Testing ocean software

What is it about?

In the current computing age, various software systems have been developed to support the ocean modelling and computations, which help our scientists simulate, model and predict ocean conditions, and thus assist the relevant industries and human activities as well as contribute to advance other disciplines. The process of testing ocean software is difficult due to a so-called "oracle problem." In a word, there is a lack of ways to test the ocean simulation and analysis software because, first and foremost, the nature of ocean is made up of many variables that interact in complex ways. Tidal signals, for instance, in the ocean can be governed by these variables, making it hard to know the true tidal patterns. Another reason is that the computed results are prone to accumulation of roundoff errors. Such errors can partly contribute to the so-called "butterfly effect" in deterministic nonlinear systems, for example, weather forecasting systems. In this paper, we're proposing a solution to this problem. We suggest using a simple yet effective testing method called metamorphic testing (MT) to test tidal analysis and prediction (TAP) software. We've found a way to make this testing method by using the repeating patterns of tides, and we were able to reveal a real issue with open-source TAP software using our method. We also did some experiments evaluate the performance of MT.

Featured Image

Photo by Marc Thunis on Unsplash

Photo by Marc Thunis on Unsplash

Why is it important?

Our Earth is actually a world of water as over 70% of its surface is covered by the ocean. Ocean science, also being referred to as marine science or oceanography, is an important discipline of the Earth sciences, covers a wide range of topics, from ocean circulation, waves, storm surge, sea level rise, heat content, water transport to ecosystem dynamics, ocean acidification and plate tectonics of the sea bed. Despite their importance, making sure ocean (and to a large extent, the Earth) software systems trustworthy is a big challenge that has been around for a long time. Our research shows that this method could also be used to test more complex ocean software systems.