You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Turns crop provisioning from a revenue proxy into a measurable service. The catalog has nci-frontiers crop revenue densities (USD/ha, CC0 from Polasky et al. 2026). That is not the same quantity SEEA needs.
Why it matters
SEEA defines crop provisioning as "the ecosystem contributions to the growth of cultivated plants" — not tonnage, and not farm-gate revenue. Booking revenue directly overstates the account by whatever share is attributable to labour, capital and purchased inputs.
The standard method (used by ARIES for SEEA itself, per UNSD Guidelines on Biophysical Modelling §6.4.1) is:
SPAM gridded production, adjusted year-on-year by FAOSTAT national production,
then the ecosystem contribution via Vallecillo et al. (2019) — the ratio of natural to natural+human inputs in energetic terms.
Step 1 needs both datasets; nothing in the catalog supplies either. Also serves #449Evaluate (agricultural land-use dependency) and pairs with GLW 4 (#447) for the livestock/grazed-biomass side.
Sources
SPAM 2020 (Spatial Production Allocation Model) — IFPRI, ~10 km, 46 crops / crop groups, physical + harvested area, production, yield, by 4 technologies (irrigated / rainfed high-input / low-input / subsistence). https://www.mapspam.info/data/. License CC-BY-4.0. (SPAM 2010 also exists; 2020 is current. A 2010→2020 pair would give a change signal.)
FAOSTAT — national production (QCL) and producer prices (PP), annual, long series. https://www.fao.org/faostat/. License CC-BY-4.0.
Scope
SPAM: raster → H3 native 8, parents 0. ⚠️Reducer depends on the variable: production (tonnes) and area (ha) are amounts → sum; yield (t/ha) is a density → mean. Ingest production + harvested area as the primary variables.
FAOSTAT:not spatial — a country × item × year × element lookup table, published as flat parquet with an ISO country key joinable to overture-divisions-countries. No hex.
SPAM production + harvested area per crop, COG + hex, with the per-variable reducer documented (amount vs density).
Crop codes documented and joinable to FAOSTAT item codes.
FAOSTAT tables as parquet with ISO keys; table:columns populated.
National SPAM production totals reconcile against FAOSTAT for the reference year (this is the meaningful validation).
verify-stac.py exits 0; licenses recorded in license-inventory.md.
Note
FAOSTAT is a non-spatial tabular import — if data-workflows is the wrong home for that artefact class, say so and unseea will carry it in-repo instead (cf. SchmidtDSE/unseea#16, which raises the same question for EXIOBASE/ESVD/ENCORE).
Requested by SchmidtDSE/unseea — SchmidtDSE/unseea#8, SchmidtDSE/unseea#9.
Turns crop provisioning from a revenue proxy into a measurable service. The catalog has
nci-frontierscrop revenue densities (USD/ha, CC0 from Polasky et al. 2026). That is not the same quantity SEEA needs.Why it matters
SEEA defines crop provisioning as "the ecosystem contributions to the growth of cultivated plants" — not tonnage, and not farm-gate revenue. Booking revenue directly overstates the account by whatever share is attributable to labour, capital and purchased inputs.
The standard method (used by ARIES for SEEA itself, per UNSD Guidelines on Biophysical Modelling §6.4.1) is:
Step 1 needs both datasets; nothing in the catalog supplies either. Also serves #449 Evaluate (agricultural land-use dependency) and pairs with GLW 4 (#447) for the livestock/grazed-biomass side.
Sources
Scope
0.sum; yield (t/ha) is a density →mean. Ingest production + harvested area as the primary variables.overture-divisions-countries. No hex.public-agriculture(register MinIO scope per Epic: Global biodiversity / nature-positive (TNFD / SBTN) catalog gaps #449 Step 7), or fold SPAM into an existing global bucket.Acceptance criteria
table:columnspopulated.verify-stac.pyexits 0; licenses recorded inlicense-inventory.md.Note
FAOSTAT is a non-spatial tabular import — if
data-workflowsis the wrong home for that artefact class, say so and unseea will carry it in-repo instead (cf. SchmidtDSE/unseea#16, which raises the same question for EXIOBASE/ESVD/ENCORE).