Scenario register

The scenario register

Every scenario is one fixed set of conditions run through the same engine. The races form a single escalation across three steps: the mild Lavaredo 120K, then the UTMB cut-off, then the Montane Winter Spine. The ambient turns colder and the wind stronger at each step, and the Reality Gap rises as a consequence. The per-device gap bands overlap rather than separating cleanly, so read this as a trend driven by the conditions, not a clean numeric break between the races.

Reference is the baseline, not a rung

The reference scenario is the spec-sheet baseline: the fixed conditions each manufacturer implies when it prints a GPS-hours claim. It is the zero point this register measures the Reality Gap against, not a step on the ladder, so it stays prose and is not listed below. It is not the softest case either: the Lavaredo day runs warmer than the reference, so the races are not simply the reference made harsher. The register holds only the live races, where a published cut-off turns the model into a pass-or-blackout question.

The register

ScenarioConditionsCut-off (h)DevicesReality gap
Lavaredo Ultra Trail
Cortina loop
8 °C · 3 m/s3064279%
UTMB
Chamonix loop
0 °C · 6 m/s46.566286%
Montane Winter Spine
Pennine Way
-3 °C · 12 m/s16867289%

Reading the escalation

Why only these six

Each scenario covers only the six devices whose display power is traced from documented sources. The AMOLED watches added later share a single baseline display figure rather than per-device measurements, so publishing scenario numbers for them would rest on an unsourced placeholder. We leave them out until that input is sourced. The gap range above is the shared statistic across scenarios; window coverage is scenario-specific and lives on each race page. To compare all devices under the reference scenario, see the matrix.

Audit your own route to see where a specific device blacks out on your GPX, segment by segment.

DATASET · HIKINGSPECS LAB · FORMULA v1.0 · CC BY 4.0

Physics-derived estimate under one fixed scenario. The model is calibrated but not yet independently validated against measured field runtime. Estimated hardware inputs are labeled as such. Full method at /methodology.