Media coverage concerning carbon dioxide (CO2) emissions into Earth’s atmosphere most often focuses on how these emissions affect climate and weather patterns. However, atmospheric CO2 is also the primary driver for ocean acidification, because the products of atmospheric CO2 dissolving into seawater reduce seawater’s pH and its concentration of carbonate ions. Since the beginning of the Industrial Revolution, the acidity of the ocean has increased by over 30%.
Some organisms in the ocean may struggle to adapt to increasingly acidified conditions, and even resilient life-forms may have a harder time finding food. Higher CO2 levels in ocean water also make it difficult for shellfish to build their shells and corals to form their reefs, both of which are made of carbonate compounds.
Ocean acidification, which affects the overall health of marine ecosystems as well as societal concerns about food security, has emerged as a major concern for decision-makers on local, regional, and global scales. Indeed, ocean acidification is now a headline climate indicator for the World Meteorological Organization.
Even though the world’s oceans are all connected, the effects of ocean acidification can unfold differently in various regions. The expression of ocean acidification can vary especially in coastal waters, where additional drivers, including both human-caused and natural processes such as nutrient runoff and biological productivity, vary greatly over time and space. Local jurisdictions, nations, and businesses seeking to implement adaptation and mitigation strategies need an accurate understanding, on global and regional scales, of how ocean acidification is progressing.
Time series observations of ocean chemistry collected by researchers are one of the most valuable tools for characterizing how the ocean is changing over time. But currently there are no standardized approaches for assessing and reporting trends from ocean acidification time series. Standardization is necessary to enable comparisons across ocean and coastal systems globally, to create an accurate record that stands the test of time, and to communicate scientific results clearly and consistently to stakeholders and policymakers.
In February 2020, NOAA’s Ocean Acidification Program and the Global Ocean Acidification Observing Network (GOA-ON) brought together a dozen scientists from the international community experienced in collecting data from time series stations scattered throughout the world’s oceans. With their expertise, and with input from scientists in other fields as well, the group began developing a set of best practices for uniformly analyzing long-term ocean acidification trends.
A Need for Standards
Developing accurate assessments of long-term change requires observational data collected using standardized measurement protocols with common reference materials. GOA-ON, established in 2012, has leveraged standard operating procedures for making ocean carbon measurements, most recently updated by Dickson et al.  and available in several languages. Uniform data quality standards and easy access to observational data have matured through several international community efforts, including the GOA-ON Implementation Strategy, the Surface Ocean CO2 Atlas (SOCAT) [Bakker et al., 2016], and the Global Ocean Data Analysis Project (GLODAP) [Olsen et al., 2019].
This scientific cooperation is further bolstered by demand for ocean acidification status and trend data to inform policy and management. For example, the United Nations Sustainable Development Goal 14.3 (UN SDG) for ocean acidity, which aims to support and build capacity to sustainably manage ocean resources, requires accurate measurements of carbon chemistry on a global basis. Some local and national governments are also beginning to focus on ways to assess ocean acidification for managing water quality or natural resources. Addressing these demands will continue to expand the community of researchers making ocean acidification observations.
A focal point of the Ocean Acidification Time-Series Analysis Workshop last February was to review each expert’s approach to trend analysis and hear from additional select experts from outside the ocean acidification community. An important part of our process was to draw upon the decades of work by and experience of researchers in the atmospheric sciences community in uniformly analyzing and reporting trends of atmospheric CO2 and temperature.
Lessons from Atmospheric Science
The workshop highlighted key distinctions between the ocean acidification and atmospheric research communities related to trend analysis and reporting. First, atmospheric measurements of CO2 and temperature are more abundant than measurements of ocean acidity. Globally, there are hundreds of fixed-location atmospheric CO2 stations collecting time series data in a variety of environments. By comparison, there are only a couple dozen stations collecting time series data of ocean acidification, and most of these are in the open ocean. Of the limited number of ocean acidification time series sites in highly variable coastal regions, most were established only in the past 10 years.
