Google says the new AI model allows for almost “immediate” weather forecasts
Google says new AI provides almost “immediate” weather forecasts that can help keep forecasts accurate in a rapidly changing climate
- A new weather forecast model makes almost ‘immediate’ forecasts possible
- Google says that ‘nowcasting’ can call up a forecast in minutes instead of hours
- The model uses machine learning and AI and has initially shown success
- Traditional models still perform better for long-term forecasts
Google throws the power of its AI and machine-learning algorithms behind the development of faster and more accurate weather forecasts.
In a blog post, Google describes a new model developed by the company called ‘nowcasting’, which is said to have been successful in accurately predicting weather patterns with ‘almost immediate’ results.
According to a new article, the method is capable of producing predictions up to six hours in advance in just five to 10 minutes – figures that, according to the method, surpass even in early stages.
Although some traditional predictions generate huge amounts of data, they can also take hours to complete.
“An important advantage of machine learning is that inference is computationally inexpensive, given an already trained model, allowing for forecasts that are almost immediate and in the original high resolution of the input data,” writes Google.
“If it takes 6 hours to calculate a forecast, it only allows 3-4 runs per day, resulting in predictions based on 6+ hour old data, which limits our knowledge of what is happening now. Nowcasting, on the other hand, is particularly useful for direct decisions from traffic routing and logistics to evacuation planning. “
When Google compared the accuracy with three current traditional models that used historical weather data between 2017 and 2019, researchers discovered that it performed equally well if not better.
Google says its model can help solve problems predicting the weather caused by climate change – namely more erratic patterns
The ability to make fast, on-the-fly predictions can prove to be particularly useful if the global climate changes, increasing weather patterns whimsical and sometimes more extreme.
Although speed is one of the key benefits of the Google model, the company says that nowcasting is still surpassed by traditional forecasting methods when it comes to long-term predictions.
It is therefore unlikely that the new Google system will completely replace traditional forecasting models, but is more likely to be used to supplement them and fill gaps for short-term coverage.
As noted by The Verge, Google is not the only one developing new and advanced ways to predict the weather. Both IBM and Monsanto are both working on their own methods.