How to run the AR6 workflow¶

Here we demonstrate how to run the workflow that was used in AR6. This is intended to demonstrate how to use the package in a familiar context. If you want a simpler interface for doing this, please see the climate-assessment package, which is a facade around gcages.

Imports¶

In [1]:

import multiprocessing
import os
import platform
from functools import partial
from pathlib import Path

import numpy as np
import openscm_units
import pandas as pd
import pandas_indexing as pix
import pandas_openscm
import pint
import seaborn as sns

from gcages.ar6 import (
    AR6Harmoniser,
    AR6Infiller,
    AR6PostProcessor,
    AR6PreProcessor,
    AR6SCMRunner,
    get_ar6_full_historical_emissions,
)

import multiprocessing import os import platform from functools import partial from pathlib import Path import numpy as np import openscm_units import pandas as pd import pandas_indexing as pix import pandas_openscm import pint import seaborn as sns from gcages.ar6 import ( AR6Harmoniser, AR6Infiller, AR6PostProcessor, AR6PreProcessor, AR6SCMRunner, get_ar6_full_historical_emissions, )

In [2]:

# Setup pint
pint.set_application_registry(openscm_units.unit_registry)

# Setup pint pint.set_application_registry(openscm_units.unit_registry)

In [3]:

pandas_openscm.register_pandas_accessors()

pandas_openscm.register_pandas_accessors()

Starting point¶

The starting point is some emissions scenario. This must be in a format like the below, i.e. a pandas DataFrame with a MultiIndex with levels: ["model", "scenario", "region", "variable", "unit"] and years as the columns.

Naming conventions¶

A complete discussion on naming conventions is provided in or our docs on naming conventions. In short, we use the gcages naming conventions throughout. However, AR6 used the IAMC naming convention as its starting point. Hence, we start from the IAMC naming convention here. However, the pre-processing step alters the naming convention to gcages names.

In [4]:

# All the code to generate these demo timeseries
# (this is obviously a scrappy demo,
# you would normally use output from somewhere else
# and be much more careful than this :)).
time = np.arange(2015, 2100 + 1)

co2_flatline = np.ones_like(time) * 38.5
co2_flatline[np.logical_and(time > 2040, time < 2075)] = 38.5 * (
    1 - 1 / (1 + np.exp(-(np.arange(2075 - 2041) - 15) / 3.0))
)
co2_flatline[time >= 2075] = 0.0
co2_flatline *= 1000.0

ch4_flatline = np.ones_like(time) * 410
ch4_flatline[np.logical_and(time > 2040, time < 2075)] = (
    (410.0 - 200.0) * (1 - 1 / (1 + np.exp(-(np.arange(2075 - 2041) - 15) / 3.0)))
) + 200.0
ch4_flatline[time >= 2075] = 200.0

co2_decline = np.ones_like(time) * 38.5
co2_decline[np.logical_and(time > 2030, time < 2050)] = 38.5 * (
    1 - 1.2 / (1 + np.exp(-(np.arange(2050 - 2031) - 10) / 3.0))
)
co2_decline[time >= 2050] = np.nan
co2_decline[time == 2100] = 0.0
co2_decline *= 1000.0

ch4_decline = np.ones_like(time) * 410
ch4_decline[np.logical_and(time > 2030, time < 2060)] = (
    (410.0 - 150.0) * (1 - 1 / (1 + np.exp(-(np.arange(2060 - 2031) - 10) / 3.0)))
) + 150.0
ch4_decline[time >= 2050] = 150.0

# All the code to generate these demo timeseries # (this is obviously a scrappy demo, # you would normally use output from somewhere else # and be much more careful than this :)). time = np.arange(2015, 2100 + 1) co2_flatline = np.ones_like(time) * 38.5 co2_flatline[np.logical_and(time > 2040, time < 2075)] = 38.5 * ( 1 - 1 / (1 + np.exp(-(np.arange(2075 - 2041) - 15) / 3.0)) ) co2_flatline[time >= 2075] = 0.0 co2_flatline *= 1000.0 ch4_flatline = np.ones_like(time) * 410 ch4_flatline[np.logical_and(time > 2040, time < 2075)] = ( (410.0 - 200.0) * (1 - 1 / (1 + np.exp(-(np.arange(2075 - 2041) - 15) / 3.0))) ) + 200.0 ch4_flatline[time >= 2075] = 200.0 co2_decline = np.ones_like(time) * 38.5 co2_decline[np.logical_and(time > 2030, time < 2050)] = 38.5 * ( 1 - 1.2 / (1 + np.exp(-(np.arange(2050 - 2031) - 10) / 3.0)) ) co2_decline[time >= 2050] = np.nan co2_decline[time == 2100] = 0.0 co2_decline *= 1000.0 ch4_decline = np.ones_like(time) * 410 ch4_decline[np.logical_and(time > 2030, time < 2060)] = ( (410.0 - 150.0) * (1 - 1 / (1 + np.exp(-(np.arange(2060 - 2031) - 10) / 3.0))) ) + 150.0 ch4_decline[time >= 2050] = 150.0

In [5]:

# Put it altogether into a DataFrame
start = pd.DataFrame(
    np.vstack(
        [
            co2_flatline,
            ch4_flatline,
            co2_flatline,
            co2_decline,
            ch4_decline,
            co2_decline * 0.85,
            co2_decline * 0.13
            + np.arange(co2_decline.size) * 10.0
            + 500.0 * np.ones_like(co2_decline),
        ]
    ),
    columns=time,
    index=pd.MultiIndex.from_tuples(
        # Note the use of IAMC names here
        [
            (
                "demo",
                "flatline",
                "World",
                "Emissions|CO2|Energy and Industrial Processes",
                "Mt CO2/yr",
            ),
            ("demo", "flatline", "World", "Emissions|CH4", "Mt CH4/yr"),
            (
                "demo",
                "flatline-co2-only",
                "World",
                "Emissions|CO2|Energy and Industrial Processes",
                "Mt CO2/yr",
            ),
            (
                "demo",
                "decline",
                "World",
                "Emissions|CO2|Energy and Industrial Processes",
                "Mt CO2/yr",
            ),
            ("demo", "decline", "World", "Emissions|CH4", "Mt CH4/yr"),
            (
                "demo",
                "decline_co2_fossil_split",
                "World",
                "Emissions|CO2|Energy",
                "Mt CO2/yr",
            ),
            (
                "demo",
                "decline_co2_fossil_split",
                "World",
                "Emissions|CO2|Industrial Processes",
                "Mt CO2/yr",
            ),
        ],
        names=["model", "scenario", "region", "variable", "unit"],
    ),
)
start = start.T.interpolate("index").T
start

