Advances in weather forecasting and manipulation could boost energy efficiency in ways we’re yet to fully comprehend.
We often hear politicians talking about the challenges of renewable energy “when the sun doesn’t shine and the wind doesn’t blow”.
Advances in energy storage are likely to solve this problem, but knowing just how much energy we will be consuming at any given time is a bigger problem, in part driven by weather.
What if we knew exactly what the weather would be doing at any moment, and could prepare to change it if we didn’t like what was coming?
It may sound far fetched, but work is progressing on both fronts.
Experts say weather predictions are constantly improving due to both developments in the physical understanding of the climate system, but also improvements in our ability to run high resolution computer simulations.
“Accurate weather forecasts will always be very important in a highly renewable grid,” says Hannah Bloomfield, Postdoctoral Research Assistant in the Meteorology Department at the UK’s University of Reading.
One area of focus for researchers, says Bloomfield, is the sub-seasonal to seasonal prediction range, which can be anything from around one week ahead to one year.
“This is generally thought of as a 'grey zone' in predictability, that is, somewhere where there isn't a lot of meteorological skill.
“But many international groups have now developed methods to extract information about large scale weather conditions, like modes of atmospheric variability (the North Atlantic Oscillation is a classic one for the Northern Hemisphere) and use past knowledge of these to make links to surface weather conditions. This may be an area where we see greatest improvements by 2040.”
And with improved weather forecasts, says Bloomfield, will come better management of energy grids. At short timescales, improvement in the representation of clouds and rain means better forecasting of solar power generation through the day.
Improved forecasting using satellite data is already being tested in South Australia to deliver real-time solar energy output.
“At timescales of days ahead, improved knowledge of the timing of weather systems passing through will result in improved anticipation of wind power ramping events,” says Bloomfield.
“Days to weeks in the future, improved forecast accuracy will result in the ability of system operators to anticipate heatwaves or cold spells, and therefore to plan for where extra generation may be needed.”
Artificial intelligence and machine learning are also playing a role, as grid operators including the National Grid in the UK turn to models to improve solar forecasting.
In the US, Google says machine learning algorithms have enabled it to predict the output from its wind farms 36 hours in advance.
Sensing a cool change
Closer to home, the SunSPOT tool created by researchers at UNSW uses spatial and weather data to calculate solar radiation on a roof surface, and tell homeowners how much money they could save by putting solar on their rooftop.
The tool also uses machine learning methods to estimate electricity consumption and understand solar savings for households without smart meter data.
Imagine a world where smart meter data, improved weather forecasting and microgrids combined to deliver even better information to both electricity generators and users. In this world, both would be empowered to manage energy more efficiently.
Building owners with access to improved wind forecasts could enable the building to shift its load in real time to match the available wind power. Battery power could be tapped at the exact moment of a forecast lull in wind speed, the building’s air-conditioning adjusted upward a few degrees to reduce demand.
Cloudy with a chance of manipulation
And if the forecast weather doesn’t match available energy - some will change it.
China recently announced plans to expand its controversial weather modification programme, and says it will have developed a system that will enable it to mitigate high temperatures by 2025.