Capybara Geek Center · Early Warning System · Situation Report

Colombian Caribbean Flood-Risk Situation Report

Seven-day model-based exposure analysis · Sucre · Córdoba · Cesar · Magdalena · Integrated Multi-Hazard Model · Google Earth Engine

HazardFlood

Horizon7-day short-term window

ModelBayesian flood class / posterior exposure

UsePlanning, prioritization and review

Classification: Unclassified / for planning use. This report describes a modelled flood-risk exposure surface for the next seven-day operational window. It is not a statement of confirmed flooding.

Method correction: the core of this report is the code-generated seven-day risk analysis, not the news search. Recent heat and drought/fire sources appear only as an external triangulation layer for compound climatic stress. They do not validate or replace the flood model output.

1. Executive summary
2. How the seven-day model run should be read
3. Exposure analysis
4. Model output maps
5. Spatial findings by zone
6. Forecasting and Decision Objects
7. External triangulation: heat and dry-condition stress
8. Limitations, next runs and sources

1. Executive summary

This situation report interprets outputs from the Integrated Multi-Hazard Model using its flood workflow over a short-term, seven-day operational horizon. The model produces a posterior-weighted exposure surface and overlays that surface with population, cropland, urban/built-up footprint, oil palm, protected areas, roads and electric-grid layers. The five screenshots are therefore the primary evidence objects.

The strongest immediate priorities are Sucre and Córdoba. Sucre has the highest proportional exposure across the panel, with more than 70% exposure in population, cropland, roads and electric grid. Córdoba has the largest exposed population count in the screenshot set — approximately 603,063 people — and very high road and electric-grid exposure.

Main operational reading: this is a seven-day flood-risk prioritization product. It should guide where to perform deeper municipal review, not be read as a definitive damage assessment.

Recommended order for detailed review: 1) Sucre / Morrosquillo–Mojana lowlands; 2) Córdoba / Sinú basin; 3) Cesar valley–Perijá corridor; 4) Magdalena / Sierra Nevada–lower Magdalena corridor.

2. How the seven-day model run should be read

The attached code is a single Google Earth Engine script with a top-level hazard selector. The relevant path here is Flood, with the model class set to either a fast simple flood-hazard index or a richer Bayesian flood class. The shared interface controls country/AOI, admin sub-region, run mode, country-size profile, forecast horizon and ENSO baseline.

For the seven-day operational reading, the important setting is the short-term forecast window. In the WeatherNext 2 branch, the code uses a lead-time window in days and the default slider is seven days. Internally, the script translates that lead time into forecast hours and filters the collection up to that maximum. The output should therefore be presented as risk expected within the modelled short-term seven-day window.

The table values shown below are not news-derived figures. They are posterior-weighted expected exposure values read directly from the model panels visible in the screenshots. News and official bulletins only enter later as external context under the JANUS claim–evidence discipline.

3. Exposure analysis

Posterior-weighted expected exposure values were transcribed from the visible model panels. Percentages are the safest headline values; absolute values are included where the model panel reports them and should be checked against the exported model table before formal publication.

AOI / likely department focus	Population exposed	Cropland exposed	Oil palm exposed	Electric grid exposed	Roads exposed	Priority
Sucre / Morrosquillo–Mojana	74.16% / 451,419 ppl	70.64% / 1,405.69 ha	69.77% / 501.43 ha	70.21% / 1,059.11 km	72.14% / 4,932.58 km	CRITICAL
Córdoba / Sinú basin	55.45% / 603,063 ppl	33.95% / 41.70 ha	62.82% / 2,304.17 ha	54.04% / 1,668.39 km	55.34% / 7,926.32 km	CRITICAL
Cesar / valley–Perijá corridor	41.11% / 284,957 ppl	55.51% / 4,182.15 ha	31.26% / 21,390.58 ha	39.90% / 835.85 km	38.40% / 4,513.49 km	HIGH
Magdalena / Sierra Nevada–lower Magdalena	38.41% / 324,746 ppl	15.33% / 457.40 ha	38.89% / 15,485.43 ha	19.90% / 355.49 km	19.90% / 2,307.59 km	HIGH

Department names are analytical labels inferred from visible geography. Before formal publication, validate each label against the original AOI polygon used in Google Earth Engine.

4. Model output maps

The maps are the core evidence for this note. They show the modelled seven-day posterior/exposure surface over terrain, drainage and settlement context. The interpretation focuses on operational consequences: who is exposed, which lifelines may fail, and which zones deserve granular municipal review.

AOI 1 — Sucre / Morrosquillo–Mojana lowlands. Model capture with the highest all-round 7-day expected exposure in the set: population 74.16%, cropland 70.64%, electric grid 70.21% and roads 72.14% inside the posterior-weighted exposure surface.

