Until recently, the lack of seafloor geodetic instrumentation and the use of unrealistically simple, half-space based forward models have resulted in poor resolution of near-trench slip in subduction zone settings. Here, we use a synthetic framework to investigate the impact of topography and geodetic data distribution on coseismic slip estimates in various subduction zone settings. We calculate surface displacements in two synthetic topographic domains that have topography similar to that of Chile and Japan, respectively. We then attempt to image target slip distributions by using a Bayesian approach to solve for slip with two sets of Green's functions—one that accounts for topography and one that does not—and five sets of 50 or more observation points selected from the synthetic surface displacements. Three of these sets of observation points are entirely onland, and two include 5–10 seafloor geodetic sites. We find that the use of topographic Green's functions always improves inferred slip models, and with seafloor geodetic data, it enables an almost perfect recovery of a target slip model, even in the near-trench region. Critically, our results demonstrate that it would be impossible for non-topographic Green's functions to properly recover the true slip distribution, particularly in the near-trench region. We also perform a parameter study with approximately 4,000 slip models estimated using a least-square approach, and find that topographic Green's functions yield significantly more accurate slip models in cases where good data (well distributed and reasonably dense) are available, even in the absence of seafloor geodetic sites.
