Can we constrain geographical variability in the biological carbon pump's transfer efficiency from observations?