WILDFIRE RISK SITUATION REPORT Cundinamarca & Bogotá D.C. (Colombia) | GFS Short-Term | Bayesian Wildfire Model | 2 July 2026
WILDFIRE RISK SITUATION REPORT
Cundinamarca & Bogotá D.C. (Colombia) | GFS Short-Term | Bayesian Wildfire Model | 2 July 2026
Special focus: the tierra caliente corridor, Sumapaz foothills, and the Bogotá periurban belt | Includes media triangulation (CAR Cundinamarca, El Tiempo, Infobae, Bogotá.gov.co, HSB Noticias, Periodismo Público)
| HORIZON | GFS short-term (≤504 h) | ENSO BASELINE | El Niño (2019, 2023, 2024) |
| MODEL CLASS | Bayes-fast (VIIRS + CHIRPS + MODIS + GFS) | RES | 500 m fine / 1 km exposure · Cundinamarca + D.C. |
1. Executive summary
The Capybara wildfire model places a broad, operationally significant 21-day fire-risk surface across Cundinamarca, concentrated in three zones: the tierra caliente corridor from Guaduas–Villeta–La Vega through Facatativá and the western Bogotá foothills; the Sumapaz–Fusagasugá–Silvania belt south of Bogotá; and the Alto Magdalena valley from Girardot–Ricaurte–Nilo down to the Tolima border near Melgar. A smaller cluster is visible around the periurban ring south of Bogotá (Soacha–Sibaté).
The model's read is corroborated by the region's own authorities. The Corporación Autónoma Regional de Cundinamarca (CAR) has activated contingency plans across 98 municipalities and named 20 municipalities of highest susceptibility to wildfires and water scarcity during the 2026–2027 El Niño window. Of the twelve Cundinamarca municipalities on the CAR list, at least ten overlap directly with the pixels flagged by the Capybara model: Girardot, Ricaurte, Nilo, Tocaima (Alto Magdalena); Nocaima, Quebradanegra, Bituima, Vianí, San Juan de Río Seco (Gualivá / Bajo Magdalena); Mosquera, Bojacá, Soacha (Sabana Occidente and periurban ring).
2. Exposure analysis
| Variable | % of total | Expected (absolute) | Total (denominator) |
|---|---|---|---|
| Population (expected exposed) | 15.44% | 1,046,286 ppl ⚠ | 6,778,193 ppl |
| Cropland area (expected exposed) | 15.85% | 509.45 ha ⚠ | 3,214.25 ha |
| Urban area | 1.99% | 1,375.45 ha | 69,099.16 ha |
| Oil palm area | 0.93% | 54.71 ha | 5,877.02 ha |
| Protected areas (expected exposed) | 15.57% | 64,657.82 ha ⚠ | 415,309.17 ha |
| Built-up footprint | 16.66% | 407,054.09 ha ⚠ | 2,444,029.89 ha |
| Electric grid length | 16.31% | 610.39 km ⚠ | 3,743.25 km |
| Roads length (expected exposed) | 17.60% | 2,925.40 km ⚠ | 16,624.83 km |
| Oil & gas sites | N/A — not present in AOI | ||
| Wind farms | N/A — not present in AOI | ||
| Solar farms | N/A — not present in AOI | ||
The pattern is coherent. Roads at 17.60 percent leads because mountain and tierra caliente transport corridors — the Vía Bogotá–Villavicencio, the Bogotá–Medellín axis through Guaduas and Honda, the Bogotá–Girardot autopista, and the network feeding the Sumapaz municipalities — thread directly through the hazard zone. Built-up footprint at 16.66 percent and the electric grid at 16.31 percent follow the same logic: this is the rural settlement and distribution infrastructure of a large department, not urban Bogotá.
Cropland at 15.85 percent and protected areas at 15.57 percent carry the loudest policy implications. Cropland exposure at that level puts smallholder subsistence plots, coffee-transition farms in the tierra fría upper margins, and the flower and horticulture belt of Sabana Occidente inside the hazard zone. Protected-area exposure at 65,000 ha covers reserves and regional parks that already absorbed the 2023–2024 páramo fire cycle.
