999-2020', 'JJA')],
marker_spec={},
labels=['1999-2020'],
label_dx={'1999-2020': 0.01},
)
ec.add_obs_constraint(
ax=axs['air_win'],
df=ac.campaign_flux,
indexes=['ATom-1', ],
marker_spec=figure_panels.marker_spec_campaigns(),
labels=['ATom-1',],
label_dx={'ATom-1': 0.03},
)
axs[subplots[-1]].legend(handles=legend_elements, ncol=4, loc=(-1.275, -0.475), frameon=False)
# add annotations
# util.subplot_col_labels(
# [axs[subplots[j]] for j in range(ncol)],
# ['Summer', '']
# )
util.subplot_row_labels(
[axs[subplots[i*ncol]] for i in range(nrow)],
['Aircraft observations', 'Surface observations']
)
fit_campaigns = {}
fit_ax_props = {}
fit_copy_properties = ['xlim', 'ylim', 'xticks', 'yticks']
for i, fit_group in enumerate(fit_groups):
fit_campaigns[fit_group] = util.list_set(
[ndx[i_campaign] for ndx in ac.df_data.xs((fit_group), level=('campaign_group')).index]
)
ax = axs[[k for k in subplots if 'air_' in k][i]]
fit_ax_props[fit_group] = {p: plt.getp(ax, p) for p in fit_copy_properties}
util.label_plots(fig, [axs[k] for k in subplots], xoff=-0.02, yoff=0.04)
return fit_campaigns, fit_ax_props
constraint_type = 'ocean_constraint'
fit_campaigns_ocn, fit_ax_props_ocn = plot_flux_v_gradient(constraint_type)
util.savefig(f'emergent-constraint-{constraint_type}.pdf')
xlim air_sum: [-1.396637 0.354277]
xlim air_win: [-0.70588275 0.30463775]
xlim srf_sum: [-0.31971167 0.01522437]
xlim srf_win: [-0.08157546 0.05080845]
xlim set span air_sum: [2.06]
WARNING: air_win: xbound[0] truncated: def: -0.7058827499999798 set: -0.705
xlim set span air_win: [1.07]
xlim set span srf_sum: [0.51]
xlim set span srf_win: [0.51]
Emergent constraints on air-sea fluxes south of 45°S. Upper panels: 90-day-mean air-sea fluxes south of 45°S versus ∆θCO2ocn from model simulations (see SM) replicating aircraft observations collected during (A) Dec-Feb and (B) Mar-Nov. Colored vertical lines show an observed value of \(\Delta_{\theta}\ce{CO}_2\) (ORCAS during Jan in panel “A” and ATom-1 in panel “B”) corrected for land and fossil-fuel influence, with shading indicating both analytical uncertainty and model spread in the correction; colored points highlight the model samples from these particular campaigns, while gray points show data from other campaigns in the (A) Dec–Feb or (B) Mar–Nov timeframe. Lower panels: Seasonal-mean surface fluxes versus ∆yCO2ocn computed from models for (C) summer (DJF) and (D) winter (JJA) over 1999–2019. Points correspond to individual models; whiskers denote the standard deviation of interannual variability. Light blue vertical lines show the observed ∆yCO2 corrected for land and fossil fuel influence; shading shows analytical uncertainty and model spread in the correction (see SM; Fig. S12A, B show ∆yCO2 time series). The sign convention for fluxes is positive upward. Diagonal lines, where significant, show the best-fit line to all data points shown: inset text shows an estimate of the slope with standard error (SM); goodness-of-fit statistics are also shown. Note that the axis bounds differ by panel. See below for a version of this plot based on total CO2.
constraint_type = 'total_constraint'
fit_campaigns_tot, fit_ax_props_tot = plot_flux_v_gradient(constraint_type)
util.savefig(f'emergent-constraint-{constraint_type}.pdf')
xlim air_sum: [-1.469505 0.501805]
xlim air_win: [-1.02552925 0.15211425]
xlim srf_sum: [-0.31791303 0.01513872]
xlim srf_win: [-0.02910749 0.17800234]
xlim set span air_sum: [1.99]
xlim set span air_win: [1.26]
xlim set span srf_sum: [0.5]
xlim set span srf_win: [0.5]
Emergent constraints on air-sea fluxes south of 45°S based on total CO2. See caption above for full explanation; the difference here is that we use total surface flux versus total CO2** and do not correct the observed gradients (colored vertical lines) for land and fossil-fuel components. Shading in these lines here solely indicates analytical uncertainties.
