Methodology
How the consensus and certainty mask work
Meteo·Models does not invent a forecast. It reads the selected models, aligns them onto a shared grid, and measures how much they agree. Here is exactly what happens.
Why combine models at all?
Decades of research show that combining several independent NWP models is usually more skilful than any single model: individual biases partly cancel, and the spread between models is itself a useful estimate of uncertainty. A value that every model agrees on is far more trustworthy than one only a single run predicts.
Aligning the models
Each model publishes on its own grid and map projection. For every selected model, Meteo·Models reads the chosen variable (temperature, precipitation, wind or pressure) at the current valid time and interpolates it onto one shared grid, so the same map cell can be compared across models.
The consensus value
At each grid cell we take the mean of the models that have data there. That average is the consensus — the single best-estimate value shown when you tap a point. Cells where too few models have coverage are left blank rather than guessed.
The certainty mask
For the same cell we compute the standard deviation — the spread — across the models, and map it to a 0–100 agreement score. A low spread means the models converge, so the mask stays bright (high certainty). A high spread means they diverge, so the mask fades (low certainty). The overall consensus gauge summarises that agreement for the current view.
What this is and isn’t
This is a transparent, model-agreement view — not a proprietary black-box forecast and not a formal probabilistic ensemble. It tells you where the world’s models line up and where they don’t, which is the fastest way to gauge confidence. Always pair it with an official forecast for safety-critical decisions.
Forecast data is retrieved through Open-Meteo, which aggregates the open data feeds of the national weather services that run these models.