The mean profiles must be as precise as possible, given
that the differences at the crossover points are one of the
error sources in the method. Indeed, in theory, for a given
mission, the difference of mean profiles at a crossover point
should be very close to zero. In reality, the instrumental
noise, the errors in the corrections and the geographically
correlated orbit errors induce differences of a few millimetres,
which are directly introduced in the altimeter bias, due
to the method. For the missions used in this study (Jason-1,
Jason-2 and Envisat), all the differences at the crossover
points between crossing tracks of the same mission are
under 8 mm (and more generally between 2 and 4 mm),
which remains in the usual error bar of the altimeter bias
estimates. The quality control and the editing of the data
and the parameters are consequently crucial to compute
accurate altimeter bias estimates.