Weather is playing an increasingly critical role as Australia transitions to net zero – both as a fuel source for renewable electricity supplies, such as wind and solar farms, and as a predictor of energy demand.
Previously, most electricity flowed into Australia’s power systems from large-scale conventional generators fuelled by coal and gas, along with large hydroelectric plants.
However, with many of the coal-fired generators reaching their end of life and accelerating climate change policies, there’s been a widespread and rapid uptake of renewable generation in Australia.
This includes wind and solar farms scattered throughout regional areas, and distributed photovoltaics (DPV or rooftop solar) in residential areas.
As a result, weather has become a key ‘fuel source’ and the role it plays is only going to become even more vital as the transition unfolds.
And because weather is variable and uncertain, forecasting weather-influenced generation and demand has become essential to maintaining a secure and reliable electricity grid.
One of the core responsibilities of the Australian Energy Market Operator (AEMO) is to operate the National Electricity Market (NEM) in the east coast of the country and the Wholesale Electricity Market (WEM) in Western Australia.
Across both power systems, AEMO matches real-time electricity production (generation) to demand (also known as load).
Part of this role is ensuring both supply and demand are balanced at all times to maintain system security and reliability.
Only 10-to-15 years ago, temperature was the primary weather factor when forecasting demand in Australia’s power systems. This was when around 90% of electricity was supplied by burning coal and gas.
Back then, a day ahead weather forecast was used when planning for the next day’s demand.
Warmer, or colder than average temperature predictions would prompt AEMO to ensure enough coal or gas plant were online to meet additional demand as consumers turned up, or down, their heating and cooling systems at home.
Now, wind and solar generation account, at times, for over 70% of Australia’s electricity supply and this is only expected to increase as aging coal-fired plants are retired, and more renewable generators come online, along with battery storage.
In the NEM, there are 22.8 million people, with homes across the region hosting about 17 gigawatts (GW) of rooftop solar capacity – equivalent to 28% of grid-scale generation capacity in the NEM. Wind and solar farms account for a further 20 GW of capacity.
In the WEM, there’s a smaller population of 2.7 million, which has around 1.96 GW of rooftop solar installed, and is equivalent to about 33.5% of total grid-scale capacity. Solar and wind farms in the region have a combined capacity of 1.17 GW.
With wind and irradiance from the sun being highly variable fuel sources, weather forecasts take into account an ever-widening range of factors and how they interplay.
Simply, a wind farm cannot generate without wind, while a solar farm cannot produce electricity without irradiance from the sun.
Other factors that come into play in forecasts include temperature, humidity, cloud distribution, dust and sand storms, bush fire smoke, lightning, rain, tornados, and cyclones. Even rare celestial events such as solar eclipses have to be planned for when maintaining the supply-demand balance.
Additionally, times of the day, days of the week, seasons and holidays can impact demand on the system and are incorporated into AEMO’s forecasts.
For example, summer is traditionally known as the peak demand period of the year, with hotter temperatures resulting in larger electricity use for cooling. Whereas rainfall can impact human behaviour with more people staying home and using heating and lighting.
Meanwhile, weekends or national holiday periods, generally have lower demand with less commercial activity.
In maintaining secure, reliable and affordable power systems, it is critical to understand how all these variables interplay.
Weather forecasting is now essential to operating power systems, in addition to other factors such as temperature and potential demand scenarios.
Roughly, AEMO produces more than 3 million point forecasts a day.
In gathering data from rooftop solar, as well as wind and solar farms, AEMO uses specialised forecasting systems.
These advanced systems use a combination of statistical and physical forecasts to predict energy output from rooftop solar, as well as wind and solar farms. Satellite imagery is also used in these forecasts.
In generating its forecasts, AEMO collates data from many sources, including global satellite feeds, numerous demand and weather forecast providers, and over 515 weather forecasting stations around the country, along with market participants’ own on-site weather stations at their respective wind or solar farm.
In the NEM, AEMO also uses an advanced model called the Forecast Uncertainty Measure (FUM) that incorporates more than 1 billion forecasts and is underpinned by Machine Learning (ML) models.
These forecasts are used to ‘train’ the FUM model based on historical errors and conditions that were present at the time the projection was produced.
This model is retrained quarterly.
All this information flows-in to AEMO’s forecasting teams every minute of every day and is collated and fed into AEMO’s models, which are becoming more complex, with new capabilities.
All these inputs are incorporated into AEMO’s Demand Forecasting System, which creates projections for each region of the NEM, and predicts how demand will change under certain conditions.
AEMO produces longer-term outlooks, and numerous shorter-term forecasts including the one week ahead short-term projected assessment of system adequacy (ST PASA).
These ST PASA forecasts are further refined as they get closer to dispatch time (AEMO’s final direction to generators) and are updated at intervals from the day ahead through to five-minutes pre-dispatch, and finally dispatch.
AEMO’s goal is to manage 100% instantaneous penetration of renewables at any moment on any day across both the WEM and NEM.
With weather potentially a 100% fuel source across the country on occasion, it is crucial to understand all variables and the impact they can have on supply, demand and risk in operating the electricity systems.
Commenting on the increasing importance of weather forecasting when planning for our electricity systems, AEMO Data Science Lead – Operational Forecasting Josh Brodrick said that in the past, the impact of temperature was the primary focus of assessment on the day prior.
“While temperature is still considered when assessing and planning for demand, other aspects of weather are also measured due to the potential impact on supply and risk posed to the system.”
“On the supply side, lack of wind will result in reduced ‘fuel’ and therefore generation at our wind farms, while very strong wind can cause damage to plant and infrastructure.
“Clouds, dust storms and bushfires can all impact the electricity generated by solar farms or rooftop PV by obstructing the sun.”
With characteristics of renewable generation and the variability of weather, the direction of supply and demand can swing quite rapidly.
“For example, in South Australia on a sunny day with a decent amount of wind, renewables can generate enough electricity to meet total demand, although synchronous generation is needed to maintain system security.”
“However, we know that weather is changeable and energy supply from variable renewable energy sources can fluctuate rapidly. This is why forecasting the weather ahead of time, while monitoring in real-time is critical to ensuring the grid can rapidly flex with the variable weather.”
“Forecasting, monitoring and modelling allows us to plan for these sudden changes in weather by having other generators and energy storage on standby to rapidly call upon to fill in the gaps and maintain a stable supply-demand balance,” Mr Brodrick explained.
With the energy transition gaining momentum and an ever increasingly weather sensitive power system, AEMO continues to build and refine its forecasting capabilities to improve situational awareness and manage risk.
New initiatives will involve deploying more weather monitoring infrastructure – particularly in renewable energy zones (REZ), where there are high levels of wind and solar generation. These REZ will then be connected to various regions via critical transmission infrastructure.
More data will be received more often, and will come from local, regional, national and global sources. It will feed into algorithms and data sets with the goal of providing even further refined and timely forecasts.
This is important for market efficiency and critical to the secure operation of the power system now and into the future.