Researchers at tech giant Microsoft have come out with a new Machine Learning based forecasting model along with a comprehensive dataset, named SubseasonalRodeo, to train the subseasonal forecasting models. It is a model system which is able to predict the temperature or precipitation 2-6 weeks in advance in the western contiguous United States. The researchers at the company have presented the details of their work in the paper titled ‘Improving Subseasonal Forecasting in the Western U.S. with Machine Learning’.
According to the research team, a large amount of high-quality historical weather data along with the existing computational power makes the process of statistical forecast modeling valuable. Additionally, associating together the physics- and statistics-based approaches lead to superior forecasts. Their Machine Learning-powered prediction system comprises the two regression models that are trained on its SubseasonalRodeo dataset. The dataset involves various weather measurements dating as far back as the year 1948. It includes precipitation, temperature, sea surface temperature, sea ice concentration, and relative humidity and pressure. This data is consolidated from the source of the National Center for Atmospheric Research, the National Centers for Environmental Prediction, and the National Oceanic and Atmospheric Administration’s Climate Prediction Center.
The Researchers at Microsoft will be further increasing its work to the Western United States and will continue its collaboration with the Bureau of Reclamation and other agencies. As the company’s researcher Lester Mackey noted, the subseasonal forecasting system is productive for Machine Learning development, and the company has just scratched the surface.