Where and how long ago was water in the western North Atlantic ventilated? Maximum entropy inversions of bottle data from WOCE line A20

A maximum entropy (ME) method is used to deconvolve tracer data for the joint distribution G of locations and times since last ventilation. The deconvolutions utilize World Ocean Circulation Experiment line A20 repeat hydrography for CFC-11, potential temperature, salinity, oxygen, and phosphate, as well as Global Ocean Data Analysis Project (GLODAP) radiocarbon data, combined with surface boundary conditions derived from the atmospheric history of CFC-11 and the World Ocean Atlas 2005 and GLODAP databases. Because of the limited number of available tracers the deconvolutions are highly underdetermined, leading to large entropic uncertainties, which are quantified using the information entropy of G relative to a prior distribution. Additional uncertainties resulting from data sparsity are estimated using a Monte Carlo approach and found to be of secondary importance. The ME deconvolutions objectively identify key water mass formation regions and quantify the local fraction of water of age tau or older last ventilated in each region. Ideal mean age and radiocarbon age are also estimated but found to have large entropic uncertainties that can be attributed to uncertainties in the partitioning of a given water parcel according to where it was last ventilated. Labrador/Irminger seawater (L water) is determined to be mostly less than similar to 40 a old in the vicinity of the deep western boundary current (DWBC) at the northern end of A20 but several decades older where the DWBC recrosses the section further south, pointing to the importance of mixing via a multitude of eddy-diffusive paths. Overflow water lies primarily below L water with young waters (tau less than or similar to 40 a) at middepth in the northern part of A20 and waters as old as similar to 600 a below similar to 3500 m.