# Put it altogether into a DataFrame start = pd.DataFrame( np.vstack( [ co2_flatline, ch4_flatline, co2_flatline, co2_decline, ch4_decline, co2_decline * 0.85, co2_decline * 0.13 + np.arange(co2_decline.size) * 10.0 + 500.0 * np.ones_like(co2_decline), ] ), columns=time, index=pd.MultiIndex.from_tuples( # Note the use of IAMC names here [ ( "demo", "flatline", "World", "Emissions|CO2|Energy and Industrial Processes", "Mt CO2/yr", ), ("demo", "flatline", "World", "Emissions|CH4", "Mt CH4/yr"), ( "demo", "flatline-co2-only", "World", "Emissions|CO2|Energy and Industrial Processes", "Mt CO2/yr", ), ( "demo", "decline", "World", "Emissions|CO2|Energy and Industrial Processes", "Mt CO2/yr", ), ("demo", "decline", "World", "Emissions|CH4", "Mt CH4/yr"), ( "demo", "decline_co2_fossil_split", "World", "Emissions|CO2|Energy", "Mt CO2/yr", ), ( "demo", "decline_co2_fossil_split", "World", "Emissions|CO2|Industrial Processes", "Mt CO2/yr", ), ], names=["model", "scenario", "region", "variable", "unit"], ), ) start = start.T.interpolate("index").T start

Out[5]:

					2015	2016	2017	2018	2019	2020	2021	2022	2023	2024	...	2091	2092	2093	2094	2095	2096	2097	2098	2099	2100
model	scenario	region	variable	unit
demo	flatline	World	Emissions|CO2|Energy and Industrial Processes	Mt CO2/yr	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	...	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.0
			Emissions|CH4	Mt CH4/yr	410.0	410.0	410.0	410.0	410.0	410.0	410.0	410.0	410.0	410.0	...	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.0
	flatline-co2-only	World	Emissions|CO2|Energy and Industrial Processes	Mt CO2/yr	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	...	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.0
	decline	World	Emissions|CO2|Energy and Industrial Processes	Mt CO2/yr	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	...	-829.133715	-737.007747	-644.881779	-552.755810	-460.629842	-368.503873	-276.377905	-184.251937	-92.125968	0.0
			Emissions|CH4	Mt CH4/yr	410.0	410.0	410.0	410.0	410.0	410.0	410.0	410.0	410.0	410.0	...	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.0
	decline_co2_fossil_split	World	Emissions|CO2|Energy	Mt CO2/yr	32725.0	32725.0	32725.0	32725.0	32725.0	32725.0	32725.0	32725.0	32725.0	32725.0	...	-704.763658	-626.456585	-548.149512	-469.842439	-391.535366	-313.228292	-234.921219	-156.614146	-78.307073	0.0
			Emissions|CO2|Industrial Processes	Mt CO2/yr	5505.0	5515.0	5525.0	5535.0	5545.0	5555.0	5565.0	5575.0	5585.0	5595.0	...	1152.212617	1174.188993	1196.165369	1218.141745	1240.118121	1262.094496	1284.070872	1306.047248	1328.023624	1350.0

7 rows × 86 columns

In [6]:

relplot_in_emms = partial(
    sns.relplot,
    kind="line",
    linewidth=2.0,
    alpha=0.7,
    facet_kws=dict(sharey=False),
    x="year",
    y="value",
    col="variable",
    col_wrap=2,
)
fg = relplot_in_emms(
    data=start.melt(ignore_index=False, var_name="year").reset_index(),
    hue="scenario",
)

fg.axes.flatten()[0].axhline(0.0, linestyle="--", color="gray")
fg.axes.flatten()[1].set_ylim(ymin=0.0)

relplot_in_emms = partial( sns.relplot, kind="line", linewidth=2.0, alpha=0.7, facet_kws=dict(sharey=False), x="year", y="value", col="variable", col_wrap=2, ) fg = relplot_in_emms( data=start.melt(ignore_index=False, var_name="year").reset_index(), hue="scenario", ) fg.axes.flatten()[0].axhline(0.0, linestyle="--", color="gray") fg.axes.flatten()[1].set_ylim(ymin=0.0)

Out[6]:

(0.0, 423.0)

Pre-process¶

If you want to run in exactly the same way as AR6, the first step is pre-processing the emissions scenario(s) (if you just want the same logic as AR6 for other steps, it is possible to skip this step and just go straight to harmonisation). This step does some variable aggregation that was needed in AR6 and extracts only those variables that are understood by the workflow.

In [7]:

pre_processor = AR6PreProcessor.from_ar6_config(
    n_processes=None,  # run serially for this demo
)

pre_processor = AR6PreProcessor.from_ar6_config( n_processes=None, # run serially for this demo )

In [8]:

# These are the variables (using the IAMC naming convention)
# understood by the workflow as was used in AR6.
pre_processor.emissions_out

# These are the variables (using the IAMC naming convention) # understood by the workflow as was used in AR6. pre_processor.emissions_out

Out[8]:

('Emissions|BC',
 'Emissions|PFC|C2F6',
 'Emissions|PFC|C6F14',
 'Emissions|PFC|CF4',
 'Emissions|CO',
 'Emissions|CO2',
 'Emissions|CO2|AFOLU',
 'Emissions|CO2|Energy and Industrial Processes',
 'Emissions|CH4',
 'Emissions|HFC|HFC125',
 'Emissions|HFC|HFC134a',
 'Emissions|HFC|HFC143a',
 'Emissions|HFC|HFC227ea',
 'Emissions|HFC|HFC23',
 'Emissions|HFC|HFC32',
 'Emissions|HFC|HFC43-10',
 'Emissions|N2O',
 'Emissions|NH3',
 'Emissions|NOx',
 'Emissions|OC',
 'Emissions|SF6',
 'Emissions|Sulfur',
 'Emissions|VOC')

Here we run the pre-processing and get the pre-processed data. Note a few things which were done:

1. renaming of variables
2. aggregation of the scenario which provided energy and industrial emissions separately

In [9]:

pre_processed = pre_processor(start)
pre_processed

pre_processed = pre_processor(start) pre_processed

/home/docs/checkouts/readthedocs.org/user_builds/gcages/envs/latest/lib/python3.11/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html
  from .autonotebook import tqdm as notebook_tqdm

For each model-scenario, calculating conditional sums: 0it [00:00, ?it/s]

For each model-scenario, calculating conditional sums: 4it [00:00, 600.02it/s]

For each model-scenario, reclassifying variables: 0it [00:00, ?it/s]

For each model-scenario, reclassifying variables: 4it [00:00, 1057.37it/s]

For each model-scenario, conditionally removing variables: 0it [00:00, ?it/s]

For each model-scenario, conditionally removing variables: 4it [00:00, 1356.39it/s]

