Interferometric aperture synthesis is an inverse problem that attempts to form an elevation map of the earth (in the case of radar) or a bathymetric map of the seafloor (in the case of sonar). In both cases, a pair of (nominally) vertically displaced transducers is configured as an interferometer. After aperture synthesis is performed to produce a pair of images, the height of each resolvable scatterer can be estimated using time delay estimation between the image pairs and knowledge of the system geometry. While interferometric synthetic aperture sonar (InSAS) seems like an obvious extension of the methods of interferometric synthetic aperture radar (InSAR), the height estimation algorithms are surprisingly different. In this paper we start with the principle of generalised correlation for optimal time delay estimation. This filters the signals to maximise their coherence since the accuracy of the time delay estimates, and thus the height estimates, strongly depends upon the signal coherence. We then consider the fundamental differences between InSAR and InSAS; namely the relative signal bandwidth, aperture sampling rate, and geometry and show how application of generalised correlation time delay estimation leads to the differences in how InSAS and InSAR signals are processed.