Coastal time series data sets require frequent sampling and longer observational records to separate long-term signals from background or periodic (daily, seasonal, and interannual) variability. Both the atmospheric and oceanic CO2-observing communities have addressed this issue by establishing baseline observatories in locations away from the primary sources of CO2 and ocean acidification variability (both natural and human influences) that are found near continents and coasts.
Another distinction between atmospheric and oceanic CO2 measurements is that the marine boundary layer of the atmosphere, where the air meets the ocean’s surface, is more well mixed than the water at the ocean surface. Thus, gaps in atmospheric CO2 data over the ocean are more easily filled through statistical interpolation based on data from surrounding areas. This is not yet possible with ocean acidification data, which are nonuniform and sparse and have uncharacterized variability.
To determine trends in ocean acidification, we are building upon the framework that the atmospheric sciences community has developed over decades of analysis and coordination. CO2 trend analysis rests on several key concepts. These include selecting fixed time series data sets whose collection follows standardized methodologies and that meet community-defined and uniform data quality requirements; determining and removing periodic signals in the time series; and applying consistent procedures for calculating and reporting trends and associated uncertainties [Tsutsumi et al., 2009]. Other important lessons we are drawing from the atmospheric sciences community’s experience involve establishing and publishing transparent protocols and meeting regularly to intercompare and reassess methods.
Factoring in Biogeochemistry
The original charge of the workshop was to determine best practices for delineating the anthropogenic CO2 signal in ocean acidification time series. However, we realized that not all time series are suitable for this purpose, particularly for highly variable coastal data, and that identifying overall trends, which may be caused by multiple drivers aside from anthropogenic CO2, is more realistic. Thus, we pivoted the mandate to focus on best practices for assessing and presenting multidecadal changes.
The multidecadal trend at a given site can be compared with the expected trend assuming ocean chemistry is changing in equilibrium with the global atmospheric CO2 increase (annual decreases of approximately 0.002 in surface ocean pH and 0.008 in surface ocean aragonite saturation state, which is a measure of the availability of dissolved carbonate). Such comparisons emphasize that investigating variability and multidecadal trends can reveal processes affecting ocean carbon chemistry at a given location. For coastal regions, these processes often include increasing atmospheric CO2 in addition to other natural processes and anthropogenic stressors, such as eutrophication, biological productivity, and ocean warming.
Ocean carbon observations are often collected along with data on other biogeochemical parameters, such as dissolved oxygen and nutrients. Extending analyses of multidecadal trends to include these other parameters can provide additional insights into the factors contributing to oceanic change, such as how increased nutrient runoff affects biological productivity and the carbon cycle.
Ultimately, participants at the workshop determined that characterizing ocean carbon and biogeochemical time series trends should involve a sequence of approaches:
- remove periodic signals (e.g., normally occurring seasonal variations) in the detrended time series
- assess linear fit to the data with the periodic signal removed
- estimate the magnitude of the minimum signal that can be detected from the time series
- propagate uncertainty estimates in the reported variability and trends.
These technical steps allow the analyst to remove the noise from the signal and assess whether the time series is long enough or otherwise suitable for detecting a trend with a quantified uncertainty. Each step is equally important to developing uniform trend estimates from all sites.
Presenting Useful Results
To provide stakeholders and scientists with context and actionable information about local processes driving changes in ocean conditions across the observing network, it is important to clearly characterize variability, explain what reported trends represent in space and time, and use metrics relevant for different stakeholder needs. For example, a trend determined at an offshore location may not represent the trend occurring closer inshore where stakeholders are concerned.
Also, some stakeholders (e.g., the persons involved in developing UN SDG 14.3) prefer trends to be presented in terms of changing seawater pH because the public is more familiar with that metric, whereas others may be interested in the relative change in ocean acidity, which is determined from the concentration of hydrogen ions and is not on a logarithmic scale like pH. In addition, some stakeholders and policymakers may be interested less in rates of change over time and more in when potential biological or ecological thresholds might be exceeded. Trends can be extrapolated backward and forward in time to examine and communicate when a threshold is projected to be crossed.