For each model-scenario, dropping variables if they are identical: 0it [00:00, ?it/s]

For each model-scenario, dropping variables if they are identical: 4it [00:00, 1386.66it/s]

Out[9]:

					2015	2016	2017	2018	2019	2020	2021	2022	2023	2024	...	2091	2092	2093	2094	2095	2096	2097	2098	2099	2100
model	scenario	region	variable	unit
demo	decline	World	Emissions|CO2|Fossil	Mt CO2/yr	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	...	-829.133715	-737.007747	-644.881779	-552.755810	-460.629842	-368.503873	-276.377905	-184.251937	-92.125968	0.0
			Emissions|CH4	Mt CH4/yr	410.0	410.0	410.0	410.0	410.0	410.0	410.0	410.0	410.0	410.0	...	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.0
	decline_co2_fossil_split	World	Emissions|CO2|Fossil	Mt CO2/yr	38230.0	38240.0	38250.0	38260.0	38270.0	38280.0	38290.0	38300.0	38310.0	38320.0	...	447.448959	547.732408	648.015857	748.299306	848.582755	948.866204	1049.149653	1149.433102	1249.716551	1350.0
	flatline	World	Emissions|CO2|Fossil	Mt CO2/yr	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	...	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.0
			Emissions|CH4	Mt CH4/yr	410.0	410.0	410.0	410.0	410.0	410.0	410.0	410.0	410.0	410.0	...	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.0
	flatline-co2-only	World	Emissions|CO2|Fossil	Mt CO2/yr	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	38500.0	...	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.0

6 rows × 86 columns

In [10]:

fg = relplot_in_emms(
    data=pre_processed.melt(ignore_index=False, var_name="year").reset_index(),
    hue="scenario",
)

fg.axes.flatten()[0].axhline(0.0, linestyle="--", color="gray")
fg.axes.flatten()[1].set_ylim(ymin=0.0)

fg = relplot_in_emms( data=pre_processed.melt(ignore_index=False, var_name="year").reset_index(), hue="scenario", ) fg.axes.flatten()[0].axhline(0.0, linestyle="--", color="gray") fg.axes.flatten()[1].set_ylim(ymin=0.0)

Out[10]:

(0.0, 423.0)

Harmonisation¶

The next step is harmonisation. This is the process of aligning the scenarios with historical emissions estimates.

In AR6, a specific set of historical emissions was used. There is no official home for this, but a copy is stored in the path below (and, as above, if you want something that pre-packages everything, see the climate-assessment package).

Under the hood, the AR6 harmonisation uses the aneris package.

In [11]:

AR6_HISTORICAL_EMISSIONS_FILE = Path(
    "tests/regression/ar6/ar6-workflow-inputs/history_ar6.csv"
)

AR6_HISTORICAL_EMISSIONS_FILE = Path( "tests/regression/ar6/ar6-workflow-inputs/history_ar6.csv" )

With this file, we can initialise a harmoniser exactly like that used in AR6.

In [13]:

harmoniser = AR6Harmoniser.from_ar6_config(
    ar6_historical_emissions_file=AR6_HISTORICAL_EMISSIONS_FILE,
    n_processes=None,  # run serially for this demo
)

harmoniser = AR6Harmoniser.from_ar6_config( ar6_historical_emissions_file=AR6_HISTORICAL_EMISSIONS_FILE, n_processes=None, # run serially for this demo )

And harmonise

In [14]:

harmonised = harmoniser(pre_processed)
harmonised

harmonised = harmoniser(pre_processed) harmonised

0it [00:00, ?it/s]

INFO:root:Harmonizing with reduce_ratio_2080

1it [00:00,  2.45it/s]

INFO:root:Harmonizing with reduce_ratio_2080

INFO:root:Harmonizing with reduce_ratio_2080

3it [00:00,  7.00it/s]

INFO:root:Harmonizing with reduce_ratio_2080

4it [00:00,  7.30it/s]

Out[14]:

					2015	2016	2017	2018	2019	2020	2021	2022	2023	2024	...	2091	2092	2093	2094	2095	2096	2097	2098	2099	2100
model	scenario	region	variable	unit
demo	decline	World	Emissions|CO2|Fossil	Mt CO2/yr	35635.286300	35679.358818	35723.431337	35767.503855	35811.576374	35855.648892	35899.721411	35943.793929	35987.866448	36031.938966	...	-829.133715	-737.007747	-644.881779	-552.755810	-460.629842	-368.503873	-276.377905	-184.251937	-92.125968	0.0
			Emissions|CH4	Mt CH4/yr	388.072796	388.410137	388.747479	389.084821	389.422162	389.759504	390.096845	390.434187	390.771529	391.108870	...	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.0
	decline_co2_fossil_split	World	Emissions|CO2|Fossil	Mt CO2/yr	35635.286300	35684.536703	35733.807989	35783.100158	35832.413211	35881.747147	35931.101967	35980.477670	36029.874257	36079.291727	...	447.448959	547.732408	648.015857	748.299306	848.582755	948.866204	1049.149653	1149.433102	1249.716551	1350.0
	flatline	World	Emissions|CO2|Fossil	Mt CO2/yr	35635.286300	35679.358818	35723.431337	35767.503855	35811.576374	35855.648892	35899.721411	35943.793929	35987.866448	36031.938966	...	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.0
			Emissions|CH4	Mt CH4/yr	388.072796	388.410137	388.747479	389.084821	389.422162	389.759504	390.096845	390.434187	390.771529	391.108870	...	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.0
	flatline-co2-only	World	Emissions|CO2|Fossil	Mt CO2/yr	35635.286300	35679.358818	35723.431337	35767.503855	35811.576374	35855.648892	35899.721411	35943.793929	35987.866448	36031.938966	...	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.0

6 rows × 86 columns

You can see the modification to the pathways as a result of the harmonisation in the plot below. In scenarios that have more emissions, the same idea is applied to all variables.

In [15]:

pdf = (
    pix.concat(
        [
            pre_processed.pix.assign(stage="pre_processed"),
            harmonised.pix.assign(stage="harmonised"),
            harmoniser.historical_emissions.pix.assign(
                stage="history", scenario="history", model="history"
            ).loc[pix.isin(variable=pre_processed.pix.unique("variable"))],
        ]
    )
    .melt(ignore_index=False, var_name="year")
    .reset_index()
)

fg = relplot_in_emms(
    data=pdf,
    hue="scenario",
    style="stage",
    dashes={
        "history": (1, 1),
        "pre_processed": (3, 3),
        "harmonised": "",
    },
)

fg.axes.flatten()[0].axhline(0.0, linestyle="--", color="gray")
fg.axes.flatten()[1].set_ylim(ymin=0.0)

pdf = ( pix.concat( [ pre_processed.pix.assign(stage="pre_processed"), harmonised.pix.assign(stage="harmonised"), harmoniser.historical_emissions.pix.assign( stage="history", scenario="history", model="history" ).loc[pix.isin(variable=pre_processed.pix.unique("variable"))], ] ) .melt(ignore_index=False, var_name="year") .reset_index() ) fg = relplot_in_emms( data=pdf, hue="scenario", style="stage", dashes={ "history": (1, 1), "pre_processed": (3, 3), "harmonised": "", }, ) fg.axes.flatten()[0].axhline(0.0, linestyle="--", color="gray") fg.axes.flatten()[1].set_ylim(ymin=0.0)

Out[15]:

(0.0, 423.1851264903377)

Infilling¶

The next step is infilling. This is the process of inferring any emissions which are not included in the scenarios, but are needed for running climate models. In our example, this is things like N2O, black carbon, sulfates.

