Tracking clouds to forecast the solar future

In many markets around the world, renewable energy mixes are regularly meeting 100% of daily demand, driven by complimentary mixes of solar and wind. Grids with high penetration of rooftop solar are seeing daytime electricity prices drop to nominal or negative values, as grid operators try to balance supply against demand. Volatility in merchant markets is driving investment in short-term storage to allow solar producers to take advantage of variable pricing and intra-day storage to leverage evening price spikes. All these changes are accelerating, as power consumers, suppliers, and grid operators seek solutions whilst investment, manufacturing capacity, and global solar capacity continue to increase at record levels.

As solar and other distributed energy resources change the supply and demand mix across grids and electricity markets, the ability to predict solar production has evolved from preferable to essential. Across the industry, historical and forecast solar irradiance data is being used by developers to assess and invest in new sites, by operators to make smarter storage and power management decisions, by power traders to maximise revenues, and by grid operators to more effectively balance supply and demand. To that end, embracing a digital, data-driven approach is becoming critical for companies that wish to harness the weather as the new fuel and build reliable grids powered by renewables.

The increasing role of technology

From distributed rooftop systems to utility scale solar farms, digitalisation and digital twin technologies can substantially improve efficiency and performance. However, the growing adoption of digital platforms means industry players require application programming interface (API)-enabled solar data sources with global scale capabilities.

As the industry gets bigger and more complex, more integrated and trusted data sources are needed as everyone from owners, operators, investors, traders, grid operators, and even residential homeowners with solar, seek to understand more about how their assets are, should be, and will be, performing. The solution lies in leveraging best-in-class technologies that make employees and workflows more efficient – and businesses more profitable. Ultimately, this digital-first approach will unlock the innovation and scale needed to realise a net-zero future.

Improved capabilities through next-generation satellites

High-resolution satellite images (Figure 1) have been fundamental to improving the scale and accuracy of solar irradiance data. In particular, they enable the accurate, granular tracking of clouds, the largest determiner of solar irradiance in most areas of the world.

The quality of satellite imagery is continually improving through updated geostationary weather satellites. The latest generation, launched globally between 2014 and 2022, has delivered an increase of around 2x in spatial resolution, 3 – 6x in temporal resolution, and 3x in spectral resolution. That equates to a remarkable 20 – 30x improvement overall.

To generate a complete solar irradiance picture, Solcast combines satellite images with other weather inputs such as atmospheric pressure, water vapour, aerosols and ozone, and surface reflectivity (albedo). Combining this network of global data sources allows the generation of a solar specific data set, turning the nightly news weather forecast into a solar-specific accurate forecast suitable for the renewable energy industry. As the industry continues to grow, new innovative applications are being developed on top of a solar specific data stack, enabling the organisations that are working to capture the weather as the new fuel.

Building accuracy through intelligent models

When it comes to solar applications, the fine resolution detail of clouds, aerosols, and terrain can have large effects on the accuracy of irradiance data. Therefore, Solcast’s team have focused on investing significant time and resources into building models running at native satellite resolution. Excluding oceans, this 1 – 2 km resolution results in about 100 million points around the world where the company’s data is updated every five or 10 minutes. That is up in the tens of billions of updates and forecasts each day. By comparison, the global weather forecast models that traditional existing forecasts use only update twice per day and at only about 10 – 20 km resolution, resulting in more than 1000 times less updates and forecasts per day.

In addition to precise cloud tracking, an accurate model requires the modelling of the physical characteristics of clouds. Models that interpolate between satellite imagery makes clouds ‘fuzzy’, which does not reflect how they alter irradiance. Instead, Solcast’s 3D cloud modelling helps to more accurately understand (and predict) how clouds form, move, and behave.

Beyond clouds, various other factors can also influence solar generation. To achieve the accuracy levels suitable for the solar industry, the company utilises a range of inputs, models, and algorithms to forecast irradiance with industry-leading accuracy. For example, it has created various models to track local aerosols such as pollution, dust, salt, smoke, and ash. Aerosol modelling is particularly valuable in desert and semi-desert regions, where atmospheric dust often surpasses clouds as the primary determiner of solar irradiance. Aerosols are also prevalent in highly urbanised countries like China and India, where emissions from industry, transport, and agriculture cause high levels of air pollution. The impacts of aerosols on irradiance were seen throughout 2023 in North America, where strong winds carried smoke from Canadian wildfires hundreds of kilometres and impacted solar generation far from its origin (Figure 2).

For a complete irradiance picture, Solcast also tracks the reflectivity of Earth’s surface, known as albedo. Specific models analyse satellite images to separate clouds from the ground, which is particularly important in snowy or sandy regions.

Overall, the methodology for arriving at a single metric like global horizontal irradiance (GHI) requires a ‘clear sky’ model that considers aerosols, water vapour, air pressure, elevation, and albedo, and then subtracts terrain and cloud effects. By processing 600 million global forecasts every hour, Solcast can generate accurate estimates – up to 14 days ahead – for almost every location on Earth.

Enjoyed what you've read so far? Read the full article and the rest of the Summer issue of Energy Global by following the link or below, or why not register today for free!

For more news and technical articles from the global renewable industry, read the latest issue of Energy Global magazine.

Energy Global's Summer 2024 issue

The Summer 2024 issue of Energy Global starts with a guest comment from Terrawatt on the streamlining of the permitting process in Italy, before moving on to a regional report from Frost & Sullivan on the energy landscape in Asia Pacific. This issue looks at key topics such as wind installation vessels, offshore wind turbine foundations, weather analysis, solar maintenance, and more!

Read the article online at: https://www.energyglobal.com/special-reports/26062024/tracking-clouds-to-forecast-the-solar-future/