What the altimeter actually measures is the average waveform of thousands of returned pulses as a function of time. These short temporal averages are further averaged over 1-intervals for T/P, resulting in the basic 1-Hz data used by most T/P investigators. Since the satellite is moving at nearly 7 km/s, the T/P range measurement is actually an along-track average for approximately 6-7 km. Other satellite altimeters have slightly different averaging periods and along-track distances, but these values do not change much from one altimeter to another. Later during ground processing, the waveform is processed to yield the time of pulse arrival (usually taken as the midpoint of the leading edge of the waveform), the significant wave height (from the slope of the leading edge of the waveform), the radar backscatter coefficient (the amplitude of the returned waveform), and many other measurement variables. If the altimeter is dual-frequency (such as on TOPEX/POSEIDON, and the future Jason-1 and ENVISAT), then this procedure is done for measurements collected on both frequencies. For a more complete discussion of these basic measurement issues, the reader is referred to Stewart (1985).
Putting aside for now altimeter path length corrections due to the atmosphere, the precision of the measurement of the time delay is primarily limited by the number of altimeter pulses averaged in the basic time interval, 1 s for T/P, or equivalently in the basic along-track averaging length, 6-7 km for T/P. Averaging for longer periods results in more precise estimates, but results in less spatial resolution and introduces errors due to short length-scale sea surface height variability that is incorrectly averaged together into a single height estimate. Of course, accuracy is a somewhat different issue than precision. To obtain the time delay, the altimeter relies on an internal time reference, or a clock. If the clock is drifting, then even a precise measurement will be biased high or low. Such clock errors are a primary concern in the range measurement, because these errors can lead to low-frequency drifts in the range measurements.
In the design of the T/P altimeter (and most of its predecessors), it was realized that drifts in the timing system could be a serious source of error and an internal calibration system was included in the system (Hayne et al., 1994). Estimates of the range stability—actually the range drift—are routinely produced by the T/P project and corrections applied to the final altimetric data provided by the project. Typically, these corrections are very low frequency in nature and have a peak-to-peak range during the T/P mission of order 10 mm. This internal calibration, however, has been brought in to question (Chambers et al., 1998), and whether it is functioning as designed should still be considered an open issue. However, we know of no compelling reason to suspect the basic range measurement, (i.e., the basic time delay measurement) to be subject to systematic drift errors.
While we will later describe a number of atmospheric and environmental corrections to the range measurement given in Eq. (1), these are not necessary to understand how satellite altimeters can measure sea level. Note that the altimeter range measurement is just the altitude of the satellite above the ocean surface. It is not a measurement of sea level, but sea level can be computed with accompanying knowledge of the orbital position of the satellite.
Was this article helpful?