Choosing a value: the axes
Numbers checked against the live data on
The default row, is_default_indicator AND is_latest_activity_year, gives one sensible number per factor. When you need another one, such as a different GWP characterization, a market-based grid mix or an earlier year, drop the flag that hides it and filter the axis you want instead. The values of one factor are always alternatives: select one per axis, never add them.
What the default collapses#
The two flags work on independent axes. You can relax one and keep the other.
| flag | axis it collapses | true on |
|---|---|---|
is_default_indicator |
Methodology: GWP set, horizon, characterization model, price basis, grid mix | One value per factor, release and reference year: the value the import elects as the sensible default |
is_latest_activity_year |
Time: the accounting years a value applies to | The rows for the newest accounting year the factor covers |
A third filter is already applied for you: factors_flat serves each library's current release only. A publisher that ships two co-valid document lines at once (a stream), or two system models, has one current release per line and the view serves each of them. Among the public libraries two do that today. AGRIBALYSE's food table (release 3.2) and its farm-gate agriculture line (stream agriculture, release agriculture-3.2-dv14) are both current, and so are ADEME's Base Carbone (release 23.11) and its Base IMPACTS line (stream base-impacts, release BI-3.0). Every other public library has one current release, and system_model is NULL throughout. ADEME keeps its previous Base Carbone edition, 23.6, and DESNZ its 2025 edition, as releases the view no longer serves.
Pin a factor value to a release shows how to hold on to a number when the next one lands.
Dropping is_default_indicator returns every methodology sibling. Dropping is_latest_activity_year returns one row per accounting year the factor covers, which is what Get the value for a past year filters.
The axes#
The populated on column names the public libraries these docs write their examples against. Other libraries in factor search fill the same columns, so read it as "where to see this axis at work", not as the full list of publishers that carry it.
| axis | column | kind | populated on |
|---|---|---|---|
| Accounting year | applies_from_year, applies_to_year |
Select one | Every factor. See Reference year. |
| GWP set and horizon | gwp_method_native, gwp_horizon_years |
Select one | Several per factor on EXIOBASE |
| Characterization model | characterization_model (on emission_factor_value) |
Select one | Several per factor on EXIOBASE; see GWP for the rest |
| Price basis | price_basis, currency_year |
Match your spend: one basis per factor | Spend factors: EXIOBASE and ADEME (basic), Open CEDA (producer and purchaser, one factor each), USEEIO (purchaser) |
| Grid mix | grid_mix |
Select one | Electricity: ADEME, AIB, DESNZ, EPA, MITECO, ÖKOBAUDAT |
| Calculation approach | calculation_approach |
Dual report: keep both, report apart | Scope 2 electricity, same libraries as grid mix |
| Radiative forcing | radiative_forcing |
Select one, and say which | Aviation: ADEME, DESNZ |
| EPD scenario and other qualifiers | value_qualifiers |
Select one | ÖKOBAUDAT, EXIOBASE |
| Supplier | supplier_code, supplier_name |
Descriptor, not an axis | ÖKOBAUDAT, MITECO |
| Representativeness | data_type |
Descriptor: specific, generic, average |
Every factor |
A dual report is still never a sum. The GHG Protocol asks for location-based and market-based Scope 2 side by side; each is one selected value.
Treatment route is not one of these axes. The landfill row and the recycling row for one waste material are separate factors, not two values of one factor, so no column in this table chooses between them. Waste treatment routes and the elected default explains which row the data elects when the route is unknown, and Which waste row to use when the route is unknown has the query.
Per library, how many top-level factors and rows the latest year holds, and how many factors carry each of the axes a value can vary on: grid mix, radiative forcing, price basis and qualifiers.
