Big Data Analytics for Weather Forecasting

Weather forecasting and more specifically precipitation forecasting is very important for our society.

Photo from


In Norway, the Norwegian Meteorological Institute ( runs its numerical weather prediction (NWP) model (Arome Norway) to forecast both long period and short period weather. The output of the weather prediction is not always accurate. Extra information such as weather observations are gathered over the whole Norwegian territory. They enable i-to quantify the error and ii- correct the output of the NWP.

The quantification and the correction of the NWP output with the observation can be done posteriori. The key challenge here is to know when the forecasting model may be wrong. The emerging big data analytics can be a powerful tool to tackle the aforementioned challenge. Through studying the connection between the forecasting data and the observation stations’ data, we may understand when to fully trust the weather forecast and, if it is not the case, how to correct future forecast.


Weather precise forecasting based on big data and predictive analytics

What will I do?

  • Collect two datasets: weather data from the Norwegian Meteorological Institute ( and the data from observation stations (
  • Compare the output from the weather forecasting model and the data from observation stations.
  • Check if the two data are same or different. If different, we need to find the connection between the two data sets.
  • Using big data analytics and predictive analytics to find out when the prediction model may be not accurate
  • Build correction model/algorithms for the forecasting model to improve the prediction accuracy
  • Predict the extreme weather
  • Develop visualization program for easy understanding
Publisert 18. aug. 2016 08:20