Merge branch 'fix-scmdf_to_emissions'

CHANGELOG.rst

@@ -3,6 +3,9 @@ Changelog
 
 master
 ------
+
+(`#77 <https://github.com/OMS-NetZero/FAIR/pull/77>`_) Fixed ``fair.tools.scmdf.scmdf_to_emissions``
+
 (`#76 <https://github.com/OMS-NetZero/FAIR/pull/76>`_) Added in the GIR carbon cycle as an option and ScmDataFrame reader, required for openscm-runner and iiasa-climate-assessment
 
 (`#69 <https://github.com/OMS-NetZero/FAIR/pull/69>`_) Added in switch to directly specify tropospheric ozone forcing time series and added in AR6 radiative efficiencies, lifetimes and molecular weights

fair/tools/scmdf.py

@@ -7,10 +7,60 @@
 
 from ..constants import molwt
 
-def scmdf_to_emissions(scmdf, include_cfcs=True, startyear=1765, endyear=2100):
+try:
+    from scmdata import ScmDataFrame
+    has_scmdata = True
+except ImportError:
+    has_scmdata = False
+
+
+EMISSIONS_SPECIES_UNITS_CONTEXT = (  # in fair 1.6, order is important
+    # @chrisroadmap can you check please
+    ('|CO2|MAGICC Fossil and Industrial', 'GtC / yr', None),
+    ('|CO2|MAGICC AFOLU', 'GtC / yr', None),
+    ('|CH4', 'MtCH4 / yr', None),
+    ('|N2O', 'MtN / yr', None),
+    ('|Sulfur', 'MtS / yr', None),
+    ('|CO', 'MtCO / yr', None),
+    ('|VOC', 'MtNMVOC / yr', None),
+    ('|NOx', 'MtN / yr', "NOx_conversions"),
+    ('|BC', 'MtBC / yr', None),
+    ('|OC', 'MtOC / yr', None),
+    ('|NH3', 'MtN / yr', None),
+    ('|CF4', 'ktCF4 / yr', None),
+    ('|C2F6', 'ktC2F6 / yr', None),
+    ('|C6F14', 'ktC6F14 / yr', None),
+    ('|HFC23', 'ktHFC23 / yr', None),
+    ('|HFC32', 'ktHFC32 / yr', None),
+    ('|HFC4310mee', 'ktHFC4310mee / yr', None),
+    ('|HFC125', 'ktHFC125 / yr', None),
+    ('|HFC134a', 'ktHFC134a / yr', None),
+    ('|HFC143a', 'ktHFC143a / yr', None),
+    ('|HFC227ea', 'ktHFC227ea / yr', None),
+    ('|HFC245fa', 'ktHFC245fa / yr', None),
+    ('|SF6', 'ktSF6 / yr', None),
+    ('|CFC11', 'ktCFC11 / yr', None),
+    ('|CFC12', 'ktCFC12 / yr', None),
+    ('|CFC113', 'ktCFC113 / yr', None),
+    ('|CFC114', 'ktCFC114 / yr', None),
+    ('|CFC115', 'ktCFC115 / yr', None),
+    ('|CCl4', 'ktCCl4 / yr', None),
+    ('|CH3CCl3', 'ktCH3CCl3 / yr', None),
+    ('|HCFC22', 'ktHCFC22 / yr', None),
+    ('|HCFC141b', 'ktHCFC141b / yr', None),
+    ('|HCFC142b', 'ktHCFC142b / yr', None),
+    ('|Halon1211', 'ktHalon1211 / yr', None),
+    ('|Halon1202', 'ktHalon1202 / yr', None),
+    ('|Halon1301', 'ktHalon1301 / yr', None),
+    ('|Halon2402', 'ktHalon2402 / yr', None),
+    ('|CH3Br', 'ktCH3Br / yr', None),
+    ('|CH3Cl', 'ktCH3Cl / yr', None),
+)
+
 
+def scmdf_to_emissions(scmdf, include_cfcs=True, startyear=1765, endyear=2100):
     """
-    Opens a ScmDataFrame and extracts the data. Interpolates linearly
+    Opens an ScmDataFrame and extracts the data. Interpolates linearly
     between non-consecutive years in the SCEN file. Fills in chlorinated gases
     from a specified SSP scenario.
 
