Add validation test for emissions factor references provided in zone configs

config/zones/CA-AB.yaml

@@ -78,7 +78,7 @@ emissionFactors:
         source: Electricity Maps, 2021 average
         value: 483.05444890890203
     unknown:
-      source: 'Battle River dual fuel power plant: 50% coal and 50% natural gas'
+      source: 'assumes Battle River dual fuel power plant: 50% coal and 50% natural gas'
       value: 655
 fallbackZoneMixes:
   powerOriginRatios:

config/zones/CH.yaml

@@ -76,7 +76,7 @@ emissionFactors:
       _comment:
         'Source ENTSOE, Swiss government and Electricity Maps: https://github.com/electricitymaps/electricitymaps-contrib/pull/2898
         and https://docs.google.com/spreadsheets/d/1FLHQ6e9Es08BIqX654BM3SEb_fuAk4k4O-6cmRXyw_E/'
-      source: 52% hydro storage, 48% unknown thermal
+      source: assumes 52% hydro storage, 48% unknown thermal
       value: 288
   lifecycle:
     battery discharge:
@@ -147,7 +147,7 @@ emissionFactors:
       _comment:
         'Source ENTSOE, Swiss government and Electricity Maps: https://github.com/electricitymaps/electricitymaps-contrib/pull/2898
         and https://docs.google.com/spreadsheets/d/1FLHQ6e9Es08BIqX654BM3SEb_fuAk4k4O-6cmRXyw_E/'
-      source: 52% hydro storage, 48% unknown thermal
+      source: assumes 52% hydro storage, 48% unknown thermal
       value: 383
 estimation_method: RECONSTRUCT_BREAKDOWN
 fallbackZoneMixes:

config/zones/GT.yaml

@@ -64,7 +64,7 @@ emissionFactors:
         value: 269.6478450016261
     unknown:
       _url: https://docs.google.com/spreadsheets/d/1CegROfej9HqRZTfihpjPpgZTYPUHrTNgTdHWbbP3w74/edit#gid=291258352
-      source: Guatemala AMM 2021-2022 and IEA 2020
+      source: Guatemala AMM 2021-2022; IEA 2020
       value: 383.240577
   lifecycle:
     battery discharge:
@@ -113,7 +113,7 @@ emissionFactors:
         value: 311.9991841222269
     unknown:
       _url: https://docs.google.com/spreadsheets/d/1CegROfej9HqRZTfihpjPpgZTYPUHrTNgTdHWbbP3w74/edit#gid=291258352
-      source: Guatemala AMM 2021-2022 and IEA 2020
+      source: Guatemala AMM 2021-2022; IEA 2020
       value: 549.9988504
 fallbackZoneMixes:
   powerOriginRatios:

config/zones/IN-KA.yaml

@@ -86,7 +86,7 @@ emissionFactors:
       _comment: 'Source: Souther Regional Load Dispatch Center Allocation Limits'
       _url:
         - http://srldc.org/2018-19srallocation.aspx
-      source: calculated 20% nuclear, 80% thermal (coal)
+      source: assumes calculated 20% nuclear, 80% thermal (coal)
       value: 658.4
 fallbackZoneMixes:
   powerOriginRatios:

config/zones/PR.yaml

@@ -129,8 +129,7 @@ emissionFactors:
         value: 604.8374514356746
     unknown:
       _comment: 'Source: https://indicadores.pr/dataset/generacion-consumo-costo-ingresos-y-clientes-del-sistema-electrico-de-puerto-rico/resource/fdad4f42-a4be-48bb-9478-a8fb75c000c6'
-      source: Based on average renewable generation October 2019-September 2020 (12%
-        hydro, 57% solar, 31% wind)
+      source: assumes 12% hydro, 57% solar, 31% wind based on average renewable generation October 2019-September 2020
       value: 31.94
 fallbackZoneMixes:
   powerOriginRatios:

config/zones/UA.yaml

@@ -90,7 +90,7 @@ emissionFactors:
       _comment: estimated yearly share of solar & wind in this mixed renewable category
         based on installed capacity in 2018
       _url: https://en.wikipedia.org/wiki/Renewable_energy_in_Ukraine
-      source: renewable mix; assumes 50% solar, 50% wind
+      source: assumes 50% solar, 50% wind renewable mix
       value: 28
 fallbackZoneMixes:
   powerOriginRatios:

electricitymap/contrib/config/co2eq_parameters.py

@@ -58,6 +58,5 @@ def generate_co2eq_parameters(
                         co2eq_parameters_lifecycle["emissionFactors"]["zoneOverrides"][
                             zone_key
                         ] = zone_config["emissionFactors"][k]
-            del zone_config["emissionFactors"]
 
     return co2eq_parameters_all, co2eq_parameters_direct, co2eq_parameters_lifecycle

electricitymap/contrib/config/model.py

@@ -96,29 +96,6 @@ class Delays(StrictBaseModel):
     productionPerUnit: PositiveInt | None
 
