← Back to Lore
Weather · Lab

Synthetic Lee Model

A small, deliberately narrow mountain-wave and lee-hazard sandbox. The aim is to explore how a synthetic ridge responds to different wind, stability, and geometry setups — before attempting anything closer to a real paragliding forecast.

It is not a CFD solver. It is not a calibrated turbulence forecast. The viewer and scores are diagnostic and schematic by design. The goal is pilot-interpretable hazard heuristics anchored in classic mountain-meteorology literature and validated against two well-studied lee-wave cases.

What the model computes

The raw benchmark-oriented model layer is kept separate from the paragliding display layer. Raw scores are diagnostic; displayed scores are a pilot-facing interpretation, clamped to the 0–10 m/s wind range that matters for free flight and gated to zero when the schematic lee field stays below a meaningful-signal threshold.

Validation

The model has been checked against two classic lee-wave cases:

Both harnesses also include a quiet_baseline and a marginal_baseline so the model is checked for not overtriggering on benign days and for keeping a usable middle of its dynamic range.

This doesn't make the model forecast-ready. It means the synthetic logic behaves plausibly against two classic benchmarks.

Explore

Why this matters for paragliding

Lee turbulence — rotors behind ridges when the wind blows across them — is one of the most consistent causes of paragliding incidents in mountain terrain. A pilot’s best defence is staying out of the lee in the first place, which means knowing where and when the lee is active for a given wind and stability profile.

Current public paragliding weather tools handle this crudely: they report wind speed and maybe direction, leaving the lee-hazard call to pilot judgement and local knowledge. This lab is a step toward a model that could, one day, power a lee-hazard layer inside Skyvarg — giving pilots a clear read on which sides of which ridges are likely to be rotor-active on a given day. For now: a sandbox, validated, published as research, and open for the community to kick the tyres on.