Written by Tyler Anderson, mathematician and R&D Data Scientist at Nebbo.
Seasonal models help us predict the average weather conditions over the coming months. This is somewhat different from the usual weather forecasts that one sees on the news, which try to predict the weather conditions for each day in the coming week or so.
One may wonder why we can’t simply run the daily models longer to predict more than just their usual duration. This could then give us a manner of getting daily predictions for a date several months ahead. The issue with this, and why there is also a limit to the seasonal models as well, is that as the model is run further into the future, the model’s predictions get worse and worse.
The chaotic nature of the weather systems is simply too great to try and predict daily weather patterns beyond more than just a few days, which is why we cannot use these daily models to predict weather patterns on the seasonal scale. We need a different type of model to do this.
The seasonal models that are used then cannot predict what the weather will be on a specific day three months in the future, but there is still some information that we can hope to get from these models so far in the future. It turns out that although we cannot predict daily events, we can predict what the weather conditions will be on average in the coming months. This information is largely due to one key system on Earth: sea surface temperature.
What Makes Seasonal Forecasting Possible
Seasonal forecasting is possible largely because of the slow fluctuations in the sea surface temperature that influence long-term weather patterns. Luckily, these long-term weather patterns are exactly what we are trying to predict when we are using these seasonal models.
These seasonal models use this connection between the sea surface temperature and the atmospheric conditions to simulate what the weather conditions will be in the coming months. It has been shown that this strong coupling between the sea surface temperature and the atmospheric conditions leads to reasonable predictability when it comes to seasonal weather patterns.
It should be emphasized here again that this coupling allows for the prediction of the average weather conditions across a reasonably long period (usually on the order of a month).
Limits of the Predictions
Since the seasonal predictions are made possible by the coupling between the atmospheric conditions and slow-moving weather conditions, one of which is the sea surface temperature, there will be some restrictions on the models based on the properties of these slow-moving conditions. Most notably, how well can we predict these slow-moving conditions, on average, in the future?
Most models can only reasonably predict these slow-moving conditions to about 6 months in the future. This restriction means that, if we can only predict these important slow-moving conditions to about 6 months in the future, then we will only be able to predict the atmospheric conditions up to about 6 months in the future as well. Beyond the 6-month limit, the models are not able to reasonably predict the slow-moving fluctuations, which will cause the other predictions to also not be very reliable.
This leads us to the very important limit of seasonal models currently: most models can only predict seasonal conditions up to 6 months in the future.
Implications of Prediction Limits
Seasonal models are used to predict the average weather conditions over a somewhat long period that ranges from at least a month to 6 months. They are not used to predict weather conditions on a time scale of much less than a month because of the chaotic nature of the weather systems, and they cannot be used to predict more than 6 months because the slow-moving conditions cannot be predicted beyond this time frame.
This is one limit to the predictive power of the seasonal models, but understanding this limit helps to put into perspective the uses of seasonal models and how they can be used to get an idea of what kind of weather patterns to expect in the coming months.