 
-class Zone(StrictBaseModelWithAlias):
-    bounding_box: list[Point] | None
-    bypass_aggregation_checks: list[ZoneKey] | None = Field(
-        [], alias="bypassedSubZones"
-    )
-    capacity: Capacity | None
-    comment: str | None = Field(None, alias="_comment")
-    contributors: list[str] | None
-    delays: Delays | None
-    disclaimer: str | None
-    parsers: Parsers = Parsers()
-    price_displayed: bool | None
-    aggregates_displayed: list[str] | None
-    sub_zone_names: list[ZoneKey] | None = Field(None, alias="subZoneNames")
-    timezone: str | None
-    key: ZoneKey  # This is not part of zones/{zone_key}.yaml, but added here to enable self referencing
-    estimation_method: str | None
-    sources: dict[str, Source] | None
-
-    def neighbors(self) -> list[ZoneKey]:
-        return ZONE_NEIGHBOURS.get(self.key, [])
-
-
 class ExchangeParsers(ParsersBaseModel):
     exchange: str | None
     exchangeForecast: str | None
@@ -292,6 +269,35 @@ class CO2eqParameters(StrictBaseModelWithAlias):
     emission_factors: EmissionFactors = Field(alias="emissionFactors")
 
 
+class ZoneEmissionFactors(StrictBaseModel):
+    direct: AllModesEmissionFactors | None
+    lifecycle: AllModesEmissionFactors
+
+
+class Zone(StrictBaseModelWithAlias):
+    bounding_box: list[Point] | None
+    bypass_aggregation_checks: list[ZoneKey] | None = Field(
+        [], alias="bypassedSubZones"
+    )
+    capacity: Capacity | None
+    comment: str | None = Field(None, alias="_comment")
+    contributors: list[str] | None
+    delays: Delays | None
+    disclaimer: str | None
+    parsers: Parsers = Parsers()
+    price_displayed: bool | None
+    aggregates_displayed: list[str] | None
+    sub_zone_names: list[ZoneKey] | None = Field(None, alias="subZoneNames")
+    timezone: str | None
+    key: ZoneKey  # This is not part of zones/{zone_key}.yaml, but added here to enable self referencing
+    estimation_method: str | None
+    sources: dict[str, Source] | None
+    emission_factors: ZoneEmissionFactors | None = Field(None, alias="emissionFactors")
+
+    def neighbors(self) -> list[ZoneKey]:
+        return ZONE_NEIGHBOURS.get(self.key, [])
+
+
 class ConfigModel(StrictBaseModel):
     exchanges: dict[str, Exchange]
     zones: dict[str, Zone]

tests/test_config_model.py

@@ -11,6 +11,45 @@ def test_pydantic_model(self):
             CONFIG_MODEL.zones["US-NW-PSCO"].parsers.get_function("production")
         )
 
+    # Add well-known sources that don't require config-based references here
+    GLOBAL_SOURCE_REFERENCES = {
+        "2018 average estimated from https://www.hops.hr/page-file/oEvvKj779KAhmQg10Gezt2/temeljni-podaci/Temeljni%20podaci%202018.pdf",
+        "2020 average by Electricity Maps. See disclaimer",
+        "BEIS 2021",
+        "CEA 2022",
+        "EIA 2020/BEIS 2021",
+        "Enerdata 2022",
+        "EU-ETS 2021",
+        "EU-ETS, ENTSO-E 2021",
+        "Guatemala AMM 2021-2022",
+        "IEA 2019",
+        "IEA 2020",
+        'Oberschelp, Christopher, et al. "Global emission hotspots of coal power generation."',
+        "Tidal (IPCC 2014)",
+        "https://www.iea.org/data-and-statistics/charts/electricity-generation-mix-in-mexico-1-jan-30-sep-2019-and-2020",
+    }
+
+    def test_zone_sources(self):
+        for _, zone in CONFIG_MODEL.zones.items():
+            for _, production_mode in zone.emission_factors or ():
+                for _, estimate in production_mode or ():
+                    if estimate is None:
+                        continue
+                    estimates = estimate if isinstance(estimate, list) else [estimate]
+                    for estimate in estimates:
+                        source_list = estimate.source
+                        if source_list is None:
+                            continue
+                        for source in source_list.split(";"):
+                            source = source.strip()
+                            if source.startswith("assumes"):
+                                continue
+                            if source.startswith("Electricity Maps"):
+                                continue
+                            if source in self.GLOBAL_SOURCE_REFERENCES:
+                                continue
+                            self.assertIn(source, zone.sources)
+
 
 if __name__ == "__main__":
     unittest.main(buffer=True)