def test_TRENDY_window():
TRENDYdf = AirborneFraction.TRENDY(co2_year, temp_year, trendy_uptake_mock)
TRENDY_co2 = TRENDYdf.co2[2:].values
TRENDY_temp = TRENDYdf.temp.sel(time=slice('1959', '2018')).Earth.values
TRENDYdf.GCP['ocean sink'] = (TRENDY_co2 + TRENDY_temp) / 2
class_af, alpha = TRENDYdf.window_af()
phi = 0.015 / 2.12
rho = 1.93
test_alpha = 1 / 2.12 + 1 * phi / rho
for decade in alpha.index:
assert alpha.mean(axis=1).loc[decade] == pytest.approx(
test_alpha, 1e-4)
assert alpha.std(axis=1).loc[decade] == pytest.approx(0.)
emission_rate = 0.02
b = 1 / np.log(1 + emission_rate)
u = 1 - b * test_alpha
test_af = 1 / u
for decade in class_af['mean'].index:
assert class_af['mean'].loc[decade] == pytest.approx(test_af, 1e-4)
assert class_af['std'].loc[decade] == pytest.approx(0.)
def test_trendy_airborne():
df = AirborneFraction.TRENDY(co2_year, temp_year, trendy_uptake_mock)
GCPdf = AirborneFraction.GCP(co2_year, temp_year)
GCPdf.GCP['land sink'] = (GCPdf.co2 + GCPdf.temp) / 2
GCPdf.GCP['ocean sink'] = (GCPdf.co2 + GCPdf.temp) / 2
phi = 0.015 / 2.12
rho = 1.93
land = df._feedback_parameters('Earth_Land')
ocean = GCPdf._feedback_parameters()['ocean']
ocean.index = ['beta', 'gamma']
ocean['beta'] /= 2.12
ocean['u_gamma'] = ocean['gamma'] * phi / rho
for parameter, divisor in zip(['beta', 'gamma'], [2.12, 1]):
assert land.loc[parameter].mean() == pytest.approx(0.5 / divisor)
assert land.loc[parameter].std() == pytest.approx(0.)
assert ocean[parameter] == pytest.approx(0.5 / divisor)
# Test function
emission_rate = 2
b = 1 / np.log(1 + emission_rate / 100)
beta = 0.5 * 2 / 2.12
u_gamma = 0.5 * 2 * phi / rho
u = 1 - b * (beta + u_gamma)
test_af = 1 / u
# Class function
params = land.add(ocean, axis=0)
class_beta = params.loc['beta']
class_u_gamma = params.loc['u_gamma']
class_b = 1 / np.log(1 + emission_rate / 100)
class_u = 1 - class_b * (class_beta + class_u_gamma)
class_af = 1 / class_u
assert class_af.mean() == pytest.approx(test_af)
def test_GCP_airborne():
df = AirborneFraction.GCP(co2_year, temp_year)
df.GCP['land sink'] = (df.co2 + df.temp) / 2
df.GCP['ocean sink'] = (df.co2 + df.temp) / 2
params = df._feedback_parameters()
for sink, param in product(['land', 'ocean'], ['CO2', 'temp']):
assert params[sink][param] == pytest.approx(0.5)
phi = 0.015 / 2.12
rho = 1.93
emission_rate = 2
b = 1 / np.log(1 + emission_rate / 100)
beta = 0.5 * 2 / 2.12
u_gamma = 0.5 * 2 * phi / rho
u = 1 - b * (beta + u_gamma)
test_af = 1 / u
assert df.airborne_fraction(emission_rate) == pytest.approx(test_af)
def test_GCP_window():
GCPdf = AirborneFraction.GCP(co2_year, temp_year)
GCPdf.GCP['land sink'] = (GCPdf.co2 + GCPdf.temp) / 2
GCPdf.GCP['ocean sink'] = (GCPdf.co2 + GCPdf.temp) / 2
class_af, alpha = GCPdf.window_af()
phi = 0.015 / 2.12
rho = 1.93
test_alpha = 1 / 2.12 + 1 * phi / rho
assert alpha.mean() == pytest.approx(test_alpha)
assert alpha.std() == pytest.approx(0)
emission_rate = 0.02
b = 1 / np.log(1 + emission_rate)
u = 1 - b * test_alpha
test_af = 1 / u
for decade in class_af.index:
assert class_af.loc[decade] == pytest.approx(test_af)
def test_INVF_window():
INVdf = AirborneFraction.INVF(co2_year, temp_year, invf_uptake_mock)
class_af, alpha = INVdf.window_af()
phi = 0.015 / 2.12
rho = 1.93
test_alpha = 1 / 2.12 + 1 * phi / rho
for decade in alpha.index:
assert alpha.mean(axis=1).loc[decade] == pytest.approx(
test_alpha, 1e-4)
assert alpha.std(axis=1).loc[decade] == pytest.approx(0.)
