Josh Lanham, a Postdoctoral Research Associate based at ICCS, and Laura Cimoli, Assistant Research Professor, are lead author and co-author of a recent paper published in Nature Communications Earth & Environment. get the inside story on how they inferred that deep-ocean heat has been marching closer to Antarctica...
Antarctica is losing ice at an increasing rate, driven largely by heat from the ocean. Beneath the surface of the Southern Ocean sits a body of relatively warm water known as Circumpolar Deep Water (CDW). CDW circulates at depth around the continent. When it gets close enough to Antarctica, it can intrude into ice shelf cavities, and melt the terminating glaciers from below. This has important implications for global sea level rise, as Antarctic ice sheets lock ice which could increase global sea level of several meters. Tracking how this warm water moves is therefore important for global climate change and adaptability.
Our new paper in Communications Earth & Environment shows that CDW has been shifting towards Antarctica over the past two decades. More warm water near the shelf means more basal melting, with direct implications for sea level rise.
Getting the data
The Southern Ocean is not an easy place to work. The most comprehensive ocean measurements come from research ships that use a combination of instruments to measure the physical and chemical properties of the water column. This typically includes high-quality measurements of temperature, salinity, oxygen, and nutrients. To infer climate variability and changes, these surveys must be repeated several times. This effort is coordinated internationally through the Global Ocean Ship-based Hydrographic Investigation Program (GO-SHIP), and a given ship transect across the Southern Ocean might be repeated once per decade. The dataset we used draws on decades of these cruises, each representing weeks at sea.
The gaps between ship surveys are filled by Argo floats. These are autonomous instruments that drift through the ocean, periodically descending to 2000m and rising to the surface, transmitting in near-real time temperature and salinity profiles. There are roughly 4,000 floats operating at any one time, globally, since the early 2000s. Ships provide the physical and chemical full water column detail needed to identify water masses; Argo provides the spatial and temporal coverage needed to track how they change. The challenge is combining the two.
Bridging the gap with machine learning
The difficulty is that ships and Argo floats measure different things. The water mass classification we use to robustly identify CDW requires chemical tracers (i.e. oxygen, nutrients) that core Argo floats (usually) don't carry.
To extend our ship-based analysis across the full Argo dataset, we trained a random forest model ensemble on water mass classifications derived from the ship data, then applied it to Argo data. The trained model identifies water masses from temperature, salinity, and position alone, without needing chemical measurements directly. This gives us a 20-year, monthly, basin-wide water mass classification. This record is of sufficient resolution in space and time to decompose the water mass variability and identify the dominant modes of change.
What we find
Both analyses (ships and Argo) tell the same story. CDW has expanded in the upper 2 km of the water column close to the Antarctic continent and contracted further north. The pattern holds across almost all longitudes and is the dominant mode of variability in the Argo record. Near the continent, the CDW expansion closely mirrors a contraction in cold, dense waters that normally limit warm water access to the shelf.
What's driving this poleward expansion of warm water isn't fully resolved. Strengthening and poleward migration of the Southern Ocean's westerly winds is one possibility. The ongoing contraction of Antarctic Bottom Water is another, as it may be allowing CDW to move into the space it previously occupied. Both mechanisms are plausible and separating them is a question for future work.
Read the paper here: Poleward migration of warm Circumpolar Deep Water towards Antarctica