In AR6, a specific infilling database was used. There is no official home for all of this, but a copy is stored in the path below (and, as above, if you want something that pre-packages everything, see the climate-assessment package).

Under the hood, the AR6 infilling uses the silicone package.

In [16]:

AR6_INFILLING_DB_FILE = Path(
    "tests/regression/ar6/ar6-workflow-inputs/infilling_db_ar6.csv"
)
AR6_INFILLING_DB_CFCS_FILE = Path(
    "tests/regression/ar6/ar6-workflow-inputs/infilling_db_ar6_cfcs.csv"
)

AR6_INFILLING_DB_FILE = Path( "tests/regression/ar6/ar6-workflow-inputs/infilling_db_ar6.csv" ) AR6_INFILLING_DB_CFCS_FILE = Path( "tests/regression/ar6/ar6-workflow-inputs/infilling_db_ar6_cfcs.csv" )

With the infilling databases, we can initialise our infiller.

In [18]:

infiller = AR6Infiller.from_ar6_config(
    ar6_infilling_db_file=AR6_INFILLING_DB_FILE,
    ar6_infilling_db_cfcs_file=AR6_INFILLING_DB_CFCS_FILE,
    n_processes=None,  # run serially for this demo
    # To make sure that our outputs remain harmonised
    # (also, turns out that the historical emissions
    # are the same as the CFCs database)
    historical_emissions=get_ar6_full_historical_emissions(AR6_INFILLING_DB_CFCS_FILE),
    harmonisation_year=harmoniser.harmonisation_year,
)

infiller = AR6Infiller.from_ar6_config( ar6_infilling_db_file=AR6_INFILLING_DB_FILE, ar6_infilling_db_cfcs_file=AR6_INFILLING_DB_CFCS_FILE, n_processes=None, # run serially for this demo # To make sure that our outputs remain harmonised # (also, turns out that the historical emissions # are the same as the CFCs database) historical_emissions=get_ar6_full_historical_emissions(AR6_INFILLING_DB_CFCS_FILE), harmonisation_year=harmoniser.harmonisation_year, )

[WARNING] 09:39:42 - pint.util: Redefining 'kt' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:42 - pint.util: Redefining 'EUR_2005' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:42 - pint.util: Redefining 'EUR' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:42 - pint.util: Redefining 'C' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:42 - pint.util: Redefining 'N' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:42 - pint.util: Redefining 'NOX' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:42 - pint.util: Redefining 'gNOX' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:42 - pint.util: Redefining 'tNOX' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:42 - pint.util: Redefining 'S' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:43 - pint.util: Redefining 'yr' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:43 - pint.util: Redefining 'a' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:43 - pint.util: Redefining 'h' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:43 - pint.util: Redefining 'd' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:43 - pint.util: Redefining 'degreeC' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:43 - pint.util: Redefining 'degreeF' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:43 - pint.util: Redefining 'kt' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:43 - pint.util: Redefining 'Tt' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

[WARNING] 09:39:43 - pint.util: Redefining 'ppm' (<class 'pint.delegates.txt_defparser.plain.UnitDefinition'>)

And infill

In [19]:

harmonised

harmonised

Out[19]:

					2015	2016	2017	2018	2019	2020	2021	2022	2023	2024	...	2091	2092	2093	2094	2095	2096	2097	2098	2099	2100
model	scenario	region	variable	unit
demo	decline	World	Emissions|CO2|Fossil	Mt CO2/yr	35635.286300	35679.358818	35723.431337	35767.503855	35811.576374	35855.648892	35899.721411	35943.793929	35987.866448	36031.938966	...	-829.133715	-737.007747	-644.881779	-552.755810	-460.629842	-368.503873	-276.377905	-184.251937	-92.125968	0.0
			Emissions|CH4	Mt CH4/yr	388.072796	388.410137	388.747479	389.084821	389.422162	389.759504	390.096845	390.434187	390.771529	391.108870	...	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.0
	decline_co2_fossil_split	World	Emissions|CO2|Fossil	Mt CO2/yr	35635.286300	35684.536703	35733.807989	35783.100158	35832.413211	35881.747147	35931.101967	35980.477670	36029.874257	36079.291727	...	447.448959	547.732408	648.015857	748.299306	848.582755	948.866204	1049.149653	1149.433102	1249.716551	1350.0
	flatline	World	Emissions|CO2|Fossil	Mt CO2/yr	35635.286300	35679.358818	35723.431337	35767.503855	35811.576374	35855.648892	35899.721411	35943.793929	35987.866448	36031.938966	...	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.0
			Emissions|CH4	Mt CH4/yr	388.072796	388.410137	388.747479	389.084821	389.422162	389.759504	390.096845	390.434187	390.771529	391.108870	...	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.0
	flatline-co2-only	World	Emissions|CO2|Fossil	Mt CO2/yr	35635.286300	35679.358818	35723.431337	35767.503855	35811.576374	35855.648892	35899.721411	35943.793929	35987.866448	36031.938966	...	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.0

6 rows × 86 columns

In [20]:

infilled = infiller(harmonised)
infilled

infilled = infiller(harmonised) infilled

Scenarios to infill: 0it [00:00, ?it/s]

/home/docs/checkouts/readthedocs.org/user_builds/gcages/envs/latest/lib/python3.11/site-packages/silicone/database_crunchers/quantile_rolling_windows.py:182: FutureWarning: DataFrame.applymap has been deprecated. Use DataFrame.map instead.
  wide_db = wide_db.applymap(lambda x: np.nan if isinstance(x, str) else x)

/home/docs/checkouts/readthedocs.org/user_builds/gcages/envs/latest/lib/python3.11/site-packages/silicone/database_crunchers/quantile_rolling_windows.py:182: FutureWarning: DataFrame.applymap has been deprecated. Use DataFrame.map instead.
  wide_db = wide_db.applymap(lambda x: np.nan if isinstance(x, str) else x)

