Frequently asked questions

ParaRidge is a free flight-planning platform for paragliding: an interactive map of thousands of flying sites worldwide, hour-by-hour weather forecasts per launch, a GO / CAUTION / NO-GO flyability verdict computed for every site, and a directory of schools, clubs, gear shops and repair workshops.

Yes. The map, forecasts, spot pages and resources directory are free, no account required. A (also free) account unlocks favorites, the flight logbook and your personal safety limits.

From national federations' open data, OpenStreetMap and community contributions. Every listing keeps its source, and pilot-submitted edits go through moderation before publishing. No data is ever invented: when a detail is unknown, it is shown as unknown.

ParaRidge does not run its own weather models: we pull, through the Open-Meteo API, the output of the major national and international numerical models (Météo-France AROME and ARPEGE, the German DWD's ICON, MeteoSwiss ICON-CH, NOAA's GFS and HRRR, ECMWF's IFS, and more). For each site, that output is normalised into an hourly bulletin: mean wind and gusts, direction, precipitation, cloud cover, temperature and dew point. That hourly bulletin is what drives the tables, the windgrams and the flyability verdict.

Simple rule: the shorter the timeframe, the finer the model should be. For the next 24–48 h, use your region's high-resolution local model (AROME 1.3 km in France, ICON-D2 2.2 km over the Alps and Central Europe, ICON-CH1 1 km in Switzerland, HRRR 3 km in the U.S., HRDPS 2.5 km in Canada, MSM 5 km in Japan): that grid size is the only one that actually "sees" valleys and breezes. For day 2 to day 5, switch to a regional model (ICON-EU 7 km, ARPEGE ~5 km, ICON-CH2 2 km, GDPS 15 km) to judge the trend. Beyond 5 days, only global models reach far enough: ECMWF (~9 km, the synoptic reference), GFS (up to +10 days) or GraphCast (Google's AI model, up to +16 days) — read those as an air-mass trend, never as a site-level forecast.

The table below is generated straight from the app's model registry — it reflects exactly what the engine serves, with each model's grid resolution, the update frequency at its weather centre, and its forecast horizon. On every site, ParaRidge automatically pre-selects the best model available for that location; you can switch models at any time in the weather tab.

Each model has its own computation cycle, shown in the table above: HRRR is recomputed every hour, AROME, ICON-D2 and ICON-CH1 every 3 h, GFS every 6 h, ECMWF every 12 h. ParaRidge refreshes its cache from those runs several times a day, and fetches fresh data when you open a site whose cache has gone stale. In practice, a local model viewed during the day is rarely more than a few hours old.

For every forecast hour, the engine compares the weather bulletin against the launch's rules (accepted orientations, wind limits) and your pilot profile. Factors come in two families: blockers — any single one yields NO-GO (thunderstorm, snow or fog, rain ≥ 0.5 mm/h, mean wind or gusts above the site limit, direction clearly off-axis, site not suited to your pilot level) — and caution factors, which yield CAUTION but never block on their own (drizzle, a gust-to-mean spread above 2.6 m/s i.e. 5 kt, very light wind, direction at the edge of the window, likely foehn, local restriction). With no blocker and no caution factor, the verdict is GO. The factor breakdown is always shown on the spot page — the verdict is never a black box.

When a site has documented rules (federation or community), those apply: max wind, max gusts, max crosswind and the orientation window of each launch. Without documented rules, the engine falls back to conservative defaults: 12 m/s mean wind, 15 m/s gusts and 5 m/s crosswind. If you have an account, your own safety limits (set in preferences) override those caps on spot pages — the verdict then adapts to YOUR maximums, not a generic pilot's.

Yes, and it is often the deciding factor. Every launch has an orientation window (say SW to NW). If the forecast direction falls clearly outside that window, it is a NO-GO blocker; at the edge of the window, it is a caution factor. The engine also computes the crosswind component at the worst of the gust (sine of the off-axis angle × max(wind, gust)) and checks it against the site's crosswind limit.

A foehn heuristic is built in: when the air is unusually dry (temperature / dew-point spread ≥ 12 °C), the wind is established (≥ 6 m/s) and the direction is not favorable for the site, the engine adds a FOEHN_LIKELY caution factor. It is deliberately a vigilance signal and never a blocker on its own: foehn is a local phenomenon that only on-site observation can confirm.

No — and no tool should claim otherwise. The verdict is computed from forecast models, which have a 1–13 km grid and do get it wrong, especially in the mountains. It replaces neither on-site observation (windsocks, clouds, pilots in the air), nor the local briefing, nor your own judgment. Use ParaRidge to shortlist the days and sites worth the drive; the decision to launch is always made at launch.

Yes, in two places. On spot pages, your level (novice, intermediate, advanced, expert) and your personal wind limits feed the computation: a site restricted to advanced pilots will read NO-GO for a novice profile. On the map, which shows thousands of sites at once, the pins use a neutral "intermediate" profile so everyone sees the same overview — open the spot page for your personalised verdict.

It's in the making. Since August 2026, ParaRidge archives one flyability verdict per site per day — the raw material for genuine "best season" statistics per site and per region, month after month. In line with our rule of never inventing data, those charts will only appear once several months of history have actually accumulated, with the covered period clearly labelled.

Every spot page has an edit form (orientations, wind limits, access, hazards...), and the Resources page lets you suggest a missing school, club, shop or repair workshop. All submissions go through moderation before publishing, and every data point keeps its source so it stays auditable.