Draft

Why different apps show different forecasts

Which model an app uses, how it picks a point from the grid, and what it adds on top.

Eight reasons two apps disagree

ReasonWhat happens
Different modelOne app shows a global model at 13 km, another a regional model at 2 km, a third a blend. They start from different data and use different physics
Different runOne app updated an hour ago, the other has the previous run
Picking a point from the gridYour position lies between grid points. An app may take the nearest one, or interpolate between several. Near a coast the nearest point may be on land
CorrectionsSome apps adjust the model for known errors, for height, for distance from the coast or for typical bias of the station; others show raw output
BlendingSome average several models or pick the model that scored best recently; the result is smoother than any single one
Which windMean wind or gust? 10 m or another height? Ten-minute mean or hourly? Without these the numbers are not comparable
Rounding and timeDifferent units and rounding; times in different time zones; hourly values that mean the hour before or the hour after
PresentationColours, icons and wording for rain probability or cloud differ even when the numbers do not

What the services in the catalog do

Many of the popular tools for sailors are aggregators: they show output from public models, and some also run their own. We list them to track how soon a new run appears in each, not to rank them, for example Windy, PredictWind, Windguru, Windfinder, Meteoblue and Open-Meteo. Each decides which models to show and how. Read the app’s own description of its sources.

How to compare two apps fairly

  1. Pick the same model in both, if each allows it. If not, find out which model each is showing.
  2. Set the same time zone (UTC is simplest) and the same units.
  3. Check the same quantity: mean wind or gusts, at 10 m.
  4. Compare the same run time.
  5. Look at the same place: zoom in so that the point is not a different grid box.
  • Apps that refresh only when you open them may show data several runs old; look for a “model run” or “updated” time.
  • A single number for “wind” on a map is the mean wind from one model; the gust is usually available on a separate layer.