meteomat / docs /methodology.json
GitHub Actions
Deploy from GitHub Actions
c83e52f
Raw
History Blame Contribute Delete
2.26 kB
{
"description": "Meteomat shows probabilistic weather forecasts anywhere in the world. Instead of a single number, it shows a range of likely outcomes so you can see not just what is most likely, but how confident the forecast is.",
"sources": [
{
"product": "ECMWF IFS Ensemble (EPS)",
"resolution_km": 18,
"members": 51,
"horizon_days": 7,
"type": "weather"
},
{
"product": "ECMWF IFS HRES",
"resolution_km": 9,
"members": 1,
"horizon_days": 7,
"type": "weather"
},
{
"product": "ECMWF Wave Model",
"resolution_km": 28,
"members": null,
"horizon_days": 7,
"type": "marine"
}
],
"parameters": {
"uncertainty_band_pct": {
"low": 10,
"high": 90
},
"rain_threshold_mm": 0.1,
"rain_blend_hres_rain": {
"ensemble": 0.5,
"hres": 0.5
},
"rain_blend_hres_no_rain": {
"ensemble": 0.75,
"hres": 0.25
}
},
"limitations": [
{
"id": "spatial_resolution",
"title": "Spatial resolution",
"body": "HRES at ~9 km and the ensemble at ~18 km cannot resolve sub-grid features such as valley-scale orography, urban heat islands, or small coastal inlets. Two nearby locations within the same grid cell will receive nearly identical forecasts."
},
{
"id": "ensemble_calibration",
"title": "Ensemble calibration",
"body": "The ECMWF ensemble is one of the best-calibrated probabilistic forecast systems in the world, but calibration is not perfect. Stated probabilities are skill estimates, not guarantees."
},
{
"id": "rainfall_uncertainty",
"title": "Rainfall uncertainty",
"body": "Precipitation is the hardest variable to forecast deterministically. The blending methodology is a pragmatic combination of two imperfect sources and should be interpreted with appropriate caution."
},
{
"id": "sst_upwelling",
"title": "SST and upwelling",
"body": "The marine SST forecast does not account for wind-driven upwelling events, which are common along the Cantabrian coast of Spain and can cause surface temperatures significantly colder than the model suggests."
}
],
"version": "2.0.1"
}