About scores

Every hour of forecast data is corrected from grid elevation up to a hill's real summit height, then combined into a 0–10 score per day using one of three weight profiles (Balanced, Safety, Views - see the forecast glossary). Hard safety gates are applied first and are exclusions, not penalties: a day that fails one is removed from the ranking entirely, regardless of how good it might otherwise score.

Right now the only wired gate is a maximum summit gust. A hill excluded on a given day shows the reason with its own forecast value, not just the threshold it breached, so an exclusion is always auditable rather than a bare verdict.

In rough order of how much they matter: the wind correction multiplier is an invented figure, not yet calibrated against real weather-station observations. The cloud-base estimate is a theoretical model, not observed cloud, and can over-call cloud on humid but calm nights while missing an elevated cloud deck sitting on a temperature inversion. The freezing-level estimate uses a fixed lapse rate that can invert under a winter anticyclone, making the summit warmer than the valley on exactly the days that matter most. Access points aren't modelled yet - a hill's summit conditions are shown, not the walk-in. The daylight window is a rough approximation, not a proper solar calculation. Region-based diversification (how Top Picks spreads across the country) is a coarse proxy for an actual walking route.

This tool informs a decision; it does not make one. Hard gates exist so a dangerous day can't be outweighed by a good score elsewhere, and the numbers behind a score are always shown rather than just a verdict - especially for wind, given the limits above.