[HughRunyan]

SWEET no longer keeps its own copy of the World Bank income groups. SWEET_python/country_income.py held What a Waste 2.0's 2018 groups, and the site-type rule fell back to it whenever a caller passed no income. Income now comes only from the caller, so every tool reads it from one place: the World Bank list the ingestion pipeline loads (RMI/waste_data_ingestion_pipeline#309).

What changed

• country_income.py and site_type_defaults.income_for are deleted. Nothing outside SWEET imported them; I checked Climate TRACE and WasteMAP.
• landfill_is_default / default_site_type: the high-income branch uses only the income_id the caller passes. A missing income counts as not high income, so only the region and named-country rules apply.
• New arguments, so WasteMAP can pass its income on each path that guesses a type:
  • City.sdst_prepopulate(..., income_by_iso3=): a country→group mapping, looked up once the clicked point's country is known.
  • City.load_andre_params(..., income_id=): the cities table's disposal-split fallback (Singapore, Brunei and others with no split of their own).
  • City.dst_baseline_blank(..., income_id=), passed on to defaults_2019.disposal_split_for: Custom Location.
• Climate TRACE's site paths already pass their row's income_id; nothing changes there.

Does this change a modelling decision?

• Yes. site-type-fill in SWEET_python/MODEL_DECISIONS.md now says the caller supplies income.

Do published numbers change?

Only for a caller that passes no income. That caller loses the high-income branch: untyped sites in Australia, Chile and the high-income Caribbean islands default to open dump, and cities in Brunei and Singapore with no split of their own fall back to all dumpsite. WasteMAP's paired PR passes income on every path, so its numbers move only where the World Bank group itself changed since 2018 (Argentina, Guyana and others; see that PR).

Tests

pytest -m "not integration": 710 passed. tests/test_site_type_defaults.py now checks that SWEET holds no table and that the new arguments exist. tests/test_custom_location_disposal_split.py covers Brunei with and without an income.

Pairs with

• RMI/WasteMAP#925: passes income on every path. Its sweet_python.ref names this branch.
• Climate TRACE needs nothing, since it already passes income on every path.
• This branch is cut from main, so the release that ships it also ships A shared site is split by the tonnes each city reports sending it #111–A site's intake series: the average sits at its years, and a shrinking site follows its line #113.

Acceptance criteria

• SWEET holds no income table.
• Every path that guesses a site type, or falls back to a type-based split, takes income from its caller.
• Tests pass.

Definition of done

• Acceptance criteria met
• Tests pass
• MODEL_DECISIONS.md and the changelog updated
• Reviewed and merged

🤖 Generated with Claude Code

[Copilot]

🟡 Changes recommended

The public City.site_only_estimate path still omits the caller-provided income_id, causing incorrect site-type guesses and leaving the active contract violation unresolved.

1 open issue

• Pass income_id through site-only site type guessing · New

What changed in this PR

This PR removes SWEET’s embedded World Bank income-group table and makes caller-provided income the sole input for site-type and disposal-split fallbacks.

Changes:

• Deletes country_income.py and removes the internal income fallback.
• Adds income parameters to city, custom-location, and site-type inference paths.
• Updates tests, modelling documentation, and the October changelog.
• Corrects the remaining public site-type path and contradictory changelog entries before approval.

This pull request requires changes.

File	Summary
tests/​test_site_type_defaults.py	Verifies removal of the embedded income table and the updated income arguments.
tests/​test_custom_location_disposal_split.py	Covers custom-location disposal splits with and without caller-provided income.
SWEET_python/​site_type_defaults.py	Uses caller-provided income for high-income rules. Moderate issue (2 votes): City.site_only_estimate still calls default_site_type without income.
SWEET_python/​MODEL_DECISIONS.md	Documents caller-supplied income as the site-type modelling decision.
SWEET_python/​defaults_2019.py	Propagates caller-provided income into disposal-split fallback logic.
SWEET_python/​country_income.py	Deletes SWEET’s duplicated World Bank income-group table.
SWEET_python/​city_params.py	Adds income inputs to city loading, custom locations, and SDST prepopulation paths.
changelog/​2026-10.md	Documents the API and modelling changes. Nit (1 vote): older entries still reference the deleted income module and outdated disposal-split API.

_{Review effort: Lite}

[Copilot]

🟢 Approval recommended

The review shows no remaining active issues, with the caller-owned income behavior implemented across affected paths and covered by the added disposal tests.

No open issues 🎉

1 resolved since last review

• Pass income_id through site-only site type guessing

_{Review effort: Lite}