/home/docs/checkouts/readthedocs.org/user_builds/gcages/envs/latest/lib/python3.11/site-packages/silicone/database_crunchers/quantile_rolling_windows.py:182: FutureWarning: DataFrame.applymap has been deprecated. Use DataFrame.map instead.
  wide_db = wide_db.applymap(lambda x: np.nan if isinstance(x, str) else x)

/home/docs/checkouts/readthedocs.org/user_builds/gcages/envs/latest/lib/python3.11/site-packages/silicone/database_crunchers/quantile_rolling_windows.py:182: FutureWarning: DataFrame.applymap has been deprecated. Use DataFrame.map instead.
  wide_db = wide_db.applymap(lambda x: np.nan if isinstance(x, str) else x)

/home/docs/checkouts/readthedocs.org/user_builds/gcages/envs/latest/lib/python3.11/site-packages/silicone/database_crunchers/quantile_rolling_windows.py:182: FutureWarning: DataFrame.applymap has been deprecated. Use DataFrame.map instead.
  wide_db = wide_db.applymap(lambda x: np.nan if isinstance(x, str) else x)

/home/docs/checkouts/readthedocs.org/user_builds/gcages/envs/latest/lib/python3.11/site-packages/silicone/database_crunchers/quantile_rolling_windows.py:182: FutureWarning: DataFrame.applymap has been deprecated. Use DataFrame.map instead.
  wide_db = wide_db.applymap(lambda x: np.nan if isinstance(x, str) else x)

/home/docs/checkouts/readthedocs.org/user_builds/gcages/envs/latest/lib/python3.11/site-packages/silicone/database_crunchers/quantile_rolling_windows.py:182: FutureWarning: DataFrame.applymap has been deprecated. Use DataFrame.map instead.
  wide_db = wide_db.applymap(lambda x: np.nan if isinstance(x, str) else x)

/home/docs/checkouts/readthedocs.org/user_builds/gcages/envs/latest/lib/python3.11/site-packages/silicone/database_crunchers/quantile_rolling_windows.py:182: FutureWarning: DataFrame.applymap has been deprecated. Use DataFrame.map instead.
  wide_db = wide_db.applymap(lambda x: np.nan if isinstance(x, str) else x)

/home/docs/checkouts/readthedocs.org/user_builds/gcages/envs/latest/lib/python3.11/site-packages/silicone/database_crunchers/quantile_rolling_windows.py:182: FutureWarning: DataFrame.applymap has been deprecated. Use DataFrame.map instead.
  wide_db = wide_db.applymap(lambda x: np.nan if isinstance(x, str) else x)

Scenarios to infill: 1it [00:38, 38.95s/it]

/home/docs/checkouts/readthedocs.org/user_builds/gcages/envs/latest/lib/python3.11/site-packages/silicone/database_crunchers/quantile_rolling_windows.py:182: FutureWarning: DataFrame.applymap has been deprecated. Use DataFrame.map instead.
  wide_db = wide_db.applymap(lambda x: np.nan if isinstance(x, str) else x)

Scenarios to infill: 2it [00:43, 18.66s/it]

Scenarios to infill: 3it [00:46, 11.37s/it]

Scenarios to infill: 4it [00:48,  7.88s/it]

Scenarios to infill: 4it [00:48, 12.16s/it]

Out[20]:

					2015	2016	2017	2018	2019	2020	2021	2022	2023	2024	...	2091	2092	2093	2094	2095	2096	2097	2098	2099	2100
model	scenario	region	variable	unit
demo	decline	World	Emissions|BC	Mt BC/yr	9.727424	9.564061	9.401881	9.240866	9.081004	8.922277	8.685341	8.705159	8.516012	8.392241	...	2.011126	1.934743	1.867127	1.840313	1.857182	1.859431	1.838243	1.811239	1.767023	1.736738
			Emissions|CO2|Biosphere	Mt CO2/yr	3517.440000	3480.237715	3443.005653	3405.743780	3368.452064	3331.130471	3080.261088	2915.893438	2797.841161	2637.810556	...	-2438.632223	-2432.661660	-2468.082193	-2494.773144	-2480.293405	-2501.699275	-2503.906392	-2547.456889	-2589.320757	-2647.317321
			Emissions|CO	Mt CO/yr	934.349885	922.094402	909.821169	897.596560	885.397506	873.198440	865.136527	855.366801	850.678222	844.114675	...	341.415033	337.727682	332.581101	329.010905	326.072438	323.085590	319.795103	317.001125	313.858938	310.880515
			Emissions|N2O	kt N2O/yr	10900.000000	11026.740972	11154.624955	11282.911591	11413.351199	11544.291932	11380.715604	11305.954853	11285.326441	11243.493856	...	7122.040000	7099.081154	7074.093619	7053.353999	7045.696238	7034.066705	7029.109720	7028.755739	7009.095071	6986.439981
			Emissions|NH3	Mt NH3/yr	65.279703	65.407914	65.523298	65.623864	65.725851	65.844146	64.818786	65.242753	65.502840	65.578378	...	45.353082	43.947514	42.428801	40.867488	39.398676	40.952544	42.506413	43.868729	44.683288	44.258182
	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
	flatline-co2-only	World	Emissions|C7F16	kt C7F16/yr	0.233800	0.245713	0.257660	0.269573	0.281487	0.293400	0.277319	0.261282	0.245246	0.229209	...	0.003650	0.003400	0.003150	0.002900	0.002650	0.002400	0.002150	0.001900	0.001650	0.001400
			Emissions|C8F18	kt C8F18/yr	0.101400	0.106577	0.111769	0.116946	0.122123	0.127300	0.120317	0.113352	0.106388	0.099424	...	0.001590	0.001480	0.001370	0.001260	0.001150	0.001040	0.000930	0.000820	0.000710	0.000600
			Emissions|cC4F8	kt cC4F8/yr	1.267200	1.307930	1.348772	1.389503	1.430233	1.470964	1.413476	1.356145	1.298815	1.241484	...	0.134054	0.131128	0.128195	0.125269	0.122343	0.119418	0.116484	0.113559	0.110633	0.107708
			Emissions|SO2F2	kt SO2F2/yr	2.531700	2.613054	2.694631	2.775986	2.857340	2.938694	2.823824	2.709268	2.594712	2.480156	...	0.267860	0.262030	0.256185	0.250355	0.244526	0.238696	0.232851	0.227021	0.221192	0.215362
			Emissions|HFC245fa	kt HFC245fa/yr	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	...	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000

202 rows × 86 columns

You can see infilled pathways compared to raw pathways in the below. A few things to notice:

• only required variables are infilled (e.g. CH₄ is only infilled for flatline-co2-only, not flatline) and there can be notable differences in the emissions used without infilling
• the infilling is largely CO₂ driven, but it's not as simple as 'higher CO₂' means higher everything else (see e.g. the N₂O plot where flatline has lower N₂O than decline in 2060)

This is not a deep analysis. If you want to see the full details of how the infilling worked in AR6, see Lamboll et al., 2020.

