Solar forecasting nonprofit Open Climate Fix awarded £500k from Google
The grant will permit the business to adapt and apply elements of its “transformer” models, which were used previously in predicting the shape of proteins and formed part of breakthroughs such as Google DeepMind’s AlphaFold-2, to solar forecasting.
The combination of technology with work will improve the ability of existing forecasting models to predict sunlight a few hours onwards, which is presently considerably impacted by clouds changing forms and obstructing sunlight.
Images will be taken every five minutes from a geostationary satellite, which machine learning algorithms will then use to determine where and how clouds move and change over a period of time.
This machine learning on satellite images will be used together with traditional weather forecasts and maps of solar generation units to create a forecast that is updated every five minutes.
The funding over the next two years will allow Open Climate Fix – which was set up by the former Google DeepMind engineer to deliver an online solar electricity forecasting service that covers the UK and Europe.
“We’re delighted that Google.org has chosen to support us in this venture,” (Dr Jack Kelly, 2021)
The business wants to do as much as they can to decrease emissions fast, with Open Climate Fix setting a target to aim towards 100m tonnes per year by 2030.
To extend further the impact of its studies, code and research outcome from the work will be entirely open, and the business will aim to collaborate with other forecasting organisations.