Go to top of page

Short-term forecasting: Improving predictability of renewables

Photo of a cloud forecasting device in the foreground with wind turbines in the background. The ground is red soil and the sky is heavy with clouds.Image credit: Fulcrum3D​​THE NEED

In the search for approaches that enable higher shares of variable renewable energy sources in the grid, ARENA has established an initiative to improve the accuracy of forecasts for the energy generated by large-scale renewables.

Accurate short-term forecasts of cloud movements and wind changes allow the Australian Energy Market Operator (AEMO) to better balance electricity supply and demand, which keeps the grid stable. They also assist AEMO in maximising the use of the lowest-cost generators, helping keep electricity prices down.

Improved forecasting approaches also benefit solar and wind farm operators, who rely on accurate forecasts to calculate how much electricity they can produce. This reduces their exposure to the costs of balancing the system when their output is less than AEMO expected.


Presently, AEMO is responsible for forecasting how much electricity will be generated by wind and solar farms. In early 2018 ARENA and AEMO established the Short-Term Forecasting initiative to trial wind and solar farms submitting their own ‘five-minute ahead’ forecasts to AEMO. The market operator can then integrate this information with its own data to improve overall accuracy.

In March 2019 ARENA awarded $9.4 million to 11 projects to implement the trial, establishing short-term forecasting at large-scale wind and solar farms across Australia. The trial comprises 35 per cent of the NEM’s registered large-scale wind and solar capacity, and collectively includes a total of 3.0 GW of renewable electricity generation.

The projects were selected to ensure diversity of forecasting technologies and generator sizes, hardware types and climate regions, including multiple solar and wind and one hybrid generator across NSW, SA and VIC.

As the trials continue there will be ongoing knowledge sharing reports and forums to communicate the learnings to the broader industry.




$9.4 million


$21.1 million

Solar 1.5 GW

Wind 1.5 GW


The objectives of ARENA’s Short-term Forecasting initiative are to:

  • demonstrate the ability for semi-scheduled generators1 to submit five-minute self-forecasts via AEMO’s web-based Application Programming Interface (API)
  • demonstrate the ability for self-forecasting to be more accurate than the equivalent forecast produced by the Australian Wind Energy Forecasting System or Australian Solar Energy Forecasting System (the systems currently used by AEMO)
  • explore the potential commercial benefits for semi-scheduled generators of investing in forecasting approaches
  • examine factors that affect the accuracy of the forecasting approaches trialed (e.g. in different weather, operational conditions or geographies)
  • improve the commercial and technical readiness of forecasting providers and technologies
  • contribute to industry knowledge sharing.

As at July 2019, the initiative is five months into the two-year trial. However, considerable outcomes have already been achieved, as described below.

More solutions for delivering secure, reliable and affordable energy and increasing the value provided by renewable energy

With 11 different forecasting approaches being trialled, the initiative will help solar and wind farms understand which technologies are likely to work best given their local situation.

Another key outcome to date is the development, testing and implementation of a new API. As part of this process, AEMO has changed its systems and procedures to give all generators participating in the NEM the option to submit their own forecasts for use in the dispatch process. This is a fundamental change for renewables, as it provides them an avenue to participate more fully in the market bidding and dispatch process.

As at 30 June 2019, 23 generators had registered to participate in the self-forecasting process and cleared the first API milestone, including 16 generators from ARENA’s short-term forecasting trial. The number of participants is expected to grow through 2019 to encompass at least the 28 generators that are involved in ARENA-supported projects.

Self-forecasting could also lead to further innovations, including more sophisticated approaches to operating renewable generators and storage. Renewable energy developers that have more visibility and control of their generation plant through short-term forecasts will have greater ability to create custom, site-specific generation entities such as solar co-located with storage plants, or to guarantee renewables firming across a portfolio of generation entities.

Increased collaboration on energy innovation

This trial is a prime example of how ARENA’s fostering of collaboration can give rise to innovation. The initiative was conceived and developed through ARENA’s A-Lab program in 2017, with key input from ARENA, AEMO and Clean Energy Finance Corporation (CEFC) representatives. There was subsequently broad stakeholder input through a series of workshops reaching more than 60 people, and input from AEMO’s relevant industry reference group.


Increase in employment in renewable energy activities

One impact already realised through the trial is the creation of a new market for the provision of short-term forecasting capability to renewable generators, in both the wind and solar sectors. Demand for self-forecasting services is expected to increase with generator appetite to connect to the API. This market will be underpinned by the growth of industry experience and capability in self-forecasting approaches.

Self-forecasting will also make it possible for greater forecast certainty to translate directly to cost and revenue improvements for semi-scheduled plants.

Secure, reliable and affordable electricity system with a significantly higher share of renewable energy

Improved accuracy in short-term forecasts will allow the electricity system to operate with a higher share of renewable energy and lower costs incurred due to forecasting errors.


  1. The NEM generator category typically applied for generators with variable output (such as large-scale wind and solar) that are 30 MW or more.