@@ -24,12 +74,12 @@ def scmdf_to_emissions(scmdf, include_cfcs=True, startyear=1765, endyear=2100):
             MAGICC files do not come loaded with CFCs (indices 24-39).
             - if True, use the values from RCMIP for SSPs (all scenarios are
                 the same).
-            - Use False to ignore and create a 24-species emission file.
+            - Use False to ignore and create a 23-species emission file.
         startyear: First year of output file.
         endyear: Last year of output file.
 
     Returns:
-        nt x 40 numpy emissions array
+        nt x 40 numpy emissions array (nt x 23 if ``include_cfcs`` is ``False``)
     """
 
     # We expect that aeneris and silicone are going to give us a nicely
@@ -39,88 +89,72 @@ def scmdf_to_emissions(scmdf, include_cfcs=True, startyear=1765, endyear=2100):
     # historical.
     # This adapter will not be tested on anything else!
 
-    n_timepoints = scmdf.shape[1]
     n_cols = 40
     nt = endyear - startyear + 1
 
     data_out = np.ones((nt, n_cols)) * np.nan
     data_out[:,0] = np.arange(startyear, endyear+1)
 
+    if not has_scmdata:
+        raise ImportError("This is not going to work without having scmdata installed")
+
+    if not isinstance(scmdf, ScmDataFrame):
+        raise TypeError("scmdf must be an scmdata.ScmDataFrame instance")
+
+    if not include_cfcs:
+        raise NotImplementedError("include_cfcs equal to False")
+
+    if scmdf[["model", "scenario"]].drop_duplicates().shape[0] != 1:
+        raise AssertionError("Should only have one model-scenario pair")
+
     # fill in 1765 to 2014 from SSP emissions
-    ssp_df = pd.read_csv(os.path.join(os.path.dirname(__file__), '../SSPs/data/rcmip-emissions-annual-means-4-0-0-ssp-only.csv'))
+    ssp_df = ScmDataFrame(os.path.join(os.path.dirname(__file__), '../SSPs/data/rcmip-emissions-annual-means-4-0-0-ssp-only.csv'))
 
