A statistical funnel-based approach to assess spatial contributions to uncertainty in organic carbon flux attenuation.

Abstract

The biological carbon pump (BCP) is one of the atmospheric CO2 pathways into the ocean and plays a major role in climate regulation. Primary production in the ocean uptakes CO2 to produce organic matter, which enters the food chain and undergoes transformations to become particulate organic carbon (POC). The POC sinks transporting carbon into the ocean, creating a POC flux. As it sinks, POC gets respired and releases CO2 back into the environment, attenuating the flux. The deeper this happens, the longer the released CO2 stays away from the atmosphere. Sediment traps and the Martin’s curve, a statistical fit of a power law, are commonly used to describe this flux attenuation with depth. The use of sediment traps assumes the flux captured by traps deployed simultaneously at different depths comes from the same region. The source of the flux plays a key role in understanding observations of the BCP. Unusual/unexpected flux profiles might be caused by variability in fields on the ocean’s surface. To know the source region of the flux, one must find its “statistical funnel”. A particle backtracking approach can be used to achieve this; however, the currently used approach does not account for uncertainties in ocean currents. Additionally, the existing method for the statistical funnel backtracks particles until they reach the ocean’s surface. This is not necessarily accurate as the surface layer of the ocean is well mixed. Particles in the mixed surface layer does not make sense as particles could come from any point within the mixed layer.
Here I suggest a method to calculate a statistical funnel considering uncertainties in ocean currents. From this funnel study I also suggest a method to estimate the contributions of spatial variability in the traps’ catchment areas to the uncertainty in Martin’s “b-value” (which represents the flux attenuation). The method also suggests a spatially adjusted b-value, which considers each trap’s catchment area. Finally, I present a case study that found that spatial variability can contribute up to 25% in the uncertainty in a b-value obtained from POC flux observation.

More information

Identifiers

ORCID for Benoit Henri Guillaume Espinola: orcid.org/0000-0003-2412-9645

ORCID for Anna Hickman: orcid.org/0000-0002-2774-3934

Catalogue record

Export record