Objectives of the service
Photo by American Public Power Association on Unsplash
Today, solar electricity generation causes problems for electricity grids. The power generated by solar generators changes rapidly as clouds move overhead. Existing forecasts struggle to predict clouds and so struggle to predict solar generation.
Due to the lack of good solar electricity forecasts, electricity grid system operators have no choice but to "firm up" solar electricity generation with fossil-fuel-powered "spinning reserve" generators. This is carbon-intensive and expensive.
It needn't be like this.
If we had more accurate predictions for solar electricity generation then we could reduce the amount of "spinning reserve" required. This would reduce carbon emissions and reduce costs to end-users, and increase the amount of solar generation the grid can handle.
Open Climate Fix proposes to create the world's best forecasts for solar electricity generation by applying machine learning to satellite imagery and numerical weather predictions. We will do cutting-edge research on vast quantities of data. But we won't stop at the research. We will also implement a functional solar forecasting system and deliver forecasts to the UK National Grid control room. We will work closely with users to ensure we meet their needs. Finally, we will measure impact on the grid's carbon intensity and costs.
Users and their needs
Our main users would be grid system operators. Other potential users include solar farm owners, energy traders, smart-home operators, and demand-side-response aggregators (who manage batteries and other assets). We intend to target the UK first and then later scale to Europe and then globally.
Service/ system concept
We will develop a Deep Learning machine learning model which takes a sequence of recent satellite images and numerical weather predictions. The model will output probabilistic solar electricity nowcasts for each PV system in the country. The nowcasts will be calibrated in near-real-time using live solar electricity data.
Deep Learning models excel when trained on huge amounts of data, so we will train the model across the entire geographical extent of the satellite imagery (not just the areas which happen to have solar electricity systems). As such, the model will be trained to predict the next few frames of satellite imagery as well as solar electricity generation.
Space Added Value
We will use geostationary satellite data from the EUMETSAT Spinning Enhanced Visible and Infrared Imager (SEVIRI) onboard the Meteosat Second Generation (MSG) satellites. Specifically, we will use the Rapid Scanning Service (RSS) onboard satellites MSG-9 & MSG-10. These provide 5-minutely images of the upper quarter of the Earth at 1km spatial resolution at nadir.
EUMETSAT SEVIRI RSS provides excellent data for nowcasting clouds. No other data source provides as wide a geographical coverage combined with high temporal resolution and good-enough spatial resolution.
‘Sky imagers’ are the main non-space alternative for nowcasting PV: fish-eye cameras pointing straight up at the sky, mounted near the PV system of interest. These cameras provide very limited spatial coverage, they cost considerable time and money for each new site, they struggle to infer the height of the clouds, and they need regular maintenance.
In contrast, SEVIRI RSS provides very wide spatial coverage, and the infrared channel allows us to infer the altitude of the cloud tops (because temperature decreases with altitude in the troposphere). SEVIRI RSS allows us, potentially, to provide PV nowcasts for any location in Europe or North Africa at minimal marginal cost. That is, once we build a nowcasting system for the UK, it should be fairly cheap and fast to scale to all other locations covered by SEVIRI RSS. Furthermore, SEVIRI RSS provides imagery of water vapour, which will probably help us to predict how much sunlight can pass through the atmosphere, and to predict the evolution of clouds.
Current Status
During the six month Kickstart Activity, Open Climate Fix conducted a total of 22 interviews to explore the user needs in greater detail. This also helped in establishing the technical and commercial feasibility of the proposed service.
Another vital part of the kickstart activity was the acquisition of early adopters and potential paying customers. This is an important first step in the now following sourcing of funding to start development of the service. Open Climate Fix is currently seeking to conduct a demonstration project together with ESA and UKSA and other private partners.
Prime Contractor(s)
