Ocean Data Need a Sea Change to Help Navigate the Warming World

Home , Ocean

US COAST GUARD/SCPO NYXOLYNO CANGEMI NYXOLYNO GUARD/SCPO US COAST A sea glider being hoisted onto a ship in the Arctic Ocean after taking various measurements, including of conductivity, temperature and depth. Ocean data need a sea change to help navigate the warming world

Annie Brett, Jim Leape, Mark Abbott & the authors of a High Level Panel for a Sustainable Ocean Economy Blue Paper on Technology, Data and New Models for Sustainably Managing Ocean Resources

Open up, share and network he ocean covers about 70% of Earth’s year’s data. Illegal fishing has proliferated, surface, regulates the climate and is devastating ecosystems and undermining information so that marine home to countless species of fish, global food supplies. stewardship can mitigate a major source of protein for more Happily, new technology platforms col- climate change, overfishing than one billion people. It is now under lected more data on the oceans in 2018 than Tthreat from climate change, overfishing and was gathered during the entire twentieth and pollution. pollution. century1. Data from satellites, autonomous To respond to these threats, those who use, underwater vehicles and other platforms have safeguard and study our seas need real-time come together with emerging data streams information. Too often, ocean management from social media, smartphones and low-cost has been undermined by the lack of data on distributed sensors. This enables a new under- human activity and on the waters themselves. standing of the impact of human activity on the Pirate fishers have plundered the high seas ocean (see ‘Data tsunami’). with impunity, knowing they cannot be traced. For instance, fishing vessels world- Crew members on legitimate fishing boats wide can now be tracked in near-real time have been tortured and even murdered, out using the website Global Fishing Watch of sight. Stocks have been overfished because (https://globalfishingwatch.org). This part- most quotas are set only annually, using last nership between Google, the international

Comment ocean-conservation organization Oceana T T and SkyTruth, an environmental watch- The rapid growth in ocean information Type of instrument: dog in Shepherdstown, West Virginia, uses in the past decade has not been Bottle accompanied by a rethink of how data Bathythermograph satellite data to monitor planetary threats. are collected, shared and accessed. CTD† The partnership combines GPS location data Historical data-management methods Buoy (moored or drifting) prevent a comprehensive understanding Attached to a marine animal from fishing vessels with machine-learning of the impact of human activities Autonomous vehicle analytics. on the ocean. Other Since 2016, it has provided information on 6 activities such as the transfer of fish between intermediate carrier vessels — a technique More data have been collected often used to disguise smuggling. It has also since 2000 than helped to catch boats that illegally dip in and over the previous out of marine protected areas. This service is 100 years possible because of advances in communica- 4 tion, such as 5G technology for mobile-phone networks, as well as improved capabilities in artificial intelligence and machine learning. This and other tools combine data from increasingly robust observation networks worldwide. More than 6,000 floating sensors, 2 satellites and other remote-sensing technol- ogies generate a real-time understanding Simple probes and of ecosystems and the risks they face (see bottles sampled water go.nature.com/3c8jcsc). Connecting disparate data sets can vastly of thousands) (hundreds per year Number of casts* boost our knowledge. For example, finding 0 and combining existing maps of the ocean 1925 1935 1945 1955 1965 1975 1985 1995 2005 2015 floor more than doubled the proportion that *A cast is a set of measurements for a single variable, such as temperature or salinity at dierent depths. †CTD, high-resolution sensor of conductivity, temperature and depth. has been mapped globally — from about 6% in

2014 to 15% in 2019 — without any new surveys DATABASE OCEAN WORLD SOURCE: (see go.nature.com/3gthqno). rarely reached by research vessels. Private Format and quality. Data are often not A major stumbling block to universal data fishing and shipping vessels have reams of interoperable. Inconsistent reporting prac- synthesis is ownership. Petabytes of ocean data information on oceanographic conditions tices, a lack of funding, concerns over sharing are under the control of government agencies, that remains locked away. Silos have serious and a lack of attribution in publications have researchers and private companies, such as consequences: illegal fishers, for example, can had two effects. First, these problems have those in oil and shipping2. This information land their catches unimpeded, knowing that hampered community efforts to create uni- must be made available — fast — to enable sus- nations don’t normally share information on versal standards. Second, they have prevented tainable management of marine resources. vessel identity or routes. Scientists have few the uptake of portals such as the Ocean Data Here we call for two things. First: feder- incentives to expend the effort necessary to Standards and Best Practices Project (www. ated data networks to connect disparate make their data sets available. oceandatastandards.org) or the World Ocean ocean databases. Second: new incentives Database (http://wod.iode.org). and business models for data sharing. These Control. Even when individual data holders can create an open, actionable and equitable realize that their assets might be useful to Fragmentation. Attempts to bring data digital ecosystem for the sustainable future others, they are often reluctant to share together often drive fragmentation — of ocean. The upcoming United Nations Decade them with centralized repositories, because data sets, communities and data norms. of Ocean Science for Sustainable Development Centralized catalogues do increase the (2021–30) must end data segregation and “Vast amounts of scientific visibility of data sets, as happened with the usher in a new era of automated access for all. Intergovernmental Oceanographic Commis- data collected by defence sion’s Ocean Data and Information System Four problems departments worldwide (http://odis.iode.org). They do not always The swelling of ocean information in the remain classified.” solve the problem of access. And the prolifera- past decade has not been accompanied by a tion of these lists makes matters worse. There rethinking of how data are collected, shared are more than 70 overlapping catalogues for and accessed. Historical data-management they want to control how the information is polar ocean data alone1. Time after time, such methods have created a highly fragmented accessed and used. The private sector keeps bespoke solutions have evolved to meet only landscape that is resistant to integration. its data close, fearing competition or public the temporary needs of managers and scien- There are four big problems. scrutiny. For instance, aquaculture farms tists. Scaling up conventional approaches record detailed information on local ocean won’t work. Silos. Government agencies, companies, conditions. They do not share it because of researchers and resource users keep vast concerns over a backlash from environmen- Three fixes stores of data that are collected and man- talists about the effects of their operations Ocean data are dispersed. So are the teams aged for their own specific purposes. These on nutrient levels and other conditions. of experts that must make sense of them. troves are inaccessible and invisible to others. Meanwhile, vast amounts of scientific data These ‘many-to-many’ networks will evolve For example, the US Navy holds extensive collected by defence departments worldwide as collaborations change. Therefore, new data oceanographic data from areas that are remain classified. architectures must enable flexible access,

usage, analysis and cooperation. Here, we committing to full traceability of all tuna for collecting and sharing data. Of course, such outline three key ingredients. products by the end of 2020 (see go.nature. data must be shared in a useful and actionable com/3c4wrpv). They are responding to grow- format, consistent with the standards used by Federated networks. A fast track to interop- ing consumer demand for better information scientists and governments. erability is networking of existing data sets. on the goods they sell and their provenance, Governments should continue to provide Global tagging standards and metadata proto- legality and social and environmental sustain- free access to raw oceanographic data for all cols specify when and how data can be stored, ability. Recognition is growing that better vis- users. Insurance companies, weather forecast- transmitted and used, and by whom. They also ibility across supply chains allows companies ers for precision agriculture, and others use describe the suitability of the data for man- to understand and manage risks. They must raw ocean and climate data to develop lucra- agement and enforcement decisions. These build on these efforts. tive knowledge products. standards support the connection of disparate Researchers must commit to collecting data Payments for these analytical services create repositories through trusted data brokers and using standardized protocols and metadata. opportunities to support research databases they streamline access; data-holders retain They can build on the existing standards of in the long term4. NOAA’s Big Data Program control. Data that meet criteria specified in the ocean-science community, including the provides a model. NOAA has partnered with the tags can be made available automatically US Integrated Ocean Observing System and Amazon and other technology firms to put to users. This provides efficient and timely Ocean Best Practices system (see go.nature. data sets in the cloud so they are freely availa- access for a broad array of managers and users. com/3ebwjtc). Work is needed to integrate ble to the public. Amazon gains insights into Federated networks are used in other these standards and incentivize their adoption. future knowledge-based services by analysing fields, such as health, to help overcome And what of credit? Assigning DOIs (digital data-usage patterns. In return, NOAA has seen confidentiality concerns. In these networks, object identifiers) to data sets can ensure that access to some data more than double, increas- the information itself is not shared — instead, scientists are recognized for contributing to ing their utility at no cost to the government. queries are submitted to gather the needed them. Networks that allow researchers to pub- However, it is risky to rely on public–private information while protecting patient privacy. lish their data in one place with global access partnerships to reduce the costs of network If users are willing to relinquish some control infrastructure. Companies such as Amazon over storage, data lakes of information held in “Sharing must be established could decide that hosting ocean data is no a raw format can be nodes in larger federated longer economically viable, or a firm might networks. Data lakes move unstructured data as a new default unless there disappear from the market. Moreover, such onto cloud architecture. This improves access are compelling constraints.” partnerships can be significantly influenced and lowers the costs of analysis. Data lakes can by financial and political pressures beyond also enable the development of services. For the scientific and stewardship needs of the instance, when the US National Oceanic and will remove many of the other logistical barriers data. Public entities can protect against these Atmospheric Administration (NOAA) pub- to sharing. Embargo windows give researchers eventualities by partnering with a variety of lished data from its Next Generation Weather time to publish scientific findings before the companies, and by ensuring that data remain Radar in the cloud, the information was used data are made available to all. For example, backed up on publicly funded servers. to analyse and track bird migrations3. the UK Natural Environment Research Council Global coordination and commitment is The UN Educational, Scientific and Cultural requires those it funds to publish all data within needed. The UN Decade of Ocean Science for Organization (UNESCO) should lead an effort two years of the end of data collection. Sustainable Development is an opportunity to develop standards for tagging and meta- to revolutionize how ocean data are collected, data. The organization should require that Business models. The provision of ocean data is stored and used. all data collected use these protocols and are an important public good, supporting sustain- made publicly available. able management globally. However, the costs associated with standardizing and disseminat- The authors Open data. Sharing must be established as a ing data are high: about 10–20% of the budget new default unless there are compelling secu- of oceanographic research projects. Most exist- Annie Brett is a fellow at the Center for rity, proprietary or privacy constraints. New ing research databases rely on public funding. Ocean Solutions, Stanford, California; and standards could allow industry and military There is an urgent need for new revenue models at the World Economic Forum Centre for the data holders to define data tags that make that can make data more broadly available. This Fourth Industrial Revolution, San Francisco, robust, long-term data sets automatically requires innovation in business models. California, USA. Jim Leape is a fellow at and available after any embargoes have expired. As private data sources proliferate, investors, co-director of the Center for Ocean Solutions Governments must lead the way by aggres- philanthropic organizations and governments at the Stanford Woods Institute for the sively declassifying and sharing data that are should invest in approaches that combine com- Environment, Stanford, California, USA. Mark relevant to ocean science and management. mercial viability with support for data manage- Abbott is president and director of Woods The technical obstacles can now be over- ment. For example, private satellite and drone Hole Oceanographic Institution, Woods Hole, come, as we have set out here. Companies and providers already sell data to governments Massachusetts, USA. A full list of the authors researchers can be incentivized to share data and large companies. They can make these of a High Level Panel for a Sustainable Ocean by making it a condition for access to public data available in delayed or slightly degraded Economy Blue Paper on Technology, Data and resources, such as funds for ocean research, form to coastal communities and developing New Models for Sustainably Managing Ocean permits for coastal development or licences countries that can’t afford to pay for them. Resources accompanies this Comment online for oil exploration or fishing. Businesses should also find ways to incen- (see go.nature.com/3cvtczd). Companies that sell or process fish and tivize better data collection throughout their e-mail: [email protected] seafood are already moving to introduce supply chains. Thai Union, a major seafood com- 1. Tanhua, T. et al. Front. Mar. Sci. 6, 440 (2019). greater transparency into their supply chains. pany in Samutsakorn, Thailand, for instance, is 2. Pendleton, L. H. et al. ICES J. Mar. Sci. 76, 1415–1423 (2019). For instance, 65 major retailers, processors, piloting the use of a blockchain ledger with a 3. Vance, T. C. et al. Front. Mar. Sci. 6, 211 (2019). 4. Organisation for Economic Co-operation and marketers, traders and harvesters signed the technology company called Fishcoin to reward Development. Business Models for Sustainable Research Tuna 2020 Traceability Declaration in 2018, small-scale fishers in mobile-phone minutes Data Repositories (OECD, 2017).