Show the SQL
SELECT f.library,
count(DISTINCT f.factor_id) AS factors,
count(*) AS values_in_latest_year,
count(DISTINCT f.factor_id) FILTER (WHERE f.grid_mix IS NOT NULL) AS with_grid_mix,
count(DISTINCT f.factor_id) FILTER (WHERE f.radiative_forcing IS NOT NULL) AS with_radiative_forcing,
count(DISTINCT f.factor_id) FILTER (WHERE f.price_basis IS NOT NULL) AS with_price_basis,
count(DISTINCT f.factor_id) FILTER (WHERE f.value_qualifiers IS NOT NULL) AS with_qualifiers
FROM open_ef.factors_flat f
WHERE f.library IN ('ademe','agribalyse','exiobase','desnz','miterd','aib','epa','useeio','okobaudat')
AND f.is_latest_activity_year
AND f.is_top_parent
GROUP BY 1
ORDER BY 1
| library | factors | values_in_latest_year | with_grid_mix | with_radiative_forcing | with_price_basis | with_qualifiers |
|---|---|---|---|---|---|---|
| ademe | 3148 | 3186 | 203 | 94 | 59 | 0 |
| agribalyse | 2879 | 2879 | 0 | 0 | 0 | 0 |
| aib | 34 | 130 | 34 | 0 | 0 | 0 |
| desnz | 1111 | 1111 | 76 | 18 | 0 | 0 |
| epa | 360 | 360 | 28 | 0 | 0 | 0 |
| exiobase | 14805 | 103635 | 0 | 0 | 14805 | 6741 |
| miterd | 730 | 731 | 534 | 0 | 0 | 0 |
| okobaudat | 4130 | 4130 | 3 | 0 | 0 | 317 |
| useeio | 392 | 392 | 0 | 0 | 392 | 0 |
Open CEDA is left out of the query for speed: its 119,076 factors carry one value each, 59,538 at producer price and 59,538 at purchaser price. Where values_in_latest_year exceeds factors, some factors have siblings; the sections below show each kind.
All 34 of AIB's top-level factors are totals open-climate.ai derived, as are 92 of EPA's 360 and 72 of MITECO's 730: the publisher's own value plus the legs it never printed, with the publisher's value one level down as a part (estimated electricity legs, estimated fuel legs). The totals and legs are is_estimated. For the publishers' own numbers, leave the estimated rows out and take the published factor beneath each total in its place: NOT is_estimated AND (is_top_parent OR parent_id IN (SELECT factor_id FROM open_ef.factors_flat WHERE is_estimated)) gives back AIB's 34, EPA's 360 and MITECO's 730 (how to leave the legs out). EPA's electricity totals carry the production mix only, so its marginal mix is on the publisher's generation factors one level down, never on a top-level row.
GWP characterizations: EXIOBASE#
EXIOBASE ships each sector's footprint under seven characterizations. The default is the AR5 100-year set; the other six stay available.
The methodologies one EXIOBASE factor is served on for its own accounting year, the default first, with the value on each.
Show the SQL
SELECT f.gwp_method_native, f.gwp_horizon_years, v.characterization_model, f.is_default_indicator,
round(f.value_co2e_native::numeric, 5) AS value_co2e_native
FROM open_ef.factors_flat f
JOIN open_ef.emission_factor_value v ON v.id = f.value_id
WHERE f.library = 'exiobase'
AND f.slug = 'ef-exiobase-animal_products_nec-fr-eur-cradle_to_gate-e2b36c2f'
AND f.applies_from_year = f.currency_year
ORDER BY f.is_default_indicator DESC, f.gwp_method_native, f.gwp_horizon_years, v.characterization_model
| gwp_method_native | gwp_horizon_years | characterization_model | is_default_indicator | value_co2e_native |
|---|---|---|---|---|
| ar5 | 100 | ipcc_ar5_direct | true | 3.76018 |
| ar4 | 20 | cml_1999 | false | 7.88503 |
| ar4 | 100 | cml_1999_baseline | false | 3.58586 |
| ar4 | 100 | cml_2001_baseline | false | 3.58583 |
| ar4 | 100 | houghton_net_max | false | 3.42572 |
| ar4 | 100 | houghton_net_min | false | 3.41286 |
| ar4 | 500 | cml_1999 | false | 1.53669 |
The 20-year value is about twice the 100-year one, because methane weighs far more over 20 years. None of the seven is wrong; they answer different questions, and an inventory uses one set throughout.
To select a sibling, replace is_default_indicator with the axis: AND f.gwp_horizon_years = 20.
Watch out. All seven rows are
unconverted: none is an AR6 number, and a total across libraries must leave them out or state the set it used. GWP explainsvalue_co2e_normalizedand when a row can be compared.
Price basis and margins: USEEIO and Open CEDA#
A spend factor means something only together with its currency year and its price basis. A purchaser price is what the buyer paid, trade and transport margins included; a producer price is what the maker received.