Flux-gradient relation by campaign¶
def plot_flux_v_gradient_by_campaign(constraint_type, fit_campaigns, fit_ax_props):
ac = acs[constraint_type]
fit_groups = ac.fit_groups
campaign_info = obs_aircraft.get_campaign_info(verbose=False)
figsize = (4.5, 4)
for fit_group in fit_groups:
plotted_elements = []
legend_elements = []
npanel = len(fit_campaigns[fit_group])
fig, axs = util.canvas(
npanel,
figsize=(5, 4),
use_gridspec=True,
#left=0, right=1,
#bottom=0., top=1,
hspace=0.15, wspace=0.1+(npanel-4)*0.02,
)
nrow, ncol = axs.shape
xmin = ac.df_data.xs((fit_group), level=('campaign_group')).gradient.min()
xmax = ac.df_data.xs((fit_group), level=('campaign_group')).gradient.max()
if constraint_type == 'ocean_constraint':
flux_type = 'Air-sea'
else:
flux_type = 'Surface'
str_co2 = 'CO$_2^{ocn}$' if constraint_type == 'ocean_constraint' else 'CO$_2$'
ax_list = []
for n, campaign in enumerate(fit_campaigns[fit_group]):
i, j = np.unravel_index(n, axs.shape)
ax = axs[i, j]
ax_list.append(ax)
xlabel = '$\Delta_{ θ}$' + str_co2 + ' [ppm]' if i == axs.shape[0]-1 else ''
ylabel = f'{flux_type} flux south of 45°S [Pg C yr$^{{-1}}$]' if j == 0 else ''
kwargs = dict(
title_left=campaign,
title_right=obs_aircraft.campaign_datestr(campaign_info[campaign]['time_bound']),
xlabel=xlabel,
ylabel=ylabel,
xhat=[xmin, xmax],
df_data=ac.df_data.xs((fit_group, campaign), level=('campaign_group', 'campaign')),
fit_dict=ec.get_fit_dict(ac.df_fits.loc[fit_group]),
)
le, pe, _ = ec.plot_constraint(ax, plotted_elements=plotted_elements,
include_equation=False,
**kwargs)
ec.add_obs_constraint(
ax=ax,
df=ac.vg_obs,
indexes=[campaign],
marker_spec=figure_panels.marker_spec_campaigns(),
)
legend_elements.extend(le)
plotted_elements.extend(pe)
plt.setp(ax, **fit_ax_props[fit_group])
if j > 0:
ax.set_yticklabels([])
if i < nrow - 1:
ax.set_xticklabels([])
if len(fit_campaigns[fit_group]) == 4:
axs[-1, -1].legend(handles=legend_elements, ncol=4, loc=(-1.275, -0.6), frameon=False)
else:
axs[-1, 1].legend(handles=legend_elements, ncol=4, loc=(-1., -0.5), frameon=False)
util.label_plots(fig, ax_list, xoff=-0.03)
util.savefig(f'emergent-constraint-by-campaign-{fit_group}-{constraint_type}.pdf')
plot_flux_v_gradient_by_campaign('ocean_constraint', fit_campaigns_ocn, fit_ax_props_ocn)
Seasonal dependence of flux-gradient slope¶
constraint_type = 'ocean_constraint'
air_parms = dict(**ec.get_parameters('default'))
ac = obj_air[constraint_type].get_ac(**air_parms, clobber=False)
parms = dict(**ec.get_parameters('default'))
parms['fit_groups']=('DJF', 'MAM', 'JJA', 'SON')
ac_tmp = obj_air[constraint_type].get_ac(**parms, clobber=False)
computing...
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/CT2017/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/CT2019B/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/CTE2018/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/CTE2020/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/MIROC/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/CAMSv20r1/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/s99oc_v2020/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/s99oc_ADJocI40S_v2020/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/s99oc_SOCCOM_v2020/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/TM5-Flux-m0f/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/TM5-Flux-mmf/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/TM5-Flux-mrf/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
assuming cache is correct
reading cached file: /glade/u/home/mclong/codes/so-co2-airborne-obs/so-co2-airborne-obs/models/data-cache/TM5-Flux-mwf/flux_ts_monthly.lat_range=-90_0_to_-45_0.nc
done.
writing /glade/work/mclong/so-co2-airborne-obs/cache-emergent-constraint/pickles/0b6288ace20409cbf0253b3123207116/1102fba1822b55fd3220063cb83e1eab.pkl
seasons = list(ac_tmp.fit_groups)
slopes = [beta[0] for beta in ac_tmp.df_fits.loc[seasons].fit_beta]
slope_err = [sd_beta[0] for sd_beta in ac_tmp.df_fits.loc[seasons].fit_stderr_beta]
plt.bar(seasons, slopes, yerr=slope_err);
plt.ylabel(r'Flux v. $\Delta_{ θ}$CO$_2$ [Pg C yr$^{{-1}}$ ppm$^{-1}$]')
plt.title('Seasonal dependence of flux v. gradient relationship');
slope_not_summer = ac.df_fits.loc['MAMJJASON'].fit_beta[0]
slope_err_not_summer = ac.df_fits.loc['MAMJJASON'].fit_stderr_beta[0]
plt.axhspan(
slope_not_summer - slope_err_not_summer,
slope_not_summer + slope_err_not_summer,
color='k', alpha=0.2, zorder=1)
plt.axhline(ac.df_fits.loc['MAMJJASON'].fit_beta[0], linestyle='--', color='k');
for season, s, s_err in zip(seasons, slopes, slope_err):
print(f'{season}: {s:0.2f} ± {s_err:0.2f}')
print(f'MAMJJASON: {slope_not_summer:0.2f} ± {slope_err_not_summer:0.2f}')
DJF: 0.96 ± 0.10
MAM: 1.98 ± 0.41
JJA: 3.20 ± 0.31
SON: 2.41 ± 0.23
MAMJJASON: 2.48 ± 0.14
print(f'Seaonsal fit groups ann mean flux: {ac_tmp.estimate_ann_mean_flux[0]:0.2f} ± {ac_tmp.estimate_ann_mean_flux[1]:0.2f}')
print(f'Summer/not-summer fit groups ann mean flux: {ac.estimate_ann_mean_flux[0]:0.2f} ± {ac.estimate_ann_mean_flux[1]:0.2f}')
Seaonsal fit groups ann mean flux: -0.53 ± 0.35
Summer/not-summer fit groups ann mean flux: -0.53 ± 0.23