GB software start-up Open Climate Fix has been awarded more than £500,000 ($690,000) to advance its open-source solar forecasting offering.
Open Climate Fix’s solution uses images taken on 5-minute timescales from a geostationary satellite, i.e. that stays above a fixed point on the equator, to monitor the changing patterns and movements of clouds over time.
The aim is to improve forecasting up to hours ahead of the production from solar PV, which fluctuates as clouds move over the panels. With more accurate predictions, the demand, supply balance can be better managed and for example, fossil fuel reserves can be optimised.
The machine learning algorithm uses the recent history of the cloud images to determine and improve how they are moving and changing over time. With the funding Open Climate Fix intends to incorporate ‘transformer’ machine learning models into its solution.
Transformer models in essence convert the machine learning process from a sequential one to a more probabilistic approach. Notably, it is at the heart of recent breakthroughs such as Google DeepMind’s AlphaFold-2 in predicting the shape of proteins, which is expected to have impacts from the treatment of diseases to the breakdown of industrial waste.
Combining machine learning on satellite images, traditional weather forecasts and maps of solar power generation units should produce more accurate forecasts updated every five minutes.
“We’re delighted that Google.org has chosen to support us in this venture. We want to do everything in our power to reduce emissions as quickly as possible,” commented Jack Kelly, a former Google DeepMind engineer and co-founder of Open Climate Fix.
“Most electricity grids simply will not be able to reduce CO2 emissions to net zero without significantly better electricity forecasts. Net zero grids will be more complex than today’s grids and will require better coordination, which in turn need better renewable generation forecasts. That’s what we aim to deliver.”
Open Climate Fix aims to deliver a state-of-the-art online solar electricity forecasting service covering the UK and Europe.
The company intends to collaborate with other forecasting companies. The code and research results will be completely open in order to have maximum climate impact and be continually improved.
The funding was awarded as part of Google.org’s Impact Challenge On Climate, which funds bold ideas that aim to use technology to accelerate Europe’s progress toward a greener, more resilient future.