First, remove the mean.
T/P computes the sea surface height based on the time it takes for a microwave pulse to go from the satellite to the ocean and back again. This time depends, though, on many factors other than the sea surface height: water vapor in the atmosphere, errors in computing the satallite orbit, and on the geoid.

The geoid is the surface of uniform gravitational potential covering the earth. It is not a perfect sphere, but has highs and lows corresponding to oceanic versus continental crust, mountains and undersea ridges, etc.  None of these are relevant to the sea surface height, but the errors in the calculation of the geoid dominate the T/P mean sea surface height.

For this reason, the mean sea surface is always removed from the T/P data before doing anything else. T/P data provides anomalies from the mean state.

Be careful when you remove the mean not to remove variability of interest, such as the annual or the ENSO cycles. It's best to:

(1)    compute the mean over an integral number of years, and
(2)    choose a time period that has no ENSO events, or the same number of El Nino and La Nina events.
 

Problems or questions? cholland at oceansci dot net