Finished comparative study of meteorological models in wind prediction
-
I am finishing a comparative study of meteorological models in wind prediction. I have taken approximately 5000 data points and 300 moment-location pairs.
Here I have put the summarized conclusions, expanded, methodology, and original Excel data (in Spanish, but you know it is easy to translate in the browser).
https://parapente.freesite.online/comparativa-meteorologica/
Methodology:
Wind forecast data were collected from various meteorological models using Windy.com (and meteo-parapente.com) for different geographical points in the Basque Country (Spain). Subsequently, these data were compared with the actual data obtained from official meteorological beacons.
Analyzed meteorological models:
ECMWF 9 km
Meteoblue 4 km
ICON-EU 7 km
AROME-HD 1.3 km
Meteo-Parapente (HREM) 2.5 kmMain conclusions:
Wind direction: The AROME-HD model is the most accurate, with the lowest average deviation in degrees.
Maximum gust: Again, AROME-HD stands out for its accuracy. "The best model is clearly Arome-HD. Moreover, it has the average closest to the actual data, both in deviation above and below." The ECMWF model, on the other hand, shows the greatest deviations.
Average speed: The results are less conclusive. The HREM model from Meteo-parapente shows the greatest deviations in excess, but also the least in deficit. With the dataset on average wind speed, we could conclude that the HREM from Meteo-parapente is the winner, as it has an overall average deviation of -1.37 km/h.
Please note that I am not an expert or a company, and I may have made analysis errors. Any suggestions, criticism, or possible improvements are welcome. I started doing this for myself (paragliding), but it could surely be useful to more people.
-
@halig
Logically the model with the lowest resolution, AROME-HD 1.3km, shows the best accuracy in mountain. It would have been interesting to show what the elevation of the model is at the chosen location compared to the real elevation. This could explain some results at this location but not transposable to others. In Windy the elevation of the model is provided in the Sounding forecast, but unfortunately not available for AROME.
