Capybara Center for Risk Analysis · essay · under JANUS

The local grammar of fire

We talk about wildfire risk as if it were one thing rising everywhere. Watch four places climb toward their fire season and you see something else: risk speaks a different dialect in every landscape.

Capybara Center for Risk Analysis · July 2026

There is a comforting fiction in the phrase "fire risk is rising." It suggests a single dial, turning up across the whole planet, that we might all watch together. The truth is stranger and more useful. When you take the same model and point it at four different corners of the Earth, then watch how modeled risk climbs month by month toward the moment fire actually erupts, the curves refuse to rhyme. Fire does not have one face. It has a local grammar, and each place speaks it differently.

This is not a story about which place is most dangerous. It is a story about shape: the particular way risk gathers, crests, and gives way to flame in the mountains above a Portuguese river, on the Colorado plateau, along the Andes of central Colombia, and in the Alpine hinterland of Provence. Read those shapes side by side and a practical lesson falls out, one that any single global rule would get wrong three times out of four.

The question is not "is risk rising" but "how does risk, in this place, hand off to fire, and how much warning does that handoff give."

A word of honesty before the shapes, because it changes how you should read them. These four are snapshots from a living model, not a controlled experiment. They were run on slightly different versions, with different climate baselines in places, and one of them, Bogota, on a different footprint entirely. So the exact heights of the curves are not strictly comparable between countries, and I will not pretend otherwise. What survives that messiness is the thing worth having: the form of each region's approach to its fire season, which is robust to the run details because it is written into the landscape and the climate, not into the settings.

---

The first faceThe Mediterranean spike

Viseu, Portugal · near-synchronous

Dormant, then all at once

For most of the year the risk curve over the hills of Viseu lies almost flat against the floor. Then, sometime in July, it does not so much rise as launch, climbing steeply to a tight peak in August and September before collapsing just as fast. And the observed fire? It arrives almost on top of that peak. Risk and flame rise together, within a few weeks of each other.

This is the Mediterranean signature, and Provence, on the far side of the same sea, is built to the same template. It is a regime of long dormancy and sudden violence. The practical consequence is double-edged. The warning window is short, because the whole season is short and explosive. But the signal is clean: when the Portuguese risk line turns upward in July, you do not have to wonder whether fire will follow. It follows almost immediately. There is little ambiguity and little time, which is its own kind of clarity.

In the Mediterranean, fire keeps its appointment. The danger is not that it surprises you, but that it gives you so little room once it arrives.

The second faceThe Western ambush

Colorado, USA · fire leads the peak

The fire that jumps the gun

Colorado breaks the intuition you just built. Here the risk curve climbs from a spring low toward a peak in autumn, a slow, reasonable ascent. But the fire does not wait for it. The first eruption comes in June, while the risk curve is only halfway up the hill, and a second pulse arrives later near the September crest. The biggest fire month is not the peak-risk month.

Why would fire arrive before peak dryness? Because dryness is only half the story. The other half is ignition, and the American West is generous with it: dry lightning marching across the high country, and a long human fuse of sparks and engines and power lines. Once the fuels cross a flammability threshold, which happens well before the seasonal maximum, all that is missing is a match, and the West supplies matches early. As I write, the model's high-risk band over southern Colorado is lit with active fires, and the national system sits at its second-highest readiness level. The landscape did not wait for its own risk peak, and neither did the flames.

This is the operationally treacherous regime, the one that punishes patience. A warning system tuned to fire when risk is highest would sleep straight through the June ambush. In the West, the actionable moment is the early-summer rise, not the late-summer summit.

The third faceThe Andean metronome

Cundinamarca, Colombia · dry-season locked

Fire that keeps time with the calendar

Above Bogota the curve does something neither of the others do: it peaks twice, once around February and again around September, tracing the two dry seasons that the tropical Andes fold into a single year. Fire follows the February peak. And right now, in the wet trough of July, the model is quiet and the satellites agree: no active fire in the last seven days. The metronome is between beats.

Here the useful question changes shape entirely. It is not "is risk rising toward a summer," because there is no single summer. It is "which dry season are we approaching, and is this year's peak running hotter or cooler than the years before it." That is a question about deviation from a rhythm, not about a climb toward a peak, and it asks for a different kind of watching. Bogota's run is the one that already carries a current-year line drawn against the baseline, so it is the one place here where you can see this season's beat laid directly over the memory of past ones.

---

What the three faces teachLead time is the fingerprint

Line the three up and the real finding is not any single curve. It is that the gap between risk rising and fire erupting is itself the identifying mark of a place:

Near-synchronous in the Mediterranean, where risk and fire crest together and the warning is clean but brief. Fire-leading in the American West, where flame jumps ahead of peak risk and rewards those who act on the early rise. Dry-season-locked in the tropical Andes, where fire keeps calendar time and the question is which season, and how this year compares.

A single global rule, "fire follows peak risk," would be right in Portugal and wrong in Colorado, wrong in Cundinamarca, and misleading in Provence. That is the whole argument for watching each place in its own grammar rather than under one planetary dial. A model earns its keep here not by producing a number but by exposing a shape, and the shapes are different because the world is.

Risk is not a single rising tide. It is a set of local dialects, and fluency means learning each landscape's particular way of handing the season over to fire.

The face that changesWatching the grammar shift

There is a second sense in which the face of risk is changing, and it is the one that should keep us honest. These regimes are not fixed monuments. Fire seasons across the Mediterranean and the American West have been lengthening and sharpening for decades, and the tropics are seeing their dry-season rhythms bent by shifting rainfall. The curves in this essay are this decade's shapes, not eternal ones. Which is exactly why the interesting instrument is not a single snapshot but the current-year line laid over the baseline, run after run, season after season, so that when a region's grammar starts to change, you see it happen rather than infer it later.

That is the work ahead: to keep pointing the same model at the same places on a steady cadence, freezing and dating each reading, so the record can show not only that Colorado ambushes and Portugal spikes and the Andes keep time, but whether, and how fast, those faces are being redrawn. The changing face of risk is not a metaphor. It is a measurement we can choose to take.

A note on the instrument. The curves here come from a seasonal diagnostic that pairs a fixed structural susceptibility with monthly climate dryness and reads it against observed satellite burned area. It is a climatological lens, not the operational forecast, and it describes how susceptibility and fire co-vary through the year rather than predicting a specific ignition. Observed burned-area data lags by roughly a year, so the current-season fire line is short until it lands. And, as noted above, the four runs are not perfectly cross-comparable in magnitude; the argument rests on the shape of each region's approach, which is what the landscape and climate write, not the run settings.

Capybara Center for Risk Analysis · Tecnologico de Monterrey · under the JANUS framework (Feged-Rivadeneira, 2025; archived release doi.org/10.5281/zenodo.21170613).
A climatological reading, not an operational forecast. Regional magnitudes are not strictly cross-comparable; the argument is about the shape of each region's progression from risk to fire.