AOI 2 — Córdoba lowlands and Sinú basin. Large absolute exposure over the 7-day model window: approximately 603,063 people, 7,926.32 km of roads and 1,668.39 km of electric grid inside the expected-exposure surface.

AOI 3 — Cesar valley and Perijá foothills. The 7-day risk surface follows the Cesar valley corridor from Valledupar toward Aguachica, with notable cropland, oil palm, roads and electric-grid exposure.

AOI 3 — Cesar / lower-valley zoom. A second zoom of the Cesar AOI confirms that exposure is not limited to Valledupar; the posterior extends south along road, river and foothill corridors.

AOI 4 — Magdalena / Sierra Nevada and lower Magdalena corridor. Moderate population exposure but high oil-palm and protected-area exposure, making this a livelihood and environmental-priority zone in the 7-day risk window.

5. Spatial findings by zone

Sucre / Morrosquillo–Mojana Critical

The highest all-round model signal. Population, cropland, roads and electric grid all exceed 70% expected exposure. Priority review should focus on low-lying settlements, drainage crossings, bridge points, rural access routes and the Mojana-transition corridor visible around Sincelejo, San Onofre, Tolú, San Marcos and Majagual.

Córdoba / Sinú basin Critical

The largest exposed-population count in the set and very high lifeline exposure. The surface follows the Sinú drainage and road network around Montería, Lorica, Cereté, Tierralta, Valencia and Planeta Rica. This is the strongest access-continuity and logistics-priority AOI.

Cesar valley / Perijá foothills High

The surface is corridor-shaped, running from Valledupar southward toward Bosconia, Chiriguaná, Curumaní, Pailitas and Aguachica. The key concern is livelihood and lifeline exposure: cropland, oil palm, roads and electric grid along valley-floor and foothill drainage lines.

Magdalena / Sierra Nevada–lower Magdalena High

Population exposure is lower than in Sucre and Córdoba, but oil-palm and protected-area exposure are high. This makes the AOI important for environmental sensitivity, rural livelihoods and access routes near the Santa Marta–Fundación–Aracataca–Plato–El Banco axis.

6. Forecasting and Decision Objects — JANUS-style

JANUS is used here as a reporting discipline: each claim is linked to a model screenshot, exposure table, code feature or external source. Model-derived claims and media/official-source claims are kept separate.

FDO 01 — Sucre is the first detailed-analysis priority

Trace: Fig. 1 exposure panel: population 74.16%, cropland 70.64%, electric grid 70.21%, roads 72.14%.

Operational meaning: Most key assets visible in the panel fall inside the seven-day expected-exposure surface.

Action: Prioritize municipal review of low-lying settlements, secondary roads, bridge crossings, shelters, health-post access and drainage bottlenecks.

FDO 02 — Córdoba is the largest exposed-population signal

Trace: Fig. 2 exposure panel: 603,063 expected exposed people, 7,926.32 km of roads and 1,668.39 km of electric grid.

Operational meaning: The Sinú basin combines human exposure with high road and grid exposure.

Action: Review access continuity, alternate routing, power-distribution vulnerability and isolation scenarios in riverine and rural settlements.

FDO 03 — Cesar is a corridor and livelihood-risk AOI

Trace: Figs. 3 and 4: cropland 55.51%, oil palm 21,390.58 ha, roads 38.40%, electric grid 39.90%.

Operational meaning: The posterior follows a long valley corridor rather than one isolated urban hotspot.

Action: Monitor agricultural losses, road cuts and electricity-distribution exposure from Valledupar southward toward Aguachica.

FDO 04 — Magdalena combines human exposure with environmental sensitivity

Trace: Fig. 5: population 38.41%, oil palm 38.89%, protected areas 35.82%, built-up footprint 23.94%.

Operational meaning: Not the highest population signal, but important for protected areas, rural livelihoods and sensitive drainage corridors.

Action: Review protected-area overlap, agricultural zones and routes near the lower Magdalena and Sierra Nevada influence area.

FDO 05 — Lifeline exposure is the cross-cutting risk

Trace: Road exposure exceeds 72% in Sucre and 55% in Córdoba; electric-grid exposure exceeds 70% in Sucre and 54% in Córdoba.

Operational meaning: The most likely operational problem is not only water in settlements, but road interruption, bridge disruption, electricity-distribution problems and access constraints.

Action: Prepare route redundancy, community communication alternatives and pre-positioning logic for rural and peri-urban nodes.

FDO 06 — External heat/dry-condition evidence is compound-stress context, not model validation

Trace: IDEAM, MinAmbiente, El Heraldo, El Universal, El Pilón and Caracol sources listed below.

