Home Environment Industry view: Gwyn Griffiths* on Undersampling the oceans is worse than just lack of data

Industry view: Gwyn Griffiths* on Undersampling the oceans is worse than just lack of data

by admin
Gwyn Griffiths

Gwyn Griffiths

The oceans are undersampled. When we read this truism, too rarely do we reflect on what it means for conclusions drawn from observations.

Join me in a thought experiment. You and I are at Southampton; we have our eyes closed, but open them briefly once a day, yours open at midnight, mine at midday. Conclusions: you live in perpetual darkness, I live in perpetual light. Let’s sample more often, every 22 hours.

We’ll both experience darkness and light, but our ‘day’ length will be wrong. Let’s sample every 12 hours. We’ll get the right number of days in a year. But, we are blind to seasonal changes in lengths of day and night. Perhaps counterintuitively, to identify and describe long-term change, the seasons, we must sample faster. That’s because unequal day and night arise from higher frequency components than sampling twice a day resolve.

We need to apply the Nyquist sampling theorem, and understand the consequences of infringement (aliasing). In the ocean, as in other domains, it comes down to understanding scales and magnitude of variability in time and space. In my experience too few have Nyquist’s theorem in mind when they design ocean observing experiments and when they interpret results and draw conclusions. Put simply, conclusions can be wrong; even in papers in prestigious journals. Nyquist is a tough master, but defensible schemes are possible.

What can we do about this problem? Marine scientists should learn Nyquist’s theorem and what it means for their science. When observing systems are designed, they should ensure that their results lead to unambiguous conclusions that pass scrutiny. An example of a well-designed system is the RAPID-Watch array of physical measurements across the Atlantic at 26˚N. Its results showed the deficiencies in earlier methods and consequent unsupported conclusions.

Here is also a tremendous challenge for technologists: provide new instruments and platforms that alleviate problems of undersampling. Reviewers of scientific papers, and readers, should keep Nyquist in mind when questioning conclusions. Finally, only through international-scale well-directed funding can we hope to properly resolve important processes and change within the oceans.

Gwyn Griffiths, Founder,  Company Autonomous Analytics
Gwyn Griffiths was previously with the National Oceanography Centre, UK

(source: Oceanology International Team via Ocean Space)

You may also like

Leave a Comment