If you haven't activated Sencrop forecasts mode, you might notice that the best model for your forecasts changes frequently.
Variability of models' performances
Different horizons
The horizon refers to the intended duration of use for a particular model.
This means that some of the models used for forecasting are intended to predict weather changes in the short term (for example, in the next few hours), while others are meant to predict changes over a longer period of time (for example, over the next several days or weeks).
Understanding the intended horizon of a particular model can help you choose the best model for your specific forecasting needs.
Different data available
Not all forecasting models provide the data you need for your daily work.
For example, the AROME2 model does not include information on wind gusts.
This means that you may need to switch to another model, even if that model is the most relevant for all other data.
💡 Therefore, by changing the forecast position, you might observe an instant switch in your forecast's best model, even if your stations are very close to each other.
Variation of the best model
Choosing the best model can vary greatly from one to another, depending on the diversity of horizons, the available data for each model, and the time at which you are viewing the forecasts.
Use case: temperature forecast
We evaluated the performance of 40 available models by comparing their predictions with actual temperature data collected over a 3-day period at one of our stations.
Here is the graph we obtained 👇
The red curve represents the best model recorded at the time of the station's data collection. It switches from one model to another almost every hour.
The best forecasts available
If you only want to see the best forecasts available for each data type without comparing models, we recommend activating Sencrop forecasts mode.