Trend signals from the oceans

October 29, 2014 - by Simone Ulmer

Between 2012 and 2013, the concentration of carbon dioxide (CO2) in the atmosphere rose more sharply within one year than it had done since 1989. According to the World Meteorological Organisation’s Greenhouse Bulletin, published on 9 September 2014, the CO2 concentration rose by 2.9 parts per million (ppm) over this period, which brought the total atmospheric concentration to 396 ppm – 40 per cent higher than in the pre-industrial era (some 250 years ago). Climate researchers are largely in agreement that humans are contributing to global warming through emissions of greenhouse gases – primarily CO2. According to the UN climate report of 2013, there is a 95% probability that humans are responsible for at least half of the temperature increase since 1950. Now, using the high-performance computers at CSCS, climate scientists at the University of Bern have analysed how long it takes until the impacts of anthropogenic CO2 emissions are distinguishable from natural variations in the oceanic carbon cycle.

CO2 acts rapidly

The researchers’ simulations demonstrate that certain signals of increasing supplies of CO2 into the oceans – such as increases in the partial pressure of CO2 and decreases in pH value – can emerge from local background noise within approximately 10 to 20 years. In contrast, the trend towards warmer temperatures in sea surface waters becomes apparent at the local scale only after 45 to 90 years. “The results correspond well to the long-term CO2 and pH measurements from the few observation stations that we have in the open seas,” says Fortunat Joos, co-author of the study, which was recently published in Biogeoscience.

For their study, the researchers used an ensemble of 17 Earth system models, whose simulations cover the climate development for the period from 1870 to 1999. They used these data to quantify the time span required for the “anthropogenic signals” to be distinguishable from natural variability in the surface ocean and for different locations worldwide. Natural variations occur in the carbon cycle-climate system due to, for example, severe volcanic eruptions or so-called internal modes such as the North Atlantic Oscillation or the El Niño-Southern Oscillation, which go hand-in-hand with characteristic large-scale changes in oceanic currents and meteorological conditions. These natural variations can intensify or even mask a signal.

As a first step, the researchers determined for each variable of interest the magnitude of the year-to-year natural variability by computing its standard deviation from the model output. Second, this variability is then compared with the mean trend over the period from 1970 to 1999. They require for trend detection that a trend signal is at least twice as large as the natural variability. In other words, the time at which the particular signal is twice as strong as the background noise is the point at which a trend emerges.

Time of Emergence in years for pH. Values represent the ensemble means from seventeen Earth System Models and are for the ocean surface.(Image: Kathrin Keller)

An illustration how the Time of Emergence (ToE) is computed from the output of the Community Earth System Model (CESM). A smoothing spline (dark blue) and a linear trend (red) over the period from 1970 to 1999 (indicated by thin-dashed lines) is computed from annually-averaged pH data (light blue) at 0°N, 145°W. The grey bar represents two times the standard deviation of the annually-averaged data from the spline and is taken as a measure of natural variability. The intersect between the red vertical line and the upper border of the grey bar at year 2010 shows when the trend leaves the envelope of background variability and, from then on, is detectable. Consequently, the ToE at this location is 11 yr (2010–1999). (Image: Kathrin Keller)

In their study, the climate researchers write that biochemical variables, such as the partial pressure of CO2 and the pH value, would exhibit similar spatial patterns to those of the level of inorganic carbon dissolved in the water. Despite this, however, their local trends emerge from the background noise faster, after about 10 to 20 years in most regions, than is the case for inorganic carbon, for which it typically takes between 10 and 30 years to detect a clear signal on the local scale. On the assumption that the natural variations are constant, they suggest that future trends, which will presumably be stronger, will emerge even faster.

Joos emphasises that the study shows once again how important it is to measure biogeochemical variables of this kind, as they are sensitive indicators that processes such as a falling pH value are occurring. Moreover, the study demonstrates that to distinguish between natural and anthropogenic signals requires regular measurements over very long periods of time – and this necessitates the development of more continuously operated measuring stations in our oceans.

Conservative estimate

“CO2 emissions have more than doubled since the 1970s and we have now reached a new record,” says Joos. Therefore, he says, the trends identified in this study covering the period from 1970 to 2000 are conservative estimates. Nevertheless, the results show that, in particular, the pH value stands out from the natural variations in the carbon cycle relatively quickly. Consequently, he says, there is a risk of this having adverse effects on marine ecosystems – for example, for coral reefs or fish whose ossicles are made of calcium carbonate. “CO2 remains in the atmosphere and the ocean for millennia, causing lasting disruptions to the climate system. If we want to limit man-made climate change and ocean acidification, we must reduce CO2 emissions,” says Joos.

Reference

Keller KM, Joos F & Raible CC: Time of emergence of trends in ocean biogeochemistry, Biogeosciences (2014), 11, 3647-3659, doi: 10.5194/bg-11-3647-2014.