FLOOD RISK SITUATION REPORT Peru (national) | GFS Short-Term (≤168 h) | Bayesian Flood Model | 2 July 2026
FLOOD RISK SITUATION REPORT
Peru (national) | GFS Short-Term (≤168 h) | Bayesian Flood Model | 2 July 2026
Special focus: the northern coast, the sierra central-and-south corridor, and the infrastructure spine | Includes media triangulation (SENAMHI, CENEPRED, La República, Infobae Perú, Diario Correo, RPP, Gestión)
| HORIZON | GFS short-term (≤168 h · 7 days) | ENSO BASELINE | El Niño (2019, 2023, 2024) |
| MODEL CLASS | Bayesian flood (balanced priors) | RES | 500 m fine / 1 km exposure · national AOI |
1. Executive summary
The Capybara Bayesian flood model places approximately 2.03 million Peruvians (9.76 percent of the 20.77 million mapped) inside the top-decile flood hazard band for the coming seven-day GFS window. The infrastructure exposure is disproportionately higher — 23.44 percent of the country's roads (42,036 km) and 18.47 percent of the electric grid (9,350 km) sit inside the flagged zones. Cropland exposure is the single most concerning asset-class figure at 24 percent (295,112 ha of 1.23 M ha) — nearly one hectare of Peruvian cropland in four falls inside the model's high-risk surface.
The model's read of Peru aligns with what the country's own institutions are already saying. CENEPRED estimates 9.3 million Peruvians live in very-high-risk zones for floods and landslides. SENAMHI has maintained a red alert on the northern coast. The 2 July 2026 emergency decree closes the loop — the executive branch has committed 60 days of extraordinary powers to the same geography the model flags.
2. Exposure analysis
| Variable | % of total | Expected (absolute) | Total (denominator) |
|---|---|---|---|
| Population (expected exposed) | 9.76% | 2,026,895 ppl ⚠ | 20,768,792 ppl |
| Cropland area (expected exposed) | 24.00% | 295,112 ha ⚠ | 1,229,808 ha |
| Urban area | 7.10% | 33,826 ha ⚠ | 476,299 ha |
| Oil palm area | 1.36% | 877 ha | 64,619 ha |
| Protected areas | 6.09% | 1,517,020 ha ⚠ | 24,894,552 ha |
| Built-up footprint | 10.07% | 12,993,963 ha ⚠ | 128,974,474 ha |
| Electric grid length | 18.47% | 9,350 km ⚠ | 50,618 km |
| Roads length (expected exposed) | 23.44% | 42,036 km ⚠ | 179,307 km |
| Oil & gas sites | N/A — sparse-point coverage | ||
| Wind farms | 0.01% | 0 of 2 sites | 2 sites |
| Solar farms | 0.01% | 0 of 4 sites | 4 sites |
The infrastructure story dominates. Roads at 23.44 percent (42,036 km) and the electric grid at 18.47 percent (9,350 km) put Peru's core logistics and power spine directly inside the seven-day flood window. That is the story El Niño 1997–98 and the 2017 Niño costero both told: the Panamericana Norte cut, Piura and Chiclayo without power, Lambayeque's irrigation channels overwhelmed. The model's read is that a similar signature is materialising again.
Cropland at 24 percent is a food-security figure. Peru's exportable agriculture — asparagus, blueberries, avocados, table grapes — concentrates on the northern and central coast, exactly where the flood surface is highest. Smallholder subsistence maize and potato in the sierra sit inside the model's Andean corridor. The combined effect of a Panamericana cut plus a quarter of national cropland under water risk is a compound supply-chain event, not a local flood.
Protected areas at 6.09 percent = 1.52 M ha covers ecosystems that already absorbed the 2023 Yaku cyclone rains. Reserva Nacional Pacaya-Samiria (Amazonian), Manu (Madre de Dios), and coastal reserves are inside the flagged corridor.
3. Where the risk concentrates — four corridors
Corridor A · Northern coast (Piura–Lambayeque–La Libertad–Áncash). The classical Niño-costero corridor. Piura alone has 65 districts already declared in emergency by regional decree. Cajamarca and the northern sierra sit at the drainage head. The most concentrated pixel cluster on the map. Priority: Panamericana Norte, Piura and Chiclayo urban drainage, Lambayeque irrigation channels, and the Chira–Piura reservoir system.
Corridor B · Central sierra (Áncash–Huánuco–Pasco–Junín–Huancavelica–Ayacucho). Andean valleys, landslide-prone slopes, and the Marañón and Mantaro river systems. Population density is lower per km² than the coast, but road exposure here concentrates on the mountain corridors — the Carretera Central and its lateral branches — that connect Lima to the rest of the country. Any Carretera Central closure is a Lima-supply event.
Corridor C · Southern highlands and altiplano (Apurímac–Cusco–Puno–Arequipa interior). Southern flood signal from the Amazon-facing basin drainage. The pixel cluster around Ayacucho–Abancay–Cusco is prominent on the map. The 2023–2024 El Niño years left this corridor with saturated soils; the current forecast interacts with that memory.
Corridor D · Southern coast (Ica–Arequipa coastal–Moquegua–Tacna). Less rainfall than the north but the desert-fed flash-flood signature (dry quebradas filling suddenly during heavy events) sits over urbanised areas — Ica in particular. Historic pattern: relatively small rain totals producing outsized flood impacts.
