Field survey with cumulus clouds at the observatory in Falkenberg, Germany. A tower with observation instruments and a custom-made solar radiation sensor in the foreground. Photo: Wouter Mol
In spring 2020, the weather gods treated us to amazingly clear skies. Chiel van Heerwaarden’s group concluded this was partly due to the cleaner air thanks to the first COVID-19 lockdown but the combination of exceptionally dry weather and few clouds was at least ten times more important. And such unusual weather conditions are in turn the result of climate change, whereby extreme heat, rainfall and drought are becoming increasingly likely. Clouds play an important but elusive role in this process. Van Heerwaarden is trying to encapsulate them in a 3D model, combining the eye for detail of the Dutch Old Master Van Ruisdael with techniques from the cinema.
Extreme weather is becoming more common – just think of the heatwaves and extreme droughts of the summers of 2018 to 2020, or the unusually heavy downpours in Limburg last year. Another example of extreme weather was the spring of 2020, which was the sunniest spring in the past century. While any given example of extreme weather can be difficult to attribute to climate change, the trend shows that such extremes are becoming more likely. Chiel van Heerwaarden, an assistant professor in the Meteorology and Air Quality chair group and associated with the Wageningen Institute for Environment and Climate Research (WIMEK), is developing more accurate weather models that will be better able to predict the erratic behaviour of the climate of the future. He focuses in particular on clouds. Figuring out clouds could not only improve our understanding of climate change, but can also help electricity grid operators to take account of peaks and troughs in solar energy.
Clouds are perhaps the most variable aspect of the weather. That is what makes it so difficult to predict what they will do, whereas precisely that is what we need to know, according to Van Heerwaarden. “We need that information for the energy transition. We want to know about fluctuations in the incident solar radiation at solar farms so that grid operators can keep the electricity grid stable. To do that, we need to understand clouds.” When there are no clouds, solar panels create a peak in the electricity grid supply, which can overload the network. Van Heerwaarden: “Current weather models are not yet able to forecast individual clouds.”
Clouds are still elusive for current weather models
The probability of extreme weather, such as heavy rainfall, is increasing. Photo: Shutterstock
So clouds are still elusive for today’s weather models. The models of the Royal Netherlands Meteorological Institute (KNMI) are accurate to two kilometres but most clouds are a lot smaller. What is more, the KNMI model gives the probability of clouds, but clouds are not static entities. They sail overhead and cast a shadow on the land for a certain time. How can the predictions be improved? “Using a two-pronged approach”, explains Van Heerwaarden. “Firstly with more accurate measurements and secondly with a better model for clouds.” His group is working on both aspects in the large Dutch Research Council-funded project ‘Shedding Light on Cloud Shadows’, which started in October 2019.
Block 1: Measuring clouds
“In this project, my fellow researcher Bert Heusinkveld has developed a cheap instrument for accurately measuring light that will help us understand the interaction between clouds and sunlight”, says Van Heerwaarden. “A light meter like this would normally cost a couple of thousand euros whereas this light meter costs about 100 euros and measures the entire light spectrum — from UV to visible light and near infrared — ten times a second. The instrument is so cheap that we can install dozens of them in a field at 50 metres from one another in a lattice pattern. That level of accuracy in time and space will show the clouds passing over in our measurements of sunlight.” Van Heerwaarden and his colleagues proved that the light meter works in two ‘field campaigns’ in Germany and Spain. Those first measurements threw up some surprising results. A cloud that casts a shadow does not always lead to less sunlight on the land’s surface. Van Heerwaarden: “Sometimes you get more light on the ground when there are broken clouds in the sky than with clear blue skies. That’s because you don’t just pick up direct sunlight, you also get the light scattered by the edges of the clouds. The light meters show us these peaks crossing the land, sometimes giving up to 20 per cent more solar energy than with clear blue skies.”
Light scattered by the edges of clouds can sometimes actually produce more solar energy. Photo: Shutterstock
Block 2: Making clouds
For the second part of the approach — a better model for clouds — Van Heerwaarden and his colleagues have created a virtual laboratory in which they simulate clouds in realistic 3D. “Our group has moved towards less abstract models, ones where the computer simulates what you see outside. I find that fascinating. It reminds me of the painting View of Haarlem with Bleaching Grounds by the Dutch Old Master Jacob van Ruisdael, with so many patterns of shadows and clouds. That’s the level of detail you should be thinking of.” In the computer simulations, Van Heerwaarden’s group looks at how light moves through the clouds. They use fluid dynamics (the physics and maths that describe the behaviour of liquids and gases) to simulate cloud movements. The model simulates the clouds to a resolution of ten metres. He is enthusiastic about the model: “For the first time, we can examine clouds in detail and run realistic simulations. That gives us the opportunity to improve our understanding of all kinds of spectacular weather phenomena, such as hurricanes, large storms and perhaps even tornadoes. Many such phenomena are difficult to measure. Our combination of measurements and simulation models will let us make real progress in our fundamental knowledge.”
Learning from the film industry
Even 3D simulations like these of clouds fail to fully capture the interaction between clouds and sunlight, though. Van Heerwaarden points to the effect mentioned earlier that they saw with the light meters: that the scattering of sunlight on the edge of clouds increases the solar energy at ground level. “In our current models, sunlight reaches the Earth’s surface in a straight line. That means the incidence of light in the shadows is not correct. We will need to resolve this issue if we want to match Van Ruisdael’s landscapes in our simulations.”
They found part of the solution in the gaming and film industry
A 3D animation of clouds and cold pools of Tropical Convection. Video: Wouter Mol
They have found the solution in the gaming and film industry. In the past, animation studios like Disney’s Pixar would draw the light sources, shadows and incident light by hand but now they simulate how light works in practice, with rays emanating from a light source onto objects, reflecting and scattering. The technique is called ray tracing. Van Heerwaarden wants to use this technique to obtain 2D pictures showing the implications of the findings for the 3D clouds for the amount of sunlight that reaches the Earth’s surface. This will let him see precisely how quickly the sunlight falling on a solar farm changes and by how much.
The weather model of the future
The 3D models do have a drawback though: running them is time-consuming. Van Heerwaarden runs his cloud simulations on the Dutch national supercomputer Snellius in Amsterdam. “It has the processing power of at least 10,000 ordinary computers. Using this supercomputer, it takes us about a day to simulate one day’s worth of clouds. That helps us understand how clouds work but it is not much use for a weather forecast — you get the results once the day has finished.” Anyway, that level of detail is not necessary, so the meteorologists are looking for a compromise. “The big question is whether we can come up with clever methods that let us create 2D pictures of incident light without all the processing power needed for ray tracing”, says Van Heerwaarden. “Perhaps we can use some sophisticated statistics or machine learning to find out what the 2D picture looks once we know the cloud pattern.” Van Heerwaarden’s work on better models extends beyond the Dutch borders. He is involved in the European NextGEMS project, ‘Next Generation Earth Modelling Systems’, a collaborative venture between 14 countries to develop more reliable weather models at the global and regional levels that include extreme weather. They want to improve the precision of climate models from 100 kilometres to 1 kilometre. “That is a much higher level of detail. If we can simulate climate change to that level, we may discover changes in cloud formation that we missed in the past.” That could let the Netherlands prepare for the next heatwave or heavy downpour, and also allow the energy network to make optimum use of every peak in solar energy.
Read more about Chiel van Heerwaarden's research project
Read more about ray tracing
Read more about the NextGEMS project
Read more about the research of the Wageningen Institute for Environment and Climate Research (WIMEK)
Share this article