Urban land (1.99 percent) and oil palm (0.93 percent) are almost entirely spared. Bogotá D.C. and the departmental urban centres sit adjacent to, not inside, the high-risk surface. The periurban ring — Soacha, Mosquera, Bojacá, Sibaté — is where the urban story lives, and it is exactly where the CAR has flagged illegal-burning alerts independently of the ENSO forecast.
3. Where the risk concentrates — three corridors
Corridor A · Alto Magdalena (Girardot–Ricaurte–Nilo–Tocaima–Melgar). The southwestern cluster. Hot lowland climate, extended dry season, high tourist traffic (Girardot's second-home belt), and the Bogotá–Girardot autopista as the primary escape route. This is the corridor where population, cropland, and roads exposure compound. Named by CAR as high-risk. Priority for water-tanker pre-positioning and highway-corridor vegetation management.
Corridor B · Gualivá + Bajo Magdalena (Villeta–Guaduas–La Vega–Nocaima–Quebradanegra–San Juan de Río Seco–Bituima–Vianí). The central-western cluster along the Bogotá–Medellín axis and the descent to the Magdalena valley. Mid-altitude, dry sub-Andean forest, sugar-cane and cattle. CAR-flagged. The road exposure here maps onto the primary logistics artery between Bogotá and the Caribbean coast — a disruption vector well beyond local livelihoods.
Corridor C · Sumapaz + Sabana Occidente periurban ring (Fusagasugá–Silvania–Soacha–Mosquera–Bojacá). The southern and southwestern periurban belt. This is where the fire risk touches the largest concentration of population. Soacha alone hosts more than a million people and sits inside the flagged zone. Illegal charcoal-burning has kept the municipality on standing alert. Priority for community-level ignition-behaviour campaigns and coordination with Bogotá's Cerros Orientales prevention programme.
4. Media triangulation
The model's read of Cundinamarca aligns with what the region's own institutions are already saying — and in some cases pre-dates the ENSO forecast entirely.
- CAR Cundinamarca (June 2026). Director general Alfred Ignacio Ballesteros requested the immediate activation of contingency plans across 98 municipalities in response to projected El Niño conditions for late 2026 / early 2027. Twenty municipalities across Cundinamarca and Boyacá were named as highest-risk for wildfire and water shortage.
- El Tiempo (Cundinamarca–Boyacá briefing, 2026). Same twenty-municipality list, cross-tabulated with water-stress projections. The Cundinamarca sub-list overlaps the Capybara-flagged pixels by at least ten of twelve municipalities.
- Infobae Colombia (16 June 2026). Cundinamarca municipalities on alert for wildfire and water shortage risk; framing emphasises that the alert is a pre-El-Niño preparatory measure, not a response to an active event.
- Bogotá city government (Distrital). Two active 2026 communications: "Tips to avoid forest fires in the Cerros Orientales" and "Avoid fires in Bogotá during the El Niño phenomenon 2026" — public campaign already in field.
- Periodismo Público. Ongoing coverage of illegal charcoal-burning ("quemas clandestinas") in Soacha and periurban Cundinamarca as an independent ignition-source alert channel.
- HSB Noticias. Framing at maximum-alert level: "Cundinamarca en riesgo crítico por incendios forestales y sequía devastadora."
The triangulation supports the model's exposure reading. Where Capybara adds value on top of the CAR alert is the per-asset quantification (2,925 km of roads, 1.05 M people, 65,000 ha of protected areas) and the 21-day operational window tied to the current GFS forecast, versus the CAR's season-scale advisory.
5. Who should act, and on what
Departmental — Gobernación de Cundinamarca + CAR. The high-risk corridor identification is already published; the operational overlay this report adds is the 21-day GFS window and the per-corridor asset counts. Suggested action: reinforce the June contingency plan with corridor-specific resource pre-positioning (fire brigades, water tankers) for corridors A, B, and C, in that order of population exposure.
