A Monte Carlo Method to Find Mass Flux Variability in the North Atlantic and Southern Oceans

Using the concept of transit-time distributions (TTDs), we estimate the volume transports and spreading rates of Labrador Sea Water (LSW), Antarctic Intermediate Water (AAIW), and North Atlantic Deep Water (NADW) through different repeat hydrographic-sections of the ocean with observations of transient tracers and model output from the National Center for Atmospheric Research (NCAR) high resolution Parallel Ocean Program (POP) model. We estimate the TTDs using inverse Gaussians (IGs) that are observationally informed from pCFC and statistically estimated Helium-3/Tritium ages. The Helium-3/Tritium ages are estimated using a nonparametric statistical model that works well with sparse observations and quantifies uncertainty associated with mapping. In areas where water masses of vastly different ages mix, this representation is extended to a mixture of two IGs. Using the NCAR POP model TTDs as priors, we arrive at a Bayesian estimate of a multiple peaked TTD and estimate the mean ages and Peclet number at each location and depth. We find that once a mean age of about 50 years is surpassed, statistically estimated Helium-3/Tritium serves to add more information than CFCs would have alone, but the uncertainties on the IG parameters below the thermocline can often be large enough to not be able to distinguish the Peclet number from zero. We also find smaller average spreading rates than those inferred from previous studies with intra-annual variability larger for NADW than AAIW and LSW with the smallest, assuming steady-state for each snapshot.