Every Last Joule · Region record

Colombia

Provenance, tier, and validation record for colombia. No live curtailment figures on this page — those are on the globe.

Region id: colombia
Country: COL
Waste modality: Hydro
Coordinates: 4.60° N, 74.10° W

Measured Data quality

Tier

live-domestic-anchored · T1b-live-domestic-anchored

The operator publishes an hourly series, but the calibration rate comes from a domestic statistical agency or a modelled share-split of a national anchor.

Live feed, domestic stat-agency rate (±50% empirical)

Uncertainty

±50% of peak GW

Published envelope for this tier. See the uncertainty model.

Source provenance

verified · Verified upstream link

The snapshot value comes from a verified upstream feed or anchor.

Source

XM SinerGox API (servapibi.xm.com.co/daily, POST MetricId=VertEner Entity=Sistema). Direct live path tried first from any Colombian-egress runner; committed CSV at data/historical/colombia-vertimientos-daily.csv is the production source-of-truth, refreshed by Britta via the elj-co WireGuard tunnel (Colombian egress). Trailing-365-day annualised total. 5-yr baseline 7.53 TWh/yr (range 0.53–13.12 TWh/yr ENSO-driven). Bimodal hydro-seasonal shape (Apr–Jun + Oct–Nov peaks). T1b, ±50% envelope.

https://servapibi.xm.com.co/daily

Tier, provenance, and source are read at build time from src/lib/regions.ts — the same table the globe draws from. The validation record below is generated separately and gated against it by npm run ci:docs-drift.

Validation record

Last updated: 2026-08-20 · Sprint: S1 + HB integration · Paper section: Technical Validation §4.2

Source

Region id: colombia
Country: COL
Tier: live-domestic-anchored
Kind: hydro
Source: XM SinerGox API (servapibi.xm.com.co/daily, POST MetricId=VertEner Entity=Sistema). Direct live path tried first from any Colombian-egress runner; committed CSV at data/historical/colombia-vertimientos-daily.csv is the production source-of-truth, refreshed by Britta via the elj-co WireGuard tunnel (Colombian egress). Trailing-365-day annualised total. 5-yr baseline 7.53 TWh/yr (range 0.53–13.12 TWh/yr ENSO-driven). Bimodal hydro-seasonal shape (Apr–Jun + Oct–Nov peaks). T1b, ±50% envelope.
Source URL: https://servapibi.xm.com.co/daily
Loader: colombia.json.ts
Structural gap: no

Calibration

Rate source documented in: docs/methodology/ (see links below)
Uniform across backfill years: n/a — no backfill

Multi-year backfill annual totals

Year	Backfill rows	Backfill annual TWh	Published TSO annual TWh	Δ %	Source
(no backfill or TSO anchors yet — will be populated after HB fan-out completes)

Published anchors

TSO annual curtailment (latest published): —
Ember annual: —
IRENA annual: —
Other: —

Discrepancy analysis

The 5-year mean of 7.53 TWh/yr is held as the T3 anchor. The natural year-on-year variance (0.53–13.12 TWh/yr) exceeds the ±40% T3 envelope, so the envelope here under-states ENSO-cycle uncertainty rather than over-stating it. The honest reading is "this region's expected annual is 7.5 TWh, but it could be 0.5 TWh in a strong El Niño year or 13 TWh in a strong La Niña year." A future T1a live loader would surface this honestly month-by-month rather than collapse it to a static mean.

Known limitations

The Colombian SIN is hydro-dominant (~60–70% of generation in normal years). Reservoir-overflow spillage (vertimientos hidráulicos) is the dominant "wasted potential renewable energy" mechanism and is conceptually identical to the spillage already modelled for Iceland (5.3 TWh/yr) and Sichuan (30 TWh/yr). VRE (solar / wind) curtailment in Colombia is currently negligible (<1%) per UPME / SER Colombia analyses — most "missing" La Guajira renewable generation is structural (transmission projects delayed) rather than operational spillage.

The XM SinerGox API is geoblocked outside Colombia, but is otherwise fully open (no auth, no rate-limit observed, structured JSON, POST body only). The loader tries the live API first; from Vercel/CI it fails and falls back to the Britta-committed CSV. When the build environment has Colombian egress (e.g. Britta with elj-co WireGuard up), the live path activates transparently. Reclassified from T1a to T1b on 2026-05-02 because the ENSO-cycle range (0.53–13.12 TWh/yr) exceeds the T1a ±15% fallback envelope and the ±50% T1b empirical band is the honest representation for a hydro-dominant grid with inter-annual ENSO swings.

The 2026-05-08 PR fixed two bugs in the loader's live path that had previously made it always fail-and-fallback (independent of geoblock): (1) the XmDailyItem interface assumed an Values: Record<string, number> field, but the actual API response uses DailyEntities: Array<{Id, Value}> with Value as a kWh string; (2) the loader requested all 365 days in a single call, but the API rejects windows >30 days with HTTP 400. The live path now uses 30-day chunks and parses DailyEntities correctly. From Britta with elj-co up, both the live API path and the relay-CSV path are now usable; production continues to read from the CSV (the relay's output) as the source-of-truth.

T3-modelled, ±40% envelope. Bimodal hydro-seasonal shape (HYDRO_SEASONAL_SHARES.colombia) lagging the rainfall peaks by reservoir-fill cycle: peak May-Jun and Nov-Dec, dry windows Jan-Feb and Jul-Aug.