Municipal civil-protection offices — the twelve CAR-listed and Capybara-overlapping municipalities. Girardot, Ricaurte, Nilo, Tocaima, Nocaima, Quebradanegra, Bituima, Vianí, San Juan de Río Seco, Mosquera, Bojacá, Soacha. Standing recommendation: verify firebreak maintenance on urban-rural interfaces, coordinate with regional fire brigades on staging locations, activate community ignition-behaviour campaigns.
Agricultural extension — flower belt, horticulture, and Sumapaz smallholders. Cropland at 15.85 percent exposure across a small denominator (3,214 ha mapped) suggests the affected units are concentrated smallholder plots and high-value horticulture — precisely the producers least able to absorb a total loss. Coordinate irrigation contingencies and pre-harvest protocols where feasible.
Protected-area authorities — Parques Nacionales + regional CARs. 65,000 ha of protected area inside the flagged zone means that the reserves absorbing the 2023–2024 páramo fire cycle are facing another El Niño-analog window. Activate ranger patrols on the fire-prone perimeter zones, verify firebreak maintenance, coordinate with adjacent communities.
Transport authorities — INVÍAS + concessionaires. The corridor most disproportionately exposed by asset class is roads, at 17.60 percent. This includes segments of the Bogotá–Medellín and Bogotá–Girardot axes. Coordinate with the concessionaires on vegetation-management schedules and incident-response staging along the flagged sections.
6. Method and uncertainty
The run uses the Bayesian short-term posterior: a fire-density prior derived from three years of FIRMS observations conditioned on El Niño years (2019, 2023, 2024), a landscape-similarity prior from Google's satellite embedding compared against the ENSO-matched fire archive, a dryness likelihood built from CHIRPS precipitation anomaly, MODIS surface signal, and terrain aspect, and a hazard likelihood built from the current GFS forecast (temperature, humidity, wind, out to 504 hours). All four combine into the posterior surface shown on the map. Exposure is the intersection of that surface with ten standard asset layers.
The result is not a prediction that fire will occur in a specific pixel — it is a probability field. Read the percentages as planning estimates, not as body counts. Uncertainty compounds at three points: the El Niño analog assumption (that 2019 / 2023 / 2024 patterns are the right reference for the coming weeks), the GFS forecast skill at 500 h, and the fire-density prior's dependence on the last three years' fire record, which for regions with rapid land-use change may over-weight the most recent ignition patterns.
Three exposure categories return N/A in this AOI: oil-and-gas sites, wind farms, and solar farms. Cundinamarca hosts no significant assets in those classes.
7. Sources
- CAR Cundinamarca — Planes de contingencia por riesgo de incendios forestales
- El Tiempo — Zonas de Cundinamarca en riesgo por El Niño
- El Tiempo — 20 municipios bajo la lupa por El Niño
- Infobae Colombia — Municipios en alerta por El Niño
- Bogotá.gov.co — Tips para evitar incendios en Cerros Orientales
- Bogotá.gov.co — Evita incendios por El Niño 2026
- Periodismo Público — Quemas clandestinas en Soacha
- El Nuevo Siglo — CAR pide planes de contingencia
- HSB Noticias — Alerta máxima Cundinamarca
- Madera y Bosques (2024) — Modelación Dinámica de incendios forestales, Cerros Orientales de Bogotá
Method reference. JANUS analytic framework for AI-assisted disaster-risk analysis: Feged (2025), DOI 10.21203/rs.3.rs-7580516/v1. This report was produced under JANUS with AI-assisted drafting and adversarial media triangulation. Quantitative claims are traceable to the model run of 2 July 2026; interpretive claims are the analyst's read.
Contact. Alejandro Feged-Rivadeneira · alejandrofeged@gmail.com · Capybara Center
Note: Only a member of this blog may post a comment.