ABSTRACTS

Oral Session 3 | Tuesday, October 4, 10:00–10:20 | Abstract 505

Detecting trends in 20 years of ocean colour

Ocean ecosystems are thought to be changing with climate. Detecting such changes is difficult over much of the ocean, though, because it is well-established that a few decades of measurements are required to detect a trend in satellite chlorophyll, whereas the satellite ocean color record is only two decades long. Remote sensing reflectance (Rrs), however, is measured at multiple wavelengths, meaning it may be possible to detect climatic trends over shorter time intervals in ocean color, because the overall signal-to-noise ratio is higher than for a single variable (i.e. chlorphyll). Here we use a multivariate regression technique that accounts for autocorrelation and correlations between Rrs wavelengths to detect trends in ocean color from the satellite record. We show that while satellite-derived chlorophyll has only changed significantly over a small fraction of the ocean in the past 20 years, ocean color has changed significantly over much of the ocean. The largest contiguous region where ocean color has changed is in the subtropical Pacific. Using a marine ecosystem model that calculates Rrs, we show that 20 years is sufficiently long to detect a climatic trend in Rrs, meaning these trends are plausibly due to climate change rather than decadal variability in ocean color. We discuss the ecological and biogeochemical implications of the observed color changes and the implications of our findings for using Rrs directly as a sentinel of surface ocean ecosystem change, rather than single-variable products derived from Rrs.

B.B. Cael, National Oceanography Centre

Kelsey Bisson, Oregon State University

Stephanie Dutkiewicz, Massachusetts Institute of Technology

Stephanie Henson, National Oceanography Centre