Cambridge researchers published a study in the journal PNAS about the creation of IcePic, an artificially intelligent algorithm. The algorithm can outperform scientists in predicting when and how different materials will form ice crystals. It could help atmospheric scientists improve future climate change predictions.
Scientists have created an artificially intelligent algorithm that can predict how and when different materials will form ice crystals, outperforming climate scientists. IcePic, the program, could help atmospheric scientists improve climate change models in the future.
Cambridge scientists have developed an artificially intelligent algorithm capable of outperforming scientists in predicting how and when different materials form ice crystals. IcePic is a program that could help atmospheric scientists improve climate change models in the future. The details were published today in the journal PNAS.
Water has some unusual properties, such as expanding when it freezes. Understanding water and how it freezes around different molecules has far-reaching implications in a variety of fields, from weather systems that can affect entire continents to storing biological tissue samples in a hospital.
Ice nucleation is extremely important for the atmospheric science community and climate modeling. Other than direct experiments or costly simulations, there is currently no viable way to predict ice nucleation. IcePic should make it easier to discover new applications.
Michael Davies
The Celsius temperature scale was designed with the assumption that it is the transition temperature between water and ice; however, while ice always melts at 0°C, water does not always freeze at that temperature. Water remains liquid at -40°C, and it is impurities in water that allow ice to freeze at higher temperatures. One of the field’s main goals has been to predict the ability of various materials to promote the formation of ice, which is known as a material’s “ice nucleation ability.”
Researchers at the University of Cambridge, have developed a ‘deep learning’ tool able to predict the ice nucleation ability of different materials – and which was able to beat scientists in an online ‘quiz’ in which they were asked to predict when ice crystals would form.
Deep learning is the process by which artificial intelligence (AI) learns to derive insights from raw data. It discovers its own patterns in the data, removing the need for human intervention and allowing it to process results more quickly and precisely. In the case of IcePic, it can infer various ice crystal formation properties around various materials. IcePic has been trained on thousands of images so that it can examine completely new systems and make accurate predictions from them.
The team created a quiz in which scientists were asked to predict when ice crystals would form in various conditions depicted by 15 different images. These results were then compared to IcePic’s performance. When put to the test, IcePic was far more accurate in determining a material’s ice nucleation ability than over 50 researchers from across the globe. Moreover, it helped identify where humans were going wrong.
Michael Davies, a Ph.D. student in the ICE lab at the Yusuf Hamied Department of Chemistry, Cambridge, and University College London, London, first author of the study, said: “It was fascinating to learn that the images of water we showed IcePic contain enough information to actually predict ice nucleation.
“Despite us — that is, human scientists — having a 75 year head start in terms of the science, IcePic was still able to do something we couldn’t.” Determining the formation of ice has become especially relevant in climate change research.
Water is constantly moving within the Earth and its atmosphere, condensing to form clouds and precipitating as rain and snow. Different foreign particles influence how ice forms in these clouds, such as smoke particles from pollution versus smoke particles from a volcano. Understanding how different conditions affect our cloud systems is critical for making more accurate weather predictions.
“Ice nucleation is extremely important for the atmospheric science community and climate modeling,” Davies said. “Other than direct experiments or costly simulations, there is currently no viable way to predict ice nucleation. IcePic should make it easier to discover new applications.”
