National Grid Electricity System Operator (ESO) and the Open Climate Fix are partnering to develop a ‘nowcasting’ model for solar output.
The initiative is intended to improve National Grid ESO’s short term forecasting in solar generation caused by cloud cover and in turn to maintaining the balance in the grid.
Nowcasting involves a machine learning model to forecast the near future, minutes and hours ahead, and has historically found use in predicting rainfall.
Open Climate Fix, a greenhouse gas emission reduction product lab founded by former DeepMind researcher Jack Kelly, applies a similar approach to sunlight prediction with a machine learning model to read satellite images and understand how and where clouds are moving in relation to the solar arrays below.
Carolina Tortora, head of innovation strategy and digital transformation at National Grid ESO, says that accurate forecasts for weather-dependent generation like solar and wind are vital for operating a low carbon electricity system.
“The more confidence we have in our forecasts, the less we’ll have to cover for uncertainty by keeping traditional, more controllable fossil fuel plants ticking over,” she says.
“We’re increasingly using machine-learning to boost our control room’s forecasts, and this latest nowcasting project with Open Climate Fix will bring another significant step forward in our capability and on our path to a zero-carbon grid.”
Changes in solar generation are difficult for grid operators to anticipate owing not only to uncertainty in forecasts, but also uncertainty around the location of many solar panels, most of which are connected to regional networks.
While work is underway by National Grid ESO and the University of Sheffield to map Britain’s solar panels, previously there’s been no way to anticipate short term swings in solar generation caused by cloud cover.
With the increased certainty in solar forecasts that should result with Open Climate Fix’s nowcasting service, National Grid ESO should be able to reduce the availability of back-up generation and bring more efficiency to balancing actions.
National Grid ESO has previously reported a 33% improvement in the accuracy of solar forecasts with machine learning.
The ESO also has just launched another initiative using machine learning to more accurately forecast the day-ahead reserve requirements.