No factor carries two price bases as two values, so there is no sibling value to select here. Open CEDA serves both bases as two factors per sector and country: the producer price factor, and beside it a purchaser price factor (activity_detail "purchaser price") that is the producer value times the sector's purchaser to producer ratio, the publisher's own rule. That ratio adds nothing for the margins' emissions, so the purchaser factor stays cradle_to_gate and says so with margins: excluded in value_qualifiers. Match it to invoice spend. The other libraries price their spend factors one way, and you match the library to how your spend is recorded. USEEIO does publish its trade and transport margins on their own, and they are served as a boundary, not as a second price. Each commodity is a cradle_to_store_gate factor, margins included, with two parts beneath it: cradle_to_gate, the product as it leaves the factory, and distribution, the margins. All three are per dollar of purchases, so all three are purchaser.
USEEIO's cheese factor and its two parts, all at purchaser price, with the value on each.
Show the SQL
SELECT f.system_boundary, f.is_top_parent, f.price_basis, f.currency_year, f.unit_code,
round(f.value_co2e_native::numeric, 5) AS value_co2e_native
FROM open_ef.factors_flat f
WHERE f.library = 'useeio'
AND f.activity_src_id = 'useeio//311513'
AND f.is_default_indicator
AND f.applies_from_year = f.currency_year
ORDER BY f.is_top_parent DESC, f.system_boundary
| system_boundary | is_top_parent | price_basis | currency_year | unit_code | value_co2e_native |
|---|---|---|---|---|---|
| cradle_to_store_gate | true | purchaser | 2024 | usd | 1.00500 |
| cradle_to_gate | false | purchaser | 2024 | usd | 0.97000 |
| distribution | false | purchaser | 2024 | usd | 0.03500 |
These are the published 2024-dollar numbers, and the parts add up to the total: 0.97000 + 0.03500 = 1.00500. The rows for other accounting years are re-priced from them. Use the top-level cradle_to_store_gate factor with an invoice amount. Use the cradle_to_gate part only when you buy at the factory gate, so the price you paid carries no trade or transport margin. Never add a part to its total: Decomposition explains the parts. Match a spend factor to your spend year covers currency_year and the re-pricing.
Grid mix and calculation approach#
A country's electricity factor can carry one value per accounting mix. AIB's France generation factor has four, residual as the default, and so does the estimated total open-climate.ai sums over it (see above): Guarantees you can rely on lists them. Select a mix with grid_mix, and a Scope 2 method with calculation_approach: location_based for the location-based total, market_based for the market-based one. Market-based vs location-based electricity says which library publishes which mix.
Radiative forcing#
For flights, ADEME publishes figures with and without the extra warming of non-CO2 effects at altitude: 47 factors each way. They are separate factors, not siblings, so every one is a default row. Choose one with radiative_forcing = 'with' or 'without', and never add a pair. DESNZ marks 18 factors with; its other flight factors leave the column NULL, which means not stated, not without.
EPD scenarios and other qualifiers#
value_qualifiers is a small jsonb bag on the value row, and it holds the discriminators that belong to one library family rather than to every factor. The keys are a closed registry, published in the code lists, such as scenario, io_table, margins, lca_method and policy_year. The loader rejects anything else, so the bag is an axis and not a dumping ground.
On ÖKOBAUDAT the key in use is scenario: the EN 15804 end-of-life and module D scenarios. Where a dataset prints several, they become sibling values under one factor, and they are far apart. This is the C4 row of one EPD, the disposal module, where the scenario decides what happens to the product at the end of its life.
The three end-of-life scenarios of one ÖKOBAUDAT product, the default first, and how many times the lowest each one is.
Show the SQL
SELECT f.value_qualifiers ->> 'scenario' AS scenario,
f.system_boundary,
f.is_default_indicator,
round(f.value_co2e_normalized::numeric, 5) AS kg_co2e_per_m2,
round((f.value_co2e_normalized / min(f.value_co2e_normalized) OVER ())::numeric, 1) AS times_the_lowest
FROM open_ef.factors_flat f
WHERE f.library = 'okobaudat'
AND f.slug = 'ef-okobaudat-floor_and_more_hollow-europe-m2-c4-fae39373'
AND f.is_latest_activity_year
ORDER BY f.is_default_indicator DESC, 4 DESC
| scenario | system_boundary | is_default_indicator | kg_co2e_per_m2 | times_the_lowest |
|---|---|---|---|---|
| S1 | C4 | true | 0.72168 | 22.9 |
| S2 | C4 | false | 0.04980 | 1.6 |
| S3 | C4 | false | 0.03147 | 1.0 |
The elected scenario is 22.9 times the lowest of the three, for the same square metre of the same product. They are alternatives: select one, never add them.
