Envitel builds probabilistic wildfire hazard models for the insurance and reinsurance market. EnviFIRE v2 runs 10,000 synthetic years of hour-by-hour fire weather, ignitions and fire growth across Spain and Portugal, and turns them into burn probability and flame-length classes for every 100-metre cell.
The problem
European wildfire losses have grown beyond the tools available to quantify them, and the Iberian Peninsula carries the continent's most active fire regime. In 2025 the EU recorded its worst fire season since harmonised records began, and Spain its worst in more than three decades.
The catastrophe models the market prices against were built for North America. That gap is why climate-risk disclosure and underwriting increasingly demand a defensible, European-first view of this hazard.
burned across the EU in 2025, its worst season on record: the size of Cyprus
insured loss from the January 2025 Los Angeles fires, the costliest on record
of European natural-catastrophe losses go uninsured
Sources: EFFIS (Copernicus / JRC) · Swiss Re · EIOPA
This is not a distant risk. It is the landscape we model, a few kilometres from where we work.
Cenicientos, Madrid region, 2019. Photo: Canizipi (CC BY-SA 4.0)
What we do
Most wildfire tools extrapolate from the footprints of past fires. Envitel simulates the fires themselves, thousands of years of them, and reads the risk from what they do.
Fire growth is solved with a physical spread model over real terrain, fuels and hourly wind, with crown fire and ember spotting. A result reflects the slope and vegetation of a place, not a regional average.
10,000 synthetic years of hourly fire weather, grounded in the 2002–2025 record, drive ignition, fire duration and spread, and accumulate into burn probability for every cell.
Every layer is produced for the 2002–2025 baseline and for twelve climate scenarios, four emissions pathways at three horizons, so a portfolio can be read under each.
The product
Hazard, delivered the way the market works with it. EnviFIRE v2 covers Spain and Portugal today, and the rest of Europe before the end of the year, on the same methodology.
Annual burn probability, return periods and flame-length classes for every 100-metre cell, portfolio-ready on the standard European grid as GeoTIFF rasters.
Locations or portfolios in, burn probabilities and return periods out, as CSV or through the API. Single points or whole books.
Integrated against an insured portfolio downstream, the layers yield the exceedance-probability view underwriting and capital modelling need.
Burn probabilities at return periods are available today, for any point in the Iberian Peninsula.
Ask for a location checkFour emissions pathways at three horizons to 2100, next to the 2002–2025 baseline.
Each cell's fires split across six flame-length classes, so the layers compare directly with the products the market already uses.
Wind and fuel moisture update every hour as a fire burns.
Every simulated fire is delivered whole, with its arrival time, flame length and intensity at each cell it burns.
Every stage of the pipeline is published in a live methodology platform, with its inputs, modelling choices, validation evidence and limitations. We walk partners through it under a simple NDA.
10,000 synthetic years of hour-by-hour fire weather, resampled from the observed record and shifted by climate projections.
Where and how often large fires start, learned from the landscape and the fire record.
How dry dead and live vegetation is, cell by cell and hour by hour.
Physical fire growth over real terrain, fuels and hourly wind, at 100 m.
How long each fire burns before it is contained, fitted to the observed fire record.
Millions of simulated fires become burn probability, return periods and flame-length layers.
The map checked against the fire record, and simulated fires compared with observed ones.
EnviFIRE v2 is validated against the observed Iberian fire record, fire by fire and climate zone by climate zone, against checks fixed in advance, and every result can be regenerated and audited. The full specification, with its layers, event data, climate scenarios and validation evidence, lives in the Product Overview and the methodology platform.
Europe's wildfire risk is still priced with models built for another continent. The Iberian Peninsula carries the most active fire regime in Europe, and European portfolios deserve a model calibrated to European fuels, weather and fire behaviour, not adapted after the fact.
2025 was the EU's worst fire season on record, extreme fires are projected to rise by around 30% globally by 2050, and climate-risk disclosure is hardening from good practice into obligation. The market needs a defensible European view of this peril now, not next decade.
Envitel was built for exactly this question: physics at ground level, validation run climate zone by climate zone, documentation end to end, and the pace of a company engineered AI-first with the checks built in. From Madrid, inside the geography we model.
The company
Envitel was founded to build one thing well: a wildfire catastrophe model the European insurance market can trust. EnviFIRE v2, released in October 2026, is that model. We work with reinsurers, brokers, MGAs and insurers with European exposure, and the same numbers serve anyone who carries wildfire risk on their books or their land.
CEO & Co-founder
CTO & Co-founder
Founding Advisor
Working with Envitel
We deliver burn probabilities at return periods for any point in the Iberian Peninsula, for the baseline and for every climate scenario. Ask for the product overview, or tell us about your portfolio.
contact@envitel.aiEnvitel Inteligencia Ambiental, S.L. · Madrid, Spain