Uncertainties in Global Mapping of Argo Drift Data at the Parking Level

We used Argo float drift data to estimate average ocean currents at 1000 dbar depth from early 2000 to early 2010. Our estimates cover the global oceans, except for marginal seas and ice-covered regions, at a resolution of 1 degree in latitude and longitude. The estimated flow field satisfies the horizontal boundary condition of no flow through the topography, and is in geostrophic balance. We also estimated the uncertainty in the average flow field, which had a typical magnitude of 0.03 ms(-1). The uncertainty is relatively large (>0.03 ms(-1) in both the zonal and meridional directions) near the Equator and in the Southern Ocean. The array bias, which is the bias due to the horizontal gradient in the spatial density of the float data, is generally negligible, with an average magnitude outside the equatorial region of 0.007 ms(-1), becoming relatively large (>0.01 ms(-1)) only near the coastal regions. The measurement uncertainty is assumed to be spatially uniform and includes errors due to the Argos positioning system, internal clock drift, unknown surface drift before submerging or after surfacing, and unknown drifts during ascent and descent between the surface and the parking depth. We found that the overall uncertainty was not sensitive to the assumed value of the measurement uncertainty (epsilon(m))(1/2) when (epsilon(m))(1/2) < 0.01 ms(-1) but it increased with (epsilon(m))(1/2) for (epsilon(m))(1/2) > 0.01 ms(-1).