You usually do not have to choose. The import already elected a scenario for the default row, preferring the publisher's unnamed base scenario, then one the publisher named "standard", then S0, then the only one available, then the lowest numbered. To take another, drop is_default_indicator and name the scenario instead: AND f.value_qualifiers ->> 'scenario' = 'S2'.
The literals are the publisher's own strings, not a code list. S0 to S4 sit beside Standard scenario, Standard Szenario, standard scenario, Recycled, Landfill and Deponierung. Match case-insensitively, or lean on the default flag rather than on the text.
Watch out. A NULL bag does not mean the factor has no siblings. ÖKOBAUDAT writes its unnamed base scenario as NULL, and the import prefers that row over every named one, so the elected value can carry no qualifier while its alternatives do. It is not the common case: of the 468 ÖKOBAUDAT factors with more than one value row in the latest year, 441 have every row named, and 27 mix an unnamed row with named ones, the unnamed one elected in all 27. On
ef-okobaudat-circularly_recyclable-de-m2-d-0f71bbdfthe elected module D value is -0.03689 kg CO2e per m2, with S2 at -0.40990 and S3 at -0.49350 beside it. Count a factor's value rows to test for siblings, never the bag for NULL.
All 468 of those factors are module rows under a parent, never an EPD's own top row, so a query that walks a decomposition is where you meet them: A building product: modules and module D keeps is_default_indicator on the C modules and so takes the elected scenario of each without showing the others.
Product or industry: EXIOBASE#
On EXIOBASE the key in use is io_table. EXIOBASE publishes its input-output model twice: by product, where a factor is the footprint per euro of one product, and by industry, where it is the footprint per euro of everything one industry sells. Both are served. A value read from the industry table carries io_table = industry, its activity_src_id starts with exiobase//A_ and its name ends in "industry output". A value read from the product table carries no qualifier and an activity_src_id that starts with exiobase//C_.
Two EXIOBASE activities in France, each read from the product table and from the industry table, with the value per euro on each.
Show the SQL
SELECT f.activity_full,
f.activity_src_id,
f.value_qualifiers ->> 'io_table' AS io_table,
round(f.value_co2e_native::numeric, 5) AS kg_co2e_per_eur
FROM open_ef.factors_flat f
WHERE f.library = 'exiobase'
AND f.region_code = 'FR'
AND f.activity_src_id IN ('exiobase//C_MILK', 'exiobase//A_MILK', 'exiobase//C_PULP', 'exiobase//A_PULP')
AND f.is_default_indicator
AND f.is_latest_activity_year
ORDER BY 1
| activity_full | activity_src_id | io_table | kg_co2e_per_eur |
|---|---|---|---|
| Pulp | exiobase//C_PULP | NULL | 0.25092 |
| Pulp - industry output | exiobase//A_PULP | industry | 0.28080 |
| Raw milk | exiobase//C_MILK | NULL | 2.18930 |
| Raw milk - industry output | exiobase//A_MILK | industry | 2.15964 |
The two readings are close here, 1 % apart for raw milk and 12 % for pulp, and they can be much further apart: an industry's output is everything it sells, by-products included, so its average euro is not the euro of its main product.
Unlike the scenarios above, these are separate factors and not sibling values of one factor, so every one of them is a default row: 8,064 product factors and 6,741 industry factors make up the 14,805 in the first table on this page. Nothing elects one table over the other for you. Use the product table when your spend is recorded by what was bought and the industry table when it is recorded by the supplier's sector, keep to one of them across an inventory, and never add a product row to an industry row. To keep to products, add AND f.value_qualifiers ->> 'io_table' IS NULL; to keep to industries, AND f.value_qualifiers ->> 'io_table' = 'industry'.
value_qualifiers is served on the factor detail response and shown as a chip on the factor page, and it is a column on factors_flat. It is not a filter on the API and not a factor search facet, and the Columns menu does not offer it, though an export asked for over the API can name it. Filtering to a named scenario is a SQL operation either way.
Gases and carbon origin#
The per-gas and per-carbon-origin columns are not an axis: they are components of the one headline on the same row. Read them with the NULL-versus-zero rule in How factors are cleaned, and never rebuild the headline from them.