Massive ice sheets with layers built up over millions of years blanket most of Greenland and Antarctica. As a result of climate change, these ice sheets have begun to melt and shrink. Scientists believe this trend is likely to continue and will contribute to sea level rise for decades to come.
Despite these predictions, significant uncertainties plague computer models that simulate the behavior of ice sheets and how they interact with the global climate. To enhance the credibility of ice sheet models, Kennedy et al. are developing an open-source software package called the Land Ice Verification and Validation toolkit (LIVVkit) that analyzes models for correctness and performance.
LIVVkit employs two sets of techniques that are widely used in a variety of other fields. It uses verification techniques to test the construction of an ice sheet model; for example, it tests whether numerical algorithms are correctly implemented in the model’s computer code. The second technique LIVVkit employs—validation—tests a model’s performance and gauges how accurately it reproduces real ice sheet phenomena. The software essentially asks, “Is the model doing what it says it’s doing, and how well is it doing it?”
Kennedy’s team is developing LIVVkit alongside the second version of the Community Ice Sheet Model (CISM 2.0), which helps predict ice sheet changes and sea level rise. CISM 2.0 developers follow a standard procedure to test new features and code using LIVVkit, and the testing results can be shared with collaborators or the public.
Ultimately, the developers of LIVVkit intend it to be used by anyone working on new ice sheet models. They have focused their efforts on making it easy to use and flexible enough to be applied to a variety of model designs, use cases, and computing platforms. New versions of LIVVkit are regularly released to the public via GitHub.
The team has also designed LIVVkit to return testing results in a straightforward format intended to be easy to share with the wider scientific community, as well as with policy makers. Over time, the team hopes that LIVVkit will foster widespread confidence in ice sheet models by illuminating their performance throughout the model development process. (Journal of Advances in Modeling Earth Systems (JAMES), https://doi.org/10.1002/2017MS000916, 2017)
—Sarah Stanley, Freelance Writer