Team Seasonal Forecasting Engine
Seasonal Forecast Engine try not to compete with the regular kind of weather forecasting. But using information from decades ago, they can predict the skiing season and bathing season for the summer. The researchers try to figure out whether the temperature will be warmer or colder, if it is going to be raining or snowing – for the next three months ahead. Read more about seasonal forecast here.
In 2017, a team of ten researchers started working on the project. The captain steering the project is Research Professor Erik Wilhelm Kolstad, from Uni Research Climate (UNI) & Bjerknes Centre for Climate Research (BCCR). The researchers use machine learning to make tailored notifications. Big Data are gathered from different sources from all over the world. And computers are taught how to recognize patterns and forecast the weather for the future. This is an enormous amount of data to handle, more than any human being can.
– We are pleased that the project has been so well received. Many companies have contacted us because they want to be involved. For example, the Meteorological Institute is interested in the notification of sea ice, says Kolstad.
Why is seasonal forecasting useful for us?
If the ocean temperature in the winter are unusually warm, this can indicate if the spring will be warmer. If they know much snow will lay on top of the mountains in the winter, they know if there will be flooding in some regions the following spring. This will be useful for counties and cities so that they can prevent and take actions before flooding occur.
– This is a really exciting project. Season forecast will allow to better manage climate related issues and give warnings to population in some cases, says Anne-Sophie Schillinger, Business Developer at BTO.
Today, a lot is happening to climate research. With the help of supercomputers researchers can now process information and large-scale data, in quantities which is not humanly possible. With data collection from all over the world, supercomputers can predict weather, climate and season forecast.
The research team are building up their methods and optimizing data. For now, they are making prototype seasonal forecasts for the project participants only, before having method validated. The aim is to develop operational forecast at the end of the project.
– Currently, we are working on a simple prototype for our forecasting system. In the first round, we will give new forecasts with a couple of months in advance, but on long term, we wish to update our forecasts for each month, says Kolstad.
BTO is supporting the project on the user interaction front. In addition, we ensure building a sustainable platform, beyond the lifetime of the project. The project results are highly valuable for many actors within research, private and public sectors. For the research sector, the project will enable better seasonal forecasting, and models that are more accurate than the ones currently available. For the public and private sectors, “climate intelligence” could be used to better manage natural disasters, such as flooding, and enable better preventive actions.
Seasonal Forecasting for summer 2018
Last winter the engine gave its first seasonal weather forecast. The predictions in the forecast turned out to be very accurate.
– We nailed the forecast for this winter. It was a special situation where the development high up in the stratosphere gave us a clear warning of a cold winter. We cannot count on getting these clear signals from the nature each winter, explains Kolstad.
– The forecast for the rest of the spring is not ready yet, but will be ready in the start of week 16. April started cold, but now there is mild weather in big parts of the country. It will be exciting to see if we can give a spot-on forecast for May and June as well, says Kolstad.
The project is a collaboration between Uni Research, Bjerknes Centre for Climate Research, Geophysical Institute, University of Bergen, Nansen Environmental and Remote Sensing Center, Norwegian Computing Center and Bergen Teknologioverføring.