Operational meaning: Heat, fire and drought-risk evidence indicates that the same northern Colombian territories may face multiple climate-sensitive pressures in the same season.

Limit: This is not a flood-model output and should not be used to claim that flooding is confirmed.

7. External triangulation: heat and dry-condition stress

This section is intentionally secondary. It helps interpret vulnerability and compound stress around the modelled AOIs, but it does not replace the seven-day flood-risk analysis.

Official IDEAM's Special Communication No. 053 reports that between May 1 and May 8, 2026, alerts for vegetation fires rose sharply in Colombia and links the deterioration to higher maximum air temperatures and reduced rainfall, especially in the Caribbean region.

Regional IDEAM reported that the Caribbean concentrated roughly 65% of municipalities with active vegetation-fire alerts, including La Guajira, Atlántico, Magdalena and Sucre. MinAmbiente/IDEAM/UNGRD later reported a national alert increase from 7 to 90 between May 1 and May 14, with the Caribbean concentrating 51.1% of active alerts.

Local Local sources add detail near or around the AOIs: Barranquilla reached 36 °C and could approach 37 °C; Cartagena was forecast between 27 °C and 37 °C; Valledupar reached 38 °C, 3.8 °C above its May average; and Córdoba reported temperatures above 40 °C with fire-risk alerts.

How to use this layer: heat and dry-condition evidence should be presented as vulnerability context for preparedness, not as proof of flooding. The blog should remain a model-derived seven-day risk analysis.

8. Limitations, next runs and sources

Limitations

• Risk, not confirmed flooding: the posterior/exposure surface is a probability-weighted planning layer.
• Screenshot transcription: exposure values were read from visible panels. For publication, use the exported CSV/table from Google Earth Engine.
• AOI naming: department labels were inferred visually from map geography and should be checked against the original AOI geometries.
• Seven-day interpretation: the report should state the exact run settings used in GEE: forecast source, lead days, ENSO phase, model class and prior balance.
• External news: heat, fire and drought stories are context only. They are not model outputs.

Next recommended model runs

• Run separate department-level AOIs for Sucre, Córdoba, Cesar and Magdalena using the same seven-day settings.
• Export the posterior raster and exposure table for each AOI.
• Add municipal overlays and rank municipalities by population, roads and electric-grid exposure.
• Run a second scenario with WeatherNext 2 and one with GFS if both data sources are available, then compare stability of the priority ranking.
• Keep a short external-source annex for heat/drought/fire stress, but do not let it become the core evidence.

Source register

Type	Source	Use in this report	Link / trace
Model code	Integrated Multi-Hazard Model — wildfire + flood single script	Primary methodology. The report uses the flood runner, shared 7-day short-term forecast configuration and posterior-weighted exposure logic.	Analyst-provided Google Earth Engine code
Model captures	Analyst-provided GEE screenshots	Primary evidence for spatial interpretation and exposure-panel values.	Figures 1–5 in this report
Method discipline	JANUS Framework — Feged 2025	Used only as a reporting discipline: claim–evidence contract, traceability, adversarial review and AI-use disclosure.	source link
Official triangulation	IDEAM Special Communication No. 053	External context: high temperatures and vegetation-fire alerts in the Caribbean. Not a flood-model output.	source link
Official triangulation	MinAmbiente / IDEAM / UNGRD heat and El Niño alert	External context: increased probability of El Niño and above-normal temperatures. Not a flood-model output.	source link
Local triangulation — Atlántico	El Heraldo — Barranquilla heat	External context for the Colombian Caribbean heat-stress layer.	source link
Local triangulation — Bolívar	El Universal — Cartagena heat	External context for coastal heat stress around Cartagena and Caribbean urban areas.	source link
Local triangulation — Cesar	El Pilón — Valledupar above historic average	External context for heat stress in the Cesar AOI.	source link
Local triangulation — Córdoba	Caracol Radio — Córdoba heat and fire risk	External context for heat/fire conditions in the Córdoba AOI.	source link
International context	WMO — Prepare for El Niño	Global ENSO context used to frame uncertainty and compound stress; not part of the flood model.	source link
Humanitarian context	Anticipation Hub / IFRC — drought early action in Colombia	Shows anticipatory action for drought in Cesar and Tolima; used only as vulnerability context.	source link

AI-use disclosure: a language model assisted with structuring, editing and translating the model outputs into a blog-ready situation report. The model did not generate the Google Earth Engine outputs, invent exposure values or determine the AOI polygons. Model-derived claims are based on the provided screenshots and attached code; external claims are separated as triangulation/context.