-    years = scmdf.columns
+    years = scmdf["year"].values
     first_scenyear = years[0]
     last_scenyear = years[-1]
-    first_row = int(first_scenyear-startyear)
-    last_row = int(last_scenyear-startyear)
-
-    species = [  # in fair 1.6, order is important
-        '|CO2|Energy and Industrial Processes',
-        '|CO2|AFOLU',
-        '|CH4',
-        '|N2O',
-        '|Sulfur',
-        '|CO',
-        '|VOC',
-        '|NOx',
-        '|BC',
-        '|OC',
-        '|NH3',
-        '|CF4',
-        '|C2F6',
-        '|C6F14',
-        '|HFC23',
-        '|HFC32',
-        '|HFC43-10',
-        '|HFC125',
-        '|HFC134a',
-        '|HFC143a',
-        '|HFC227ea',
-        '|HFC245ca',
-        '|SF6',
-    ]
-
-    emissions_file_species = species.copy()
-    emissions_file_species[0] = '|CO2|MAGICC Fossil and Industrial'
-    emissions_file_species[1] = '|CO2|MAGICC AFOLU'
-    emissions_file_species[16] = '|HFC4310mee'
-    emissions_file_species[21] = '|HFC245fa'
-    emissions_file_species.extend([
-        '|CFC11',
-        '|CFC12',
-        '|CFC113',
-        '|CFC114',
-        '|CFC115',
-        '|CCl4',
-        '|CH3CCl3',
-        '|HCFC22',
-        '|HCFC141b',
-        '|HCFC142b',
-        '|Halon1211',
-        '|Halon1202',
-        '|Halon1301',
-        '|Halon2402',
-        '|CH3Br',
-        '|CH3Cl',
-    ])
-
-    # Assume that units coming out of aneris don't change. One day I'll do unit parsing
-    unit_convert = np.ones(40)
-    unit_convert[1] = molwt.C/molwt.CO2/1000
-    unit_convert[2] = molwt.C/molwt.CO2/1000
-    unit_convert[4] = molwt.N2/molwt.N2O/1000
-    unit_convert[5] = molwt.S/molwt.SO2
-    unit_convert[8] = molwt.N/molwt.NO2
-
-    years_future = [2015] + list(range(2020,2501,10))
-    for i, specie in enumerate(emissions_file_species):
-        data_out[:first_row,i+1] = ssp_df.loc[(ssp_df['Scenario']=='ssp245')&(ssp_df['Variable'].str.endswith(specie)),str(startyear):'2014']*unit_convert[i+1]
-        if i<23:
-            f = interp1d(years, scmdf[scmdf.index.get_level_values('variable').str.endswith(species[i])])
-            data_out[first_row:(last_row+1), i+1] = f(np.arange(first_scenyear, last_scenyear+1))*unit_convert[i+1]
+    first_scen_row = int(first_scenyear-startyear)
+    last_scen_row = int(last_scenyear-startyear)
+
+    for i, (specie, unit, context) in enumerate(EMISSIONS_SPECIES_UNITS_CONTEXT):
+        data_out[:first_scen_row, i+1] = ssp_df.filter(
+            variable="*{}".format(specie),
+            region="World",
+            scenario="ssp245",
+            year=range(startyear, 2015)
+        ).convert_unit(unit, context=context).values.squeeze()
+
+        if i < 23:
+            if not any([specie in v for v in scmdf.get_unique_meta("variable")]):
+                raise AssertionError("{} not available in scmdf".format(specie))
+
+            f = interp1d(
+                years,
+                scmdf.filter(
+                    variable="*{}".format(specie),
+                    region="World"
+                ).convert_unit(unit, context=context).values.squeeze()
+            )
+            data_out[first_scen_row:(last_scen_row+1), i+1] = f(
+                np.arange(first_scenyear, last_scenyear+1)
+            )
+
         else:
-            f = interp1d(years_future, ssp_df.loc[(ssp_df['Scenario']=='ssp245')&(ssp_df['Variable'].str.endswith(specie)),'2015':'2500'].dropna(axis=1))
-            data_out[first_row:(last_row+1), i+1] = f(np.arange(first_scenyear, last_scenyear+1))*unit_convert[i+1]
+            if not any([specie in v for v in ssp_df.get_unique_meta("variable")]):
+                raise AssertionError("{} not available in ssp_df".format(specie))
+
+            filler_data = ssp_df.filter(
+                scenario="ssp245",
+                variable="*{}".format(specie),
+                year=range(2015, 2500 + 1),
+            )
+
+            f = interp1d(
+                filler_data["year"].values,
+                filler_data.convert_unit(unit, context=context).values.squeeze()
+            )
+            data_out[first_scen_row:(last_scen_row+1), i+1] = f(
+                np.arange(first_scenyear, last_scenyear+1)
+            )
 
     return data_out

setup.py

@@ -38,7 +38,7 @@ def readme():
     zip_safe=False,
     extras_require={
         'docs': ['sphinx>=1.4', 'nbsphinx'],
-        'dev' : ['notebook', 'wheel', 'twine'],
+        'dev' : ['notebook', 'scmdata<0.6', 'wheel', 'twine'],
         'test': ['pytest>=4.0', 'nbval', 'pytest-cov', 'codecov']
     }
 )