Researchers have used deep learning to model more precisely than ever before how ice crystals form in the atmosphere. Their paper, published this week in PNAS, points to the potential to significantly increase the accuracy of weather and climate forecasts.
Researchers used deep learning to predict how atoms and molecules behave. First, models were trained on small-scale simulations of 64 water molecules to help them predict how electrons in atoms interact. The models then replicated these interactions on a larger scale, with more atoms and molecules. It is this ability to accurately simulate electron interactions that allowed the team to accurately predict physical and chemical behavior.
“The properties of matter arise from how the electrons behave,” says Pablo Piaggi, a Princeton University researcher and lead author of the study. “Explicitly simulating what happens at this level is a way to capture much richer physical phenomena.”
It is the first time this method has been used to model something as complex as ice crystal formation, also known as ice nucleation. This is one of the first steps in cloud formation, where all precipitation comes from.
Xiaohong Liu, a professor of atmospheric sciences at Texas A&M University who was not involved in the study, says that half of all precipitation events, whether snow, rain or sleet, start as ice crystals, which then grow and give rise to precipitation. If researchers could model ice nucleation more accurately, it could give a big boost to weather prediction in general.
Ice nucleation is currently predicted from laboratory experiments. Researchers collect data on ice formation under different laboratory conditions, and this data is fed into weather prediction models under similar real-world conditions. This method works well enough at times, but often ends up being inaccurate due to the large number of variables involved in actual weather conditions. If even a few factors vary between the lab and the real world, the results can be quite different.
“Your data is only valid for a certain region, temperature, or type of lab setup,” says Liu.
Predicting ice nucleation from the way electrons interact is much more accurate, but it is also very computationally expensive. It requires researchers to model at least 4,000 to 100,000 water molecules, and even on supercomputers, such a simulation could take years to run. Even this could only model interactions for 100 picoseconds, or 10-10 seconds, not enough time to observe the ice nucleation process.
With deep learning, however, the researchers were able to run the calculations in just 10 days. The time duration was also 1,000 times longer, still a fraction of a second, but enough to see nucleation.
Of course, more accurate models of ice nucleation alone won’t make the forecast perfect, Liu says, since it’s only a small but critical component of weather modeling. Other aspects are also important: understanding how water droplets and ice crystals grow, for example, and how they move and interact together under different conditions.
Still, the ability to more accurately model how ice crystals form in the atmosphere would significantly improve weather predictions, especially those involving whether and how much rain or snow is likely to fall. It could also help weather forecasting by improving the ability to model clouds, which affect the planet’s temperature in complex ways.
Piaggi says future research could model ice nucleation when substances such as smoke are present in the air, further improving the accuracy of the models. Because of deep learning techniques, it is now possible to use electron interactions to model larger systems over longer timescales.
“This has essentially opened up a new field,” says Piaggi. “It is already playing and will play an even bigger role in chemistry simulations and in our simulations of materials.”
