five billion tonnes

of carbon, unmeasured

Problem statement

The ocean's biological carbon pump transports 5–10 Gt C/year from the surface to depth, sequestering CO2 from the atmosphere for decades to millennia. But we cannot accurately measure this flux. Recent work shows that century-scale carbon sequestration (0.9–2.6 Pg C/yr) may be up to 6x larger than previous estimates because significant sequestration occurs above 1,000m — a depth zone that models systematically ignore. The organic composition of carbon pools below 500m is largely unknown. The field still relies on the Martin curve, a 1987 empirical power-law approximation for flux attenuation with depth, despite knowing it oversimplifies diverse ocean provinces.

Why this matters

The biological pump is one of the largest active carbon fluxes in the Earth system. Getting its magnitude and climate sensitivity wrong propagates directly into carbon budget calculations and emissions pathway projections. If warming weakens the pump (through temperature-dependent recycling and stratification), a positive feedback could accelerate climate change. If the pump is stronger than currently modeled, more carbon is already being sequestered than we account for. Earth system models that assess flux at a fixed reference depth systematically underestimate pump efficiency in regions with shallow euphotic zones and overestimate it in deep ones — meaning regional projections are unreliable.

What’s been tried and why it hasn’t worked

Moored sediment traps — the primary direct measurement tool — disrupt fragile marine particle aggregates during collection, creating persistent measurement artifacts that have never been fully resolved. Diel vertical migration by zooplankton and fish creates a major carbon transport pathway that is nearly impossible to quantify with nets because organisms actively avoid capture. Three distinct export pathways (gravitational sinking, migrant transport, and physical mixing pumps) are regulated by completely different mechanisms and have never been measured simultaneously at the same location. Satellite remote sensing observes only surface ocean properties, requiring uncertain chains of proxy relationships to estimate subsurface fluxes. The Martin curve parameterization, used in most climate models, does not account for regional differences in particle composition, microbial activity, or food web structure that control how much carbon actually reaches sequestration depth.

What would unlock progress

Autonomous profiling floats (like the GO-BGC Array) that measure biogeochemical properties from surface to 2,000m throughout the global ocean, providing the first spatially and temporally resolved view of subsurface carbon fluxes. New sensors for in-situ measurement of particle flux, composition, and transformation at depth without the artifacts of sediment traps. Machine learning approaches to integrate satellite, float, and ship-based data into unified flux estimates. Acoustic and optical methods for quantifying diel vertical migration without relying on net catches.

Entry points for student teams

A student team could analyze publicly available BGC-Argo float data (biogeochemical profiles of oxygen, nitrate, pH, chlorophyll) to estimate biological carbon export in a specific ocean basin and compare their estimates to satellite-derived predictions, quantifying the discrepancy. Alternatively, a team could prototype a low-cost particle imaging system for deployment on moorings or floats to characterize sinking particles without the disruption of physical collection. Relevant skills: oceanography, biogeochemistry, sensor design, image processing, data science.