feat: Add default scaler

CHANGELOG.md

@@ -17,8 +17,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Added
 
+- Add `default_scaler` to the configuration for the scaler to be used if no specific scaler configuration is provided [#9](https://github.com/climate-resource/spaemis/pull/9)
 - Move test configuration to `test-data` directory  [#8](https://github.com/climate-resource/spaemis/pull/8)
-- Add functionality to write out an xarray dataset as a set of CSVs that are formatted the same as the input emissions inventory data  [#7](https://github.com/climate-resource/spaemis/pull/7)
+- Add functionality to write out a xarray dataset as a set of CSVs that are formatted the same as the input emissions inventory data  [#7](https://github.com/climate-resource/spaemis/pull/7)
 - Add relative_change scaler and reading of Input4MIPs data [#3](https://github.com/climate-resource/spaemis/pull/3)
 - Added CLI command `project` and a framework for scalers [#2](https://github.com/climate-resource/spaemis/pull/2)
-- Initial commit and repository setup
+- Initial commit and repository setup

src/spaemis/commands/project_command.py

@@ -4,20 +4,22 @@
 
 import logging
 import os.path
+from itertools import product
+from typing import Dict, Tuple
 
 import click
 import xarray as xr
 
 from spaemis.commands.base import cli
-from spaemis.config import DownscalingScenarioConfig, VariableConfig, load_config
+from spaemis.config import DownscalingScenarioConfig, VariableScalerConfig, load_config
 from spaemis.inventory import EmissionsInventory, load_inventory, write_inventory_csvs
 from spaemis.scaling import get_scaler_by_config
 
 logger = logging.getLogger(__name__)
 
 
 def scale_inventory(
-    cfg: VariableConfig, inventory: EmissionsInventory, target_year: int
+    cfg: VariableScalerConfig, inventory: EmissionsInventory, target_year: int
 ) -> xr.Dataset:
     """
     Scale a given variable/sector
@@ -68,10 +70,35 @@ def calculate_projections(
 
         The dimensionality of the output variables is (sector, year, lat, lon)
     """
+    scaling_configs: Dict[Tuple[str, str], VariableScalerConfig] = {
+        (cfg.variable, cfg.sector): cfg for cfg in config.scalers
+    }
+
+    if config.default_scaler:
+        # Add in additional scalers for each missing variable/sector
+        variables = inventory.data.data_vars.keys()
+        sectors = inventory.data["sector"].values
+
+        for variable, sector in product(variables, sectors):
+            scaling_configs.setdefault(
+                (variable, sector),
+                VariableScalerConfig(
+                    variable=variable,
+                    sector=sector,
+                    method=config.default_scaler,
+                ),
+            )
+
     projections = []
-    for variable_config in config.variables:
+
+    for variable_config in scaling_configs.values():
         for slice_year in config.timeslices:
-            logger.info(f"Processing year={slice_year}")
+            logger.info(
+                "Processing variable=%s sector=%s year=%i",
+                variable_config.variable,
+                variable_config.sector,
+                slice_year,
+            )
             res = scale_inventory(variable_config, inventory, slice_year)
             projections.append(res)
 

src/spaemis/config.py

@@ -2,7 +2,7 @@
 Description of the configuration
 """
 
-from typing import Any, ClassVar, Literal, Type, Union, get_args
+from typing import Any, ClassVar, Literal, Optional, Type, Union, get_args
 
 from attrs import define
 from cattrs.preconf.pyyaml import make_converter
@@ -42,7 +42,7 @@ def _discriminate_scaler(value: Any, _klass: Type) -> ScalerMethod:
 
