It depends on what you're interested in. Admittedly, if you want, say, a global map of all hydrothermal vents, 500-2000m is not going to cut it. But the time and cost required to map the entire ocean with multibeam sonar is huge, on the order of 120 ship-years for the deep oceans and 750 ship-years for the shallow (<500m) continental margins [1].

Global coverage at 500-2000m resolution from satellite gravitometers represents a very cost effective middle ground between that huge effort and what is currently available, which is a patchwork of soundings that has gaps 10s to 100s of km thoughout much of the ocean. (See Fig 2 of [1]).

[1] Becker at al. (2009) http://topex.ucsd.edu/sandwell/publications/124_MG_Becker.pd...