The TOPEX/POSEIDON altimeter measurement system is evaluated for the f
irst 46 repeat cycles (September 23, 1992-December 23, 1993) using tra
cks over the Great Lakes. The temporal variations in lake level are re
moved from the altimeter measurements using in situ lake level measure
ments, thus permitting the performance of the altimeter system to be a
ssessed. For the NASA altimeter, the root-mean-square (RMS) scatter of
the residuals is 3.95 cm using all the tracks over the lakes. However
, some of the scatter in this result is probably due to lake tides or
seiche, which can amount to a few centimeters amplitude near the ends
of the lakes. When the seven best tracks are used, which cross the cen
ter of the lakes where tides/seiche effects are minimal, the RMS error
is reduced to either 2.9 or 3.0 cm, depending on whether the Centre N
ational d'Etudes Spatiales (CNES) or NASA orbit is used. This places a
n upper limit on the error budget of the altimeter system, excluding o
cean tides and inverse barometer effect. There are several short-perio
d variations in the residuals. The most pronounced is a 55-day period,
with a 1-cm amplitude, which we believe is (at least in part) due to
orbit error. When the model-derived wet tropospheric correction is sub
stituted for the TOPEX microwave radiometer correction, the RMS error
increases significantly, possibly resulting in an annual cycle of a fe
w centimeters. Evaluation of the ionospheric correction indicates that
the dual-frequency correction provides an average improvement of 0.85
cm over the Doppler orbitography and radiopositioning integrated by s
atellite (DORIS) correction. Although there are insufficient data to d
irectly assess the CNES altimeter, the relative bias between the altim
eters is estimated to be either -14.3 or -15.6 cm (NASA altimeter meas
uring short), depending on whether the DORIS or dual-frequency ionosph
eric correction is applied to the NASA altimeter.