Retrieval of the ice sheet topography by satellite radar altimeter echoes

This thesis describes a novel method, multi-channel processing, for measuring the elevation of the continental ice sheets from satellite altimeter echoes. Unlike previous methods, which use a single echo to measure the surface height, the multichannel processing method uses a set of correlated echoes and the entire echo waveform to perform a single measurement. A theoretical framework for the multi-channel processing method is presented, based on linear optimal estimation theory. This theory requires a description of the covariance of the altimeter echoes and the cross-covariance of the altimeter echo and the surface. These covariances and their numerical computation are described. A separable solution is also described for which the computational requirement can be considerably reduced. As part of the solution, a narrow-beam echo is introduced. This is the echo that would result were the altimeter to have a line-beam in the along-track plane. The a priori covariances of the separable estimation are described. The vertical and horizontal scales of the surface undulations, and the number of echoes which are averaged on-board in order to reduce the speckle noise, have effects on the accuracy of the measurement. In addition, the spatial density, coverage, and the duration of the echo which are selected as the input data for the multi-channel processing method can also affect the accuracy of the measurement. All these effects are studied through the calculation of the least RMS error of the optimal estimation, and the computer simulation of the surface retrieval process. By computer simulation, three different scales of undulating features of surface are recovered using the multi-channel processing method, and the results are compared with the conventional retracking method. Three different levels of speckle noise of the echoes are used in the simulation. The results of these two methods are compared. Finally, the feasibility of the multi-channel processing method is confirmed, and the results of the research are summarised and concluded