 
 @define
-class VariableConfig:
+class VariableScalerConfig:
     variable: str
     sector: str
     method: ScalerMethod
@@ -57,7 +57,8 @@ class DownscalingScenarioConfig:
     inventory_name: str
     inventory_year: int
     timeslices: list[int]
-    variables: list[VariableConfig]
+    scalers: list[VariableScalerConfig]
+    default_scaler: Optional[ScalerMethod] = None
 
 
 def load_config(config_file: str) -> DownscalingScenarioConfig:

src/spaemis/config/scenarios/ssp245.yaml

@@ -5,26 +5,30 @@ timeslices:
   - 2020
   - 2040
   - 2060
+  - 2080
 
-variables:
+default_scaler:
+  name: constant
+
+scalers:
   - variable: NOx
     sector: industry
     method:
       name: relative_change
-      source_id: IAMC-MESSAGE-GLOBIOM-ssp245-1-1
+      source_id: &source IAMC-MESSAGE-GLOBIOM-ssp245-1-1
       variable_id: NOx-em-anthro
       sector: Industrial Sector
   - variable: NOx
     sector: motor_vehicles
     method:
       name: relative_change
-      source_id: IAMC-MESSAGE-GLOBIOM-ssp245-1-1
+      source_id: *source
       variable_id: NOx-em-anthro
       sector: Transportation Sector
   - variable: CO
     sector: industry
     method:
       name: relative_change
-      source_id: IAMC-MESSAGE-GLOBIOM-ssp245-1-1
+      source_id: *source
       variable_id: CO-em-anthro
       sector: Industrial Sector

tests/test-data/config/test-config.yaml

@@ -6,7 +6,7 @@ timeslices:
   - 2040
   - 2060
 
-variables:
+scalers:
   - variable: NOx
     sector: industry
     method:

tests/unit/commands/test_project.py

@@ -6,7 +6,12 @@
 
 from spaemis.commands import cli
 from spaemis.commands.project_command import calculate_projections, scale_inventory
-from spaemis.config import VariableConfig, converter, load_config
+from spaemis.config import (
+    ConstantScaleMethod,
+    VariableScalerConfig,
+    converter,
+    load_config,
+)
 
 
 def test_cli_project(runner, config_file, tmpdir, mocker, inventory):
@@ -59,7 +64,7 @@ def test_scale_inventory_missing_variable(inventory):
             "sector": "Industrial",
             "method": {"name": "constant"},
         },
-        VariableConfig,
+        VariableScalerConfig,
     )
     with pytest.raises(ValueError, match="Variable missing not available in inventory"):
         scale_inventory(config, inventory, 2040)
@@ -72,7 +77,7 @@ def test_scale_inventory_missing_sector(inventory):
             "sector": "unknown",
             "method": {"name": "constant"},
         },
-        VariableConfig,
+        VariableScalerConfig,
     )
     with pytest.raises(ValueError, match="Sector unknown not available in inventory"):
         scale_inventory(config, inventory, 2040)
@@ -85,7 +90,7 @@ def test_scale_inventory_constant(inventory):
             "sector": "rail",
             "method": {"name": "constant"},
         },
-        VariableConfig,
+        VariableScalerConfig,
     )
     res = scale_inventory(config, inventory, 2040)
     assert isinstance(res, xr.Dataset)
@@ -111,7 +116,7 @@ def test_scale_inventory_relative(inventory):
                 "sector": "Transportation Sector",
             },
         },
-        VariableConfig,
+        VariableScalerConfig,
     )
     res = scale_inventory(config, inventory, 2040)
     assert isinstance(res, xr.Dataset)
@@ -143,3 +148,25 @@ def test_calculate_projections(config, inventory):
     assert res["CO"].sel(sector="motor_vehicles").isnull().all()
     # but CO|industry should have data
     assert not res["CO"].sel(sector="industry").isnull().all()
+
+
+def test_calculate_projections_with_default(config, inventory):
+    config.default_scaler = ConstantScaleMethod()
+
+    res = calculate_projections(config, inventory)
+
+    assert (res["sector"] == inventory.data["sector"]).all()
+
+    # CO|architect_coating should be held constant
+    xr.testing.assert_allclose(
+        res["CO"]
+        .sel(sector="architect_coating", year=2040)
+        .reset_coords("year", drop=True),
+        inventory.data["CO"].sel(sector="architect_coating"),
+    )
+    # CO|industry should be scaled
+    with pytest.raises(AssertionError):
+        xr.testing.assert_allclose(
+            res["CO"].sel(sector="industry", year=2040).reset_coords("year", drop=True),
+            inventory.data["CO"].sel(sector="industry"),
+        )