Researchers of Karlsruhe Institute of Technology (KIT) are now using a novel approach using statistics and artificial intelligence to correct systematic errors in weather models.
The new approach can accurately forecast wind gusts much more reliably when compared to other traditional methods.
According to the researchers, incorporating geographical information along with extra meteorological variables such as temperature enhances forecast quality dramatically.
They were able to achieve the best results while utilizing artificial intelligence methods based on advanced neural networks.
This one-of-a-kind technique can drastically help in raising quick alerts in disastrous conditions like squalls, which have a speed of more than 65 kilometers per hour and have the potential to cause massive damage to the region.
A doctoral researcher at the KIT’s institute of Stochastics, Benedikt Schulz, said, “Wind gusts are difficult to model, as they are driven by small-scale processes and are locally limited.”
He further added that their capacity to anticipate the weather using numerical weather forecast models utilized by weather agencies is limited and unpredictable. Therefore, the new approach can turn out to be of great value in accurately predicting dangerous wind gusts.
The researchers want to use AI to address different systematic flaws, enhance forecasts, and more accurately predict catastrophic weather events. Researchers said that they scrutinized available methods for statistical post-processing of numerical weather forecasts and systematically compared their forecast qualities.
Dr. Sebastian Lerch, head of the junior research group named AI Methods for Probabilistic Weather Forecasts, said, “Yet, AI methods are far superior to classical statistical approaches and produce far better results, as they allow for a better consideration of new information sources, such as geographical conditions or other meteorological variables, such as temperature and solar radiation.”
He also mentioned that, on average, AI approaches to cut the forecasting errors of weather models by 36 percent.