4. Media triangulation
- Decreto Supremo N.º 097-2026-PCM (2 July 2026). Executive declares 796 districts across 22 regions plus Callao in state of emergency for 60 days. The named regions are a near-exact overlay of the model's flagged pixel clusters. This is the strongest possible external validation: the model and the executive branch published within hours of each other pointing to the same geography.
- Infobae Perú (2 July 2026). "Más de 9 millones de peruanos en riesgo muy alto por inundaciones y deslizamientos ante el fenómeno El Niño." CENEPRED estimate of 9.3 million in very-high-risk zones. Capybara's 2 M is a 7-day acute subset of that chronic figure.
- La República (2 July 2026). Detailed coverage of the decree, jurisdiction list, and the operational implications for the 60-day emergency window.
- Diario Correo. Regional context on the affected departments; frames the decree as a preparatory move ahead of the intense-rain window.
- RPP. "Gobierno declara estado de emergencia en 22 regiones ante lluvias previstas por Fenómeno El Niño." Broadcast framing at national scale.
- Infobae Perú (March 2026). SENAMHI investigator: rain period could extend to July or August — matching the current model's short-term window.
- Gestión. Business-press coverage; explicit tie to the Fenómeno El Niño projection through austral spring 2026.
- SENAMHI (ongoing). Red alert on the northern coast; national aviso meteorológico portal actively publishing.
The triangulation is unusually strong. On the same day the Bayesian flood surface identifies these corridors, the national government has declared them in emergency. Where Capybara adds value on top of the decree is the per-asset quantification at 500-metre resolution — the 42,036 km of roads, the 9,350 km of grid, the 295,112 ha of cropland — and the seven-day operational window tied to the current GFS forecast, versus the decree's 60-day preparatory horizon.
5. Who should act, and on what
National — Presidencia del Consejo de Ministros + INDECI. The emergency decree is in force. The Capybara exposure figures translate into resource-allocation deltas: prioritise pre-positioning of pump equipment, mobile generators, and highway-repair capacity along Corridor A first, then B (Carretera Central), then C, then D. The 2 M acute-exposure population is the operational subset of the 9.3 M CENEPRED chronic-exposure figure.
Regional — the 22 named regional governments. Each named region should identify its top-decile districts and issue evacuation-readiness guidance for the seven-day window. Piura, Lambayeque, La Libertad, and Áncash carry the heaviest coastal load; Cusco, Puno, and Apurímac the sierra load.
Agricultural extension — MIDAGRI, SENASA, regional agricultural directorates. Cropland at 24 percent means a quarter of the country's mapped agricultural surface is in the seven-day window. Priority interventions: irrigation-channel monitoring in the north, harvest-scheduling contingencies for exportable crops (asparagus, blueberries), smallholder communication in sierra corridors, and post-event damage-assessment protocols ready to deploy.
Transport authorities — MTC + concessionaires. The road exposure of 42,036 km is Peru's largest infrastructure figure in the run. Prioritise pre-positioning of debris-removal equipment along the Panamericana Norte, the Carretera Central, and the Interoceánica Sur — the three arteries where a closure has the largest downstream effect.
Electricity operators — COES + regional distributors. 9,350 km of grid in the top-decile hazard band. Pre-positioning of pole and transformer replacement stock is warranted in the Corridor A and B geographies.
Protected-area authorities — SERNANP. 1.52 M ha inside the flagged zone. Pacaya-Samiria, Manu, and coastal reserves should activate flood-response protocols, including patrol repositioning and communication with adjacent communities.
6. Method and uncertainty
The run uses the Capybara Bayesian flood posterior with balanced priors: an inundation-frequency prior from JRC Global Surface Water yearly history conditioned on El Niño years (2019, 2023, 2024), a landscape-similarity prior from Google's satellite embedding, a saturation likelihood built from CHIRPS antecedent precipitation and SMAP surface soil moisture, and a hazard likelihood built from the current GFS forecast precipitation out to 168 hours. All four combine into the posterior surface shown on the map. Exposure is the intersection of that surface with ten standard asset layers at 1-km grid.
The result is a probability field, not a prediction of flood at a specific pixel. Read the percentages as planning estimates. Uncertainty concentrates at three points: the El Niño analog assumption (that 2019 / 2023 / 2024 patterns generalise to the coming week), GFS forecast skill at 168 h (better than at 500 h but not perfect), and the JRC prior's dependence on historical inundation frequency, which may under-represent recently-urbanised drainage basins where new flood risk has emerged since 2020.
Three exposure categories return near-zero or N/A: oil-and-gas sites (sparse point coverage), wind farms, and solar farms. Peru's renewable-generation buildout has not yet produced large mapped assets inside the flagged corridors.
7. Sources
- Diario Oficial El Peruano — Decreto Supremo N.º 097-2026-PCM (2 July 2026)
- La República — Gobierno declara en emergencia a casi 800 distritos por El Niño
- Infobae Perú — Más de 9 millones de peruanos en riesgo muy alto (CENEPRED)
- Infobae Perú — Estado de emergencia en 796 distritos
- Diario Correo — Emergencia en 796 distritos por El Niño
- RPP — Emergencia en 22 regiones
- Gestión — SENAMHI alerta roja por lluvias y huaicos
- Infobae Perú (Enero 2026) — SENAMHI alerta de lluvias en Lima y 15 regiones
- Infobae Perú (Marzo 2026) — Periodo de lluvias podría extenderse a julio–agosto
- El Tiempo (Piura) — 65 distritos de Piura en emergencia por El Niño
- SENAMHI — Aviso Meteorológico (portal nacional)
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.