A New Improved Estimation of Agulhas Leakage Using Observations and Simulations of Lagrangian Floats and Drifters

Abstract A new estimate of Agulhas leakage transport is calculated using profiling floats and drifters. Since Richardson's seminal estimate of 15 Sv in 2007, the number of floats and drifters passing through the Agulhas Current has quadrupled. Within uncertainties we find the same leakage percentages as Richardson, with 34% of drifters leaking at the surface and 21% of floats leaking at 1,000 m depth. We find that the drifters tend to follow a northward leakage pathway via the Benguela Current compared to the northwestward leakage pathway of the floats along the Agulhas Ring corridor. We simulate the isobaric and profiling behavior of the floats and drifters using two high resolution models and two offline Lagrangian tracking tools, quantifying for the first time the sampling biases associated with the observations. We find that the isobaric bias cannot be robustly simulated but likely causes an underestimate of observed leakage by one or two Sverdrups. The profiling behavior of the floats causes no significant bias in the leakage. Fitting a simulated vertical leakage profile to the observed leakage percentages from the floats and drifters and using the mean Agulhas transport observed by a moored array at 34°S we find an improved Agulhas leakage transport of 21.3 Sv, with an estimated error of 4.7 Sv. Our new leakage transport is higher primarily because we account for leakage at depths down to 2,000 m, while Richardson considered only the top 1,000 m of the water column.