In [21]:

pdf = (
    pix.concat(
        [
            pre_processed.pix.assign(stage="pre_processed"),
            harmonised.pix.assign(stage="harmonised"),
            infilled.pix.assign(stage="infilled"),
        ]
    )
    .loc[pix.ismatch(variable=["**CO2|Fossil", "**CH4", "**N2O", "**SOx"])]
    .melt(ignore_index=False, var_name="year")
    .reset_index()
)

fg = relplot_in_emms(
    data=pdf,
    hue="scenario",
    style="stage",
    dashes={
        "pre_processed": (3, 3),
        "harmonised": "",
        "infilled": (1, 1),
    },
)

fg.axes.flatten()[0].axhline(0.0, linestyle="--", color="gray")
fg.axes.flatten()[1].set_ylim(ymin=0.0)

pdf = ( pix.concat( [ pre_processed.pix.assign(stage="pre_processed"), harmonised.pix.assign(stage="harmonised"), infilled.pix.assign(stage="infilled"), ] ) .loc[pix.ismatch(variable=["**CO2|Fossil", "**CH4", "**N2O", "**SOx"])] .melt(ignore_index=False, var_name="year") .reset_index() ) fg = relplot_in_emms( data=pdf, hue="scenario", style="stage", dashes={ "pre_processed": (3, 3), "harmonised": "", "infilled": (1, 1), }, ) fg.axes.flatten()[0].axhline(0.0, linestyle="--", color="gray") fg.axes.flatten()[1].set_ylim(ymin=0.0)

Out[21]:

(0.0, 423.1851264903377)

The returned data is just the infilled timeseries. We can combine these with the harmonised data to create complete scenarios, ready for running our simple climate models.

In [22]:

complete_scenarios = pd.concat([harmonised, infilled])
complete_scenarios

complete_scenarios = pd.concat([harmonised, infilled]) complete_scenarios

Out[22]:

					2015	2016	2017	2018	2019	2020	2021	2022	2023	2024	...	2091	2092	2093	2094	2095	2096	2097	2098	2099	2100
model	scenario	region	variable	unit
demo	decline	World	Emissions|CO2|Fossil	Mt CO2/yr	35635.286300	35679.358818	35723.431337	35767.503855	35811.576374	35855.648892	35899.721411	35943.793929	35987.866448	36031.938966	...	-829.133715	-737.007747	-644.881779	-552.755810	-460.629842	-368.503873	-276.377905	-184.251937	-92.125968	0.000000
			Emissions|CH4	Mt CH4/yr	388.072796	388.410137	388.747479	389.084821	389.422162	389.759504	390.096845	390.434187	390.771529	391.108870	...	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000	150.000000
	decline_co2_fossil_split	World	Emissions|CO2|Fossil	Mt CO2/yr	35635.286300	35684.536703	35733.807989	35783.100158	35832.413211	35881.747147	35931.101967	35980.477670	36029.874257	36079.291727	...	447.448959	547.732408	648.015857	748.299306	848.582755	948.866204	1049.149653	1149.433102	1249.716551	1350.000000
	flatline	World	Emissions|CO2|Fossil	Mt CO2/yr	35635.286300	35679.358818	35723.431337	35767.503855	35811.576374	35855.648892	35899.721411	35943.793929	35987.866448	36031.938966	...	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000
			Emissions|CH4	Mt CH4/yr	388.072796	388.410137	388.747479	389.084821	389.422162	389.759504	390.096845	390.434187	390.771529	391.108870	...	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000	200.000000
	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
	flatline-co2-only	World	Emissions|C7F16	kt C7F16/yr	0.233800	0.245713	0.257660	0.269573	0.281487	0.293400	0.277319	0.261282	0.245246	0.229209	...	0.003650	0.003400	0.003150	0.002900	0.002650	0.002400	0.002150	0.001900	0.001650	0.001400
			Emissions|C8F18	kt C8F18/yr	0.101400	0.106577	0.111769	0.116946	0.122123	0.127300	0.120317	0.113352	0.106388	0.099424	...	0.001590	0.001480	0.001370	0.001260	0.001150	0.001040	0.000930	0.000820	0.000710	0.000600
			Emissions|cC4F8	kt cC4F8/yr	1.267200	1.307930	1.348772	1.389503	1.430233	1.470964	1.413476	1.356145	1.298815	1.241484	...	0.134054	0.131128	0.128195	0.125269	0.122343	0.119418	0.116484	0.113559	0.110633	0.107708
			Emissions|SO2F2	kt SO2F2/yr	2.531700	2.613054	2.694631	2.775986	2.857340	2.938694	2.823824	2.709268	2.594712	2.480156	...	0.267860	0.262030	0.256185	0.250355	0.244526	0.238696	0.232851	0.227021	0.221192	0.215362
			Emissions|HFC245fa	kt HFC245fa/yr	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	...	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000

208 rows × 86 columns

SCM Running¶

The next step is running a simple climate model (SCM). This provides information about the climate implications of the scenario(s).

In AR6, MAGICCv7.5.3 was used for categorisation, but other climate models were also available. MAGICCv7.5.3 can be downloaded from magicc.org, but a copy is also stored in the path below so we can run these docs. Please go and download from magicc.org to help the MAGICC developers see the usage of their model.

Under the hood, the AR6 SCM running uses the OpenSCM-Runner package.

In [23]:

MAGICC_EXE_PATH = Path("tests/regression/ar6/ar6-workflow-inputs/magicc-v7.5.3/bin")
MAGICC_AR6_PROBABILISTIC_CONFIG_FILE = Path(
    "tests/regression/ar6/ar6-workflow-inputs/magicc-ar6-0fd0f62-f023edb-drawnset/0fd0f62-derived-metrics-id-f023edb-drawnset.json"
)

MAGICC_EXE_PATH = Path("tests/regression/ar6/ar6-workflow-inputs/magicc-v7.5.3/bin") MAGICC_AR6_PROBABILISTIC_CONFIG_FILE = Path( "tests/regression/ar6/ar6-workflow-inputs/magicc-ar6-0fd0f62-f023edb-drawnset/0fd0f62-derived-metrics-id-f023edb-drawnset.json" )

In [25]:

if platform.system() == "Darwin":
    if platform.processor() == "arm":
        MAGICC_EXE = MAGICC_EXE_PATH / "magicc-darwin-arm64"
        os.environ["DYLD_LIBRARY_PATH"] = "/opt/homebrew/lib/gcc/current/"

elif platform.system() == "Linux":
    MAGICC_EXE = MAGICC_EXE_PATH / "magicc"

elif platform.system() == "Windows":
    MAGICC_EXE = MAGICC_EXE_PATH / "magicc.exe"

if platform.system() == "Darwin": if platform.processor() == "arm": MAGICC_EXE = MAGICC_EXE_PATH / "magicc-darwin-arm64" os.environ["DYLD_LIBRARY_PATH"] = "/opt/homebrew/lib/gcc/current/" elif platform.system() == "Linux": MAGICC_EXE = MAGICC_EXE_PATH / "magicc" elif platform.system() == "Windows": MAGICC_EXE = MAGICC_EXE_PATH / "magicc.exe"

With the MAGICC executable and config file, we can initialise our SCM runner.

