Origin of the shallow slip deficit on a strike slip fault: Influence of elastic structure, topography, data coverage, and noise

Abstract

Numerous fault slip models for several large strike-slip earthquakes indicate a reduction of slip along the uppermost part of the fault (Shallow Slip Deficit, SSD). In this paper, we investigate the possibility that the SSD is an artifact due to the simplifications made when we model the medium. Using a set of synthetic data for the 1992 Landers earthquake, we show that while neglecting the topography of the Landers area has a moderate impact on the retrieved slip model, neglecting variations in elastic properties, in particular the variation with depth, leads to significant bias on the slip model and can produce 25% of artificial SSD (whereas neglecting the fault damage zone has a smