A three-dimensional mapping of the ocean based on environmental data

Abstract

Oceanographers have long recognized and described persistent, large, water masses which partition the global ocean into chemically and physically distinct volumetric regions. We constructed a regularly spaced ocean point mesh grid from sea surface to seafloor, and attributed these points with the 2013 World Ocean Atlas (WOA) dataset, version 2, 57 year average values for six physical and chemical environment parameters (temperature, salinity, dissolved oxygen, nitrate, phosphate, and silicate). The database of over 52 million points represented the global ocean in x, y, and z dimensions. These points were statistically clustered to define 37 distinct volumetric units, here called Ecological Marine Units (EMUs). The EMUs represent physically and chemically distinct water volumes based on spatial variation in the six marine environmental characteristics used. Twenty two of the 37 EMUs are globally or regionally extensive, and accounted for 99% of the ocean volume, while the remaining 15 were smaller and shallower, and occurred around coastal features. We characterized the horizontal and vertical dimensions of EMUs and mapped distinct marine regions of varying size and depth. We found vertical separation into three broad depth zones, and general spatial correspondence with the major global water masses. The EMUs are an open access resource, and are intended to be useful for disturbance assessments, ecosystem accounting exercises, conservation priority setting and marine protected area network design, and other research and management applications.

Mapping near surface global marine ecosystems through cluster analysis of environmental data

“…In addition, it maximized objectivity and repeatability by using PCA and similarity analyses, and mapped at a finer spatial resolution to date (5 arc‐minutes = 9.2 km at the equator). The results align well spatially with the previous objective classifications, notably the EMUs (Sayre, Dangermond, et al, ; Sayre, Wright, et al, ). The outputs are publicly available at Figshare for use in future research and environmental management.…”

“…Sayre, Dangermond, et al (); Sayre, Wright, et al, () used the pseudo F ‐statistic to decide on the number of EMUs. They chose 37 but arguably 17 or 28 could have been chosen as the number of EMUs.…”

“…Both different underlying data and methods may result in different classifications, suggesting the relations between them. Thus, to explore the generality of the ecosystems defined in the present study we compared them to three contrasting global marine classifications: the EMUs (Sayre, Dangermond, et al, ; Sayre, Wright, et al, ) based on k ‐means clustering analysis by Euclidean Distance using temperature, salinity, oxygen, nitrate, phosphate, and silicate as variables; biogeographic realms (Realms) defined by cluster analysis of 65,000 marine species distributions (Costello et al, ); and biogeochemical provinces (Provinces) defined by expert interpretation of ocean color and other pelagic information (Longhurst, ).…”

“…Thus, despite appearances, only two fully global classifications of the ocean based on quantitative environmental data analysis have been published. Oliver and Irwin () applied a method, which used hierarchical and k ‐means clustering on sea surface temperature and radiance to map 81 “provinces.” Sayre, Dangermond, et al (); Sayre, Wright, et al, () used k ‐means clustering on six variables (temperature, salinity, oxygen, nitrate, phosphate, and silicate) to a depth of 5,500 m. They distinguished 37 three‐dimensional regions named “Ecological Marine Units” (EMUs), of which 22 comprised 99% of the ocean volume. However, there are 20 variables now available for ocean surface waters at a global scale (Basher, Bowden, & Costello, ).…”

3D spatial conservation prioritisation: Accounting for depth in marine environments

“…There are also important advances in the amount of global marine environmental data with a depth component, through the World Ocean Atlas (Boyer et al, 2013). All this information with a 3D component can be used to create species distribution models (Duffy & Chown, 2017) and ecological units (Sayre et al, 2017) in 3D, although this is still not common practice. Scientists are now able to better map cumulative anthropogenic impacts in the ocean, estimate whether these impacts are increasing or decreasing (Halpern et al, 2015), and how they can affect different stakeholders .…”