Forecasting renewable energy during a solar eclipse

A solar eclipse is one of the few events you can pinpoint years in advance, down to the minute. Forecasting what it does to solar and wind power is another matter. Because eclipses are so rare, there is little precedent to lean on, which makes them both a real challenge and a genuine chance to learn. That is exactly why we studied this one closely and built its effects into our forecasts, well before the day itself.

On August 12, 2026, a solar eclipse will move across parts of Europe. The path of totality runs over the North Atlantic, Iceland, and northern Spain, while a deep partial eclipse will be visible across much of the continent. In many of the markets we forecast for, the Moon will cover a large share of the Sun in the early evening. Over Belgium and the wider region, for example, that means roughly 85 to 90% coverage around 20:12 CEST, with the Sun already sitting low in the western sky.

For most people it's a striking thing to watch (with proper eye protection). For a team that forecasts renewable energy for a living, it's something else entirely: a rare, natural stress test for the models we rely on every day.

Here is why this matters for forecasting, and how we prepared for it.

Renewable forecasts live and die by weather parameters

Solar and wind generation forecasts are built on top of Numerical Weather Prediction (NWP) models. A handful of variables do most of the heavy lifting:

• SSR (Surface Solar Radiation): the total solar energy reaching the ground.
• DNI (Direct Normal Irradiance): direct sunlight hitting a surface pointed straight at the Sun.
• DHI (Diffuse Horizontal Irradiance): sunlight scattered by the atmosphere before it reaches the surface.
• Surface temperature.
• Wind speed and wind direction.

Get these right, and a solar or wind forecast has a solid foundation. An eclipse briefly disturbs almost all of them at once, which is exactly why it deserves attention.

What an eclipse does to the atmosphere

During an eclipse, the atmosphere suddenly loses part of its primary energy source: incoming sunlight. It is a bit like a sunset, except the drop is far more abrupt, and then it reverses just as quickly.

The measurable effects are well documented:

• Solar irradiance falls sharply.
• Surface temperatures can dip by roughly 1 to 6 degrees Celsius.
• Wind speeds may briefly ease, and local wind directions can shift as the lower atmosphere becomes more stable.

None of this is dramatic in isolation. Taken together, and compressed into a short window, it is a pattern that rarely shows up in normal weather data, so it is worth handling deliberately.

Making sure the models see it

Capturing an eclipse correctly is not automatic for every forecasting setup. A weather model only reflects the event if its radiation calculation is told that the Sun is being progressively blocked, and by how much, at each moment. Not every operational NWP model does this on its own.

That is precisely the kind of detail we account for in advance. Part of preparing for August 12 is making sure the reduced solar input is represented for the short window when it actually matters, rather than letting the forecast treat it as an ordinary evening. Knowing which inputs handle the eclipse, and correcting for the ones that do not, is what keeps a forecast reliable when conditions step outside the everyday.

So what does it mean for forecasts that day?

The honest answer: the overall impact should be modest, and that is mostly down to timing.

This eclipse peaks in the evening, shortly before sunset, when solar production is already in the steep, declining tail of the day. Output is low and falling fast anyway, so removing part of an already small number has a limited effect on total generation. Because it happens during that evening transition, the impact on wind generation is expected to be limited too.

For traders and system operators, the market impact should therefore stay contained. The more interesting challenge is not the drop in generation itself. It is the unusually rapid ramp down and ramp back up of solar power around the event. Sudden swings like that can create frequency control challenges for grids carrying a lot of photovoltaics. Had this eclipse fallen around noon, with the Sun high and solar output near its peak, the operational story would have been very different.

Prepared, not surprised

None of this is left to chance on the day. Because an eclipse is fully predictable, we studied this one ahead of time and made sure its effects are reflected in the forecasts our clients rely on across European markets.

That said, we are genuinely curious to see how the real world compares to what we prepared for. Rare events like this are a valuable check: they confirm what our models already handle well and highlight the details worth sharpening for next time. Forecasting is never finished, and moments like this are exactly why we keep refining it.

References

• Harrison & Gray (2017), *The weather's response to a solar eclipse. academic.oup.com
• Aplin et al. (2016), *Atmospheric changes from solar eclipses. arxiv.org
• Palomaki et al. (2019), *Observations of thermally-driven winds during the 2017 solar eclipse. mdpi.com
• Recent studies on eclipse impacts on temperature, humidity, and wind fields. mdpi.com