The recommended sequence of approaches for characterizing time series will also be useful in assessing how different observing strategies (e.g., ship-based versus autonomous observing, and directly measured versus calculated ocean acidification parameters) affect trend detection. Sampling frequency, measurement uncertainty, and time series length vary among observing strategies and are all factors that contribute to the ability to detect statistically significant trends. Applying the proposed approaches on observational data or model output may inform new sampling strategies and provide stakeholders with estimates of when trends may be detectable at a given location.
Making Best Practices Better
Through this workshop, we started to outline best practices for uniform analysis of ocean acidification time series. We based these best practices on input from the community of researchers working with ocean biogeochemical time series data and on lessons learned from established statistical techniques and best practices used by the atmospheric sciences community.
The most common approaches for removing periodic signals prior to estimating trends require data sets that can be collected continuously and uniformly (e.g., atmospheric CO2 measurements). But data sets in the ocean acidification community are typically discontinuous and nonuniform, so these common approaches do not work. Thus, we are pursuing additional research to determine the best approach for analyzing trends in ocean acidification time series.
We also plan to partner with statisticians to properly apply statistical approaches in our proposed best practices and gather input from the broader ocean biogeochemical community. We envision that our continuing efforts will result in an open-access paper detailing best practices for analysis of ocean acidification trends, including computational code to implement trend analysis, and the development of a community dedicated to meeting regularly to reassess methods. A consistent methodology will enable comparisons across diverse locations for understanding this global issue. The goal of this work is to make trend detection more accessible to the research community so we can best meet society’s need to manage and adapt to ocean acidification impacts.
This workshop was funded by NOAA’s Ocean Acidification Program and the Global Ocean Acidification Observing Network. We acknowledge the workshop attendees who contributed their expertise to the discussion and provided feedback on this article: Simone Alin, Nick Bates, Wei-Jun Cai, Brendan Carter, Kim Currie, Wiley Evans, Richard A. Feely, Christopher Sabine, Toste Tanhua, and Rik Wanninkhof. We are greatly appreciative of the time our invited speakers offered to provide insight into the standards and statistical approaches taken by the atmospheric community: Pieter Tans of NOAA’s Earth System Research Laboratories and Peter Guttorp of the Norwegian Computing Center and the University of Washington. We also thank additional attendees who provided feedback during a remote session in which recommendations were presented to a broader audience: Dorothee Bakker, Dwight Gledhill, Burke Hales, Peter Landschuster, Nico Lange, Kathy Tedesco, and Bronte Tilbrook. This is PMEL contribution 5106.
Bakker, D. C. E., et al. (2016), A multi-decade record of high-quality fCO2 data in version 3 of the Surface Ocean CO2 Atlas (SOCAT), Earth Syst. Sci. Data, 8, 383–413, https://doi.org/10.5194/essd-2016-15.
Dickson, A. G., C. L. Sabine, and J. R. Christian (Eds.) (2007), Guide to Best Practices for Ocean CO2 Measurements, PICES Spec. Publ. 3, 191 pp., North Pac. Mar. Sci. Organ., Sidney, B. C., Canada, https://cdiac.ess-dive.lbl.gov/ftp/oceans/Handbook_2007/Guide_all_in_one.pdf.
Olsen, A., et al. (2019), GLODAPv2.2019—An update of GLODAPv2, Earth Syst. Sci. Data, 11, 1,437–1,461, https://doi.org/10.5194/essd-11-1437-2019.
Tsutsumi, Y., et al. (2009), Technical report of global analysis method for major greenhouse gases by the World Data Center for Greenhouse Gases, Global Atmos. Watch Rep. 184, WMO/TD-No. 1473, World Meteorol. Organ., Geneva, Switzerland, https://library.wmo.int/index.php?lvl=notice_display&id=12631.
Adrienne Sutton (firstname.lastname@example.org), NOAA Pacific Marine Environmental Laboratory, Seattle, Wash.; and Jan A. Newton, Applied Physics Laboratory, University of Washington, Seattle
Sutton, A.,Newton, J. A. (2020), Reaching consensus on assessments of ocean acidification trends, Eos, 101, https://doi.org/10.1029/2020EO150944. Published on 29 October 2020.
Text not subject to copyright.
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.