emission_rate = 0.02
b = 1 / np.log(1 + emission_rate)
u = 1 - b * test_alpha
test_af = 1 / u
for decade in class_af['mean'].index:
assert class_af['mean'].loc[decade] == pytest.approx(test_af, 1e-4)
assert class_af['std'].loc[decade] == pytest.approx(0.)
def test_invf_airborne():
df = AirborneFraction.INVF(co2_year, temp_year, invf_uptake_mock)
land = df._feedback_parameters('Earth_Land')
ocean = df._feedback_parameters('Earth_Ocean')
for region, param in product([land, ocean], ['beta', 'gamma']):
assert region.loc[param].mean() == pytest.approx(0.5)
assert region.loc[param].std() == pytest.approx(0.)
phi = 0.015 / 2.12
rho = 1.93
emission_rate = 2
b = 1 / np.log(1 + emission_rate / 100)
beta = 0.5 * 2 / 2.12
u_gamma = 0.5 * 2 * phi / rho
u = 1 - b * (beta + u_gamma)
test_af = 1 / u
assert df.airborne_fraction(emission_rate)['mean'] == pytest.approx(
test_af)
assert df.airborne_fraction(emission_rate)['std'] == pytest.approx(0.)
return af.mean()
""" RESULTS """
def af_dataframe(data):
af_df = pd.Series(data.airborne_fraction())
af_df['alpha_5'] = af_df['alpha_mean'] - af_df['alpha_std'] * 1.645
af_df['alpha_95'] = af_df['alpha_mean'] + af_df['alpha_std'] * 1.645
return af_df
""" EXECUTIONS """
GCPdf = AirborneFraction.GCP(co2['year'], temp['year'])
GCPdf.airborne_fraction()
pd.Series(GCPdf.airborne_fraction())
GCP_window = pd.DataFrame(GCPdf.window_af())
GCP_window
stats.linregress(range(len(GCP_window.index)), GCP_window.af)
INVdf = AirborneFraction.INVF(co2['year'], temp['year'], invf_uptake['year'])
af_dataframe(INVdf)
inv_window = pd.DataFrame(INVdf.window_af())
inv_window['alpha_cf'] = inv_window['alpha_std'] * 1.645
inv_window['alpha_5'] = inv_window['alpha_mean'] - inv_window['alpha_cf']
inv_window['alpha_95'] = inv_window['alpha_mean'] + inv_window['alpha_cf']
inv_window
"month": {
model_name: xr.open_dataset(
OUTPUT_DIR +
f'TRENDY/spatial/output_all/{model_name}_S3_nbp/month.nc')
for model_name in trendy_models
}
}
}
phi, rho = 0.0071, 1.93
""" DEVS """
from importlib import reload
reload(AirborneFraction)
df = AirborneFraction.GCP(co2['year'], temp['year']) #, invf_uptake['year'])
df.window_af()
def plot():
fig = plt.figure(figsize=(10, 8))
alpha = beta + u_gamma
ax1 = fig.add_subplot(211)
ax1.bar(alpha.mean(axis=1).index - 3, alpha.mean(axis=1).values, width=3)
ax1.axhline(linestyle='--', alpha=0.5, color='k')
ax1.axhline(emission_rate, linestyle='--', alpha=0.5, color='r')
ax2 = fig.add_subplot(212)
x = af_results['mean'].index
y = af_results['mean'].values