In [26]:

scm_runner = AR6SCMRunner.from_ar6_config(
    # Generally, you want to run SCMs in parallel
    n_processes=multiprocessing.cpu_count(),
    magicc_exe_path=MAGICC_EXE,
    magicc_prob_distribution_path=MAGICC_AR6_PROBABILISTIC_CONFIG_FILE,
    historical_emissions=get_ar6_full_historical_emissions(AR6_INFILLING_DB_CFCS_FILE),
    harmonisation_year=2015,
    output_variables=("Surface Air Temperature Change", "Effective Radiative Forcing"),
)

scm_runner = AR6SCMRunner.from_ar6_config( # Generally, you want to run SCMs in parallel n_processes=multiprocessing.cpu_count(), magicc_exe_path=MAGICC_EXE, magicc_prob_distribution_path=MAGICC_AR6_PROBABILISTIC_CONFIG_FILE, historical_emissions=get_ar6_full_historical_emissions(AR6_INFILLING_DB_CFCS_FILE), harmonisation_year=2015, output_variables=("Surface Air Temperature Change", "Effective Radiative Forcing"), )

If you're reading this on RtD, note that we run a greatly reduced number of ensemble members. You will likely want to skip this step if running yourself.

In [27]:

if os.environ.get("READTHEDOCS", False):
    scm_runner.climate_models_cfgs["MAGICC7"] = scm_runner.climate_models_cfgs[
        "MAGICC7"
    ][:10]

if os.environ.get("READTHEDOCS", False): scm_runner.climate_models_cfgs["MAGICC7"] = scm_runner.climate_models_cfgs[ "MAGICC7" ][:10]

And then run

In [28]:

scm_results = scm_runner(complete_scenarios)
scm_results

scm_results = scm_runner(complete_scenarios) scm_results

Climate models:   0%|          | 0/1 [00:00<?, ?it/s]

Scenario batch:   0%|          | 0/1 [00:00<?, ?it/s]

Climate models:   0%|          | 0.00/1.00 [00:00<?, ?it/s]

[WARNING] 09:40:35 - openscm_runner.adapters.magicc7.magicc7: Historical data has not been checked

/home/docs/checkouts/readthedocs.org/user_builds/gcages/envs/latest/lib/python3.11/site-packages/scmdata/run.py:2632: FutureWarning: The behavior of DataFrame concatenation with empty or all-NA entries is deprecated. In a future version, this will no longer exclude empty or all-NA columns when determining the result dtypes. To retain the old behavior, exclude the relevant entries before the concat operation.
  pd.concat([ret._meta.to_frame(), *to_join_metas]).astype("category")

Writing SCEN7 files:   0%|          | 0.00/4.00 [00:00<?, ?it/s]

Writing SCEN7 files: 100%|██████████| 4.00/4.00 [00:00<00:00, 7.53it/s]

Front serial:   0%|          | 0.00/2.00 [00:00<?, ?it/s]

Front serial: 100%|██████████| 2.00/2.00 [00:01<00:00, 1.61it/s]

Front parallel:   0%|          | 0.00/2.00 [00:00<?, ?it/s]

Front parallel: 100%|██████████| 2.00/2.00 [00:01<00:00, 1.79it/s]

Parallel runs:   0%|          | 0.00/36.0 [00:00<?, ?it/s]

Parallel runs:  25%|██▌       | 9.00/36.0 [00:05<00:16, 1.64it/s]

Parallel runs:  50%|█████     | 18.0/36.0 [00:10<00:10, 1.71it/s]

Parallel runs:  75%|███████▌  | 27.0/36.0 [00:16<00:05, 1.68it/s]

Parallel runs: 100%|██████████| 36.0/36.0 [00:21<00:00, 1.72it/s]

Parallel runs: 100%|██████████| 36.0/36.0 [00:21<00:00, 1.71it/s]

Climate models: 100%|██████████| 1.00/1.00 [00:26<00:00, 26.8s/it]

Climate models: 100%|██████████| 1.00/1.00 [00:26<00:00, 26.8s/it]

Scenario batch: 100%|██████████| 1/1 [00:26<00:00, 26.82s/it]

Scenario batch: 100%|██████████| 1/1 [00:26<00:00, 26.82s/it]

Climate models: 100%|██████████| 1/1 [00:26<00:00, 26.83s/it]

Climate models: 100%|██████████| 1/1 [00:26<00:00, 26.83s/it]

Out[28]:

						time	1750	1751	1752	1753	1754	1755	1756	1757	1758	1759	...	2091	2092	2093	2094	2095	2096	2097	2098	2099	2100
climate_model	model	region	run_id	scenario	unit	variable
MAGICCv7.5.3	demo	World	0	decline	W/m^2	Effective Radiative Forcing	0.024862	0.118897	0.116943	0.101121	0.073664	0.010479	-0.012348	0.037891	0.084870	0.115642	...	2.507267	2.514207	2.511824	2.505728	2.498638	2.487797	2.483538	2.479657	2.480580	2.484025
					K	Surface Air Temperature Change	0.000000	0.025523	0.035245	0.038705	0.037403	0.035409	0.028824	0.013995	0.019164	0.031390	...	1.735712	1.738574	1.740756	1.741576	1.741184	1.739511	1.737953	1.736753	1.736441	1.737189
			1	decline	W/m^2	Effective Radiative Forcing	0.040476	0.188135	0.175385	0.153909	0.121071	0.013471	-0.034662	0.058462	0.133571	0.179039	...	2.447416	2.461051	2.457011	2.446263	2.435107	2.418911	2.412117	2.405989	2.408235	2.415898
					K	Surface Air Temperature Change	0.000000	0.031040	0.042205	0.048116	0.050462	0.050914	0.043648	0.027046	0.034479	0.048696	...	2.842276	2.849517	2.855698	2.859535	2.861623	2.862003	2.862165	2.862608	2.864175	2.867480
			2	decline	W/m^2	Effective Radiative Forcing	0.045260	0.197474	0.192773	0.165397	0.121218	0.011162	-0.031142	0.057258	0.136335	0.189962	...	2.593371	2.605526	2.597954	2.584181	2.571334	2.553665	2.545758	2.538146	2.539882	2.548180
			...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
			7	flatline-co2-only	K	Surface Air Temperature Change	0.000000	0.027917	0.037654	0.038351	0.031970	0.024944	0.016686	0.002665	0.008787	0.025196	...	2.727328	2.736132	2.741962	2.743607	2.742882	2.740108	2.737626	2.735960	2.736296	2.739440
			8	flatline-co2-only	W/m^2	Effective Radiative Forcing	0.027534	0.145660	0.143361	0.126693	0.097244	0.016528	-0.014465	0.051436	0.107386	0.143041	...	3.391445	3.389782	3.376594	3.361716	3.348160	3.331062	3.320102	3.308980	3.302919	3.300152
					K	Surface Air Temperature Change	0.000000	0.021493	0.027976	0.030634	0.030421	0.029793	0.024080	0.012661	0.018823	0.028366	...	1.828989	1.829130	1.828113	1.826124	1.823665	1.820430	1.817394	1.814462	1.812108	1.810528
			9	flatline-co2-only	W/m^2	Effective Radiative Forcing	0.089170	0.247113	0.224403	0.175709	0.109414	-0.024954	-0.078921	0.031023	0.144344	0.230172	...	3.295516	3.317332	3.295804	3.262913	3.235503	3.202411	3.187120	3.172362	3.176341	3.193911
					K	Surface Air Temperature Change	0.000000	0.038188	0.048259	0.050171	0.046314	0.041752	0.030263	0.011439	0.022490	0.041930	...	2.167417	2.175027	2.178616	2.177330	2.173765	2.168354	2.163687	2.159884	2.158575	2.160791

80 rows × 351 columns

With these outputs, we can look at raw (i.e. before pre-processing) variables.

In [29]:

scm_results.loc[
    pix.isin(variable=["Effective Radiative Forcing"])
].openscm.plot_plume_after_calculating_quantiles(
    style_var="variable",
    quantile_over="run_id",
    quantiles_plumes=(
        (0.5, 0.8),
        ((0.05, 0.95), 0.3),
    ),
)

scm_results.loc[ pix.isin(variable=["Effective Radiative Forcing"]) ].openscm.plot_plume_after_calculating_quantiles( style_var="variable", quantile_over="run_id", quantiles_plumes=( (0.5, 0.8), ((0.05, 0.95), 0.3), ), )

Out[29]:

<Axes: xlabel='time', ylabel='W/m^2'>

Post-processing¶

The last step is post-processing. This handles calculation of key pieces of metadata.

In [30]:

post_processor = AR6PostProcessor.from_ar6_config(n_processes=None)
post_processed_results = post_processor(scm_results)

post_processor = AR6PostProcessor.from_ar6_config(n_processes=None) post_processed_results = post_processor(scm_results)

For example, the scenario category.

In [31]:

post_processed_results.metadata_categories.unstack("metric")

post_processed_results.metadata_categories.unstack("metric")

Out[31]:

		metric	category	category_name
climate_model	model	scenario
MAGICCv7.5.3	demo	decline	C3	C3: limit warming to 2°C (>67%)
		decline_co2_fossil_split	C3	C3: limit warming to 2°C (>67%)
		flatline	C5	C5: limit warming to 2.5°C (>50%)
		flatline-co2-only	C5	C5: limit warming to 2.5°C (>50%)

Exceedance thresholds.

In [32]:

post_processed_results.metadata_exceedance_probabilities.unstack("threshold")

post_processed_results.metadata_exceedance_probabilities.unstack("threshold")

Out[32]:

						threshold	1.0	1.5	2.0	2.5	3.0	3.5	4.0
climate_model	model	region	scenario	unit	variable	threshold_unit
MAGICCv7.5.3	demo	World	decline	%	Exceedance probability	K	100.0	90.0	30.0	10.0	0.0	0.0	0.0
			decline_co2_fossil_split	%	Exceedance probability	K	100.0	80.0	20.0	10.0	0.0	0.0	0.0
			flatline	%	Exceedance probability	K	100.0	100.0	60.0	20.0	10.0	10.0	0.0
			flatline-co2-only	%	Exceedance probability	K	100.0	90.0	60.0	20.0	10.0	10.0	0.0

Key warming metrics.

In [33]:

post_processed_results.metadata_quantile.loc[
    pix.isin(quantile=[0.05, 0.5, 0.95])
].unstack(["quantile", "metric"]).round(2).sort_index(axis="columns")

post_processed_results.metadata_quantile.loc[ pix.isin(quantile=[0.05, 0.5, 0.95]) ].unstack(["quantile", "metric"]).round(2).sort_index(axis="columns")

Out[33]:

					quantile	0.05			0.50			0.95
					metric	2100	max	max_year	2100	max	max_year	2100	max	max_year
climate_model	model	region	scenario	unit	variable
MAGICCv7.5.3	demo	World	decline	K	Surface Temperature (GSAT)	1.07	1.40	NaN	1.61	1.82	NaN	2.45	2.52	NaN
				yr	Surface Temperature (GSAT)	NaN	NaN	2042.25	NaN	NaN	2047.0	NaN	NaN	2076.60
			decline_co2_fossil_split	K	Surface Temperature (GSAT)	1.13	1.37	NaN	1.69	1.79	NaN	2.56	2.57	NaN
				yr	Surface Temperature (GSAT)	NaN	NaN	2042.70	NaN	NaN	2048.0	NaN	NaN	2077.05
			flatline	K	Surface Temperature (GSAT)	1.44	1.66	NaN	2.11	2.20	NaN	3.29	3.29	NaN
				yr	Surface Temperature (GSAT)	NaN	NaN	2059.45	NaN	NaN	2061.0	NaN	NaN	2100.00
			flatline-co2-only	K	Surface Temperature (GSAT)	1.38	1.58	NaN	2.05	2.11	NaN	3.20	3.20	NaN
				yr	Surface Temperature (GSAT)	NaN	NaN	2058.45	NaN	NaN	2061.0	NaN	NaN	2097.30

Assessed surface temperatures.

In [34]:

post_processed_results.timeseries_quantile.loc[:, 2000:].openscm.plot_plume(
    style_var="variable",
    quantiles_plumes=(
        (0.5, 0.8),
        ((0.05, 0.95), 0.3),
    ),
)

post_processed_results.timeseries_quantile.loc[:, 2000:].openscm.plot_plume( style_var="variable", quantiles_plumes=( (0.5, 0.8), ((0.05, 0.95), 0.3), ), )

Out[34]:

<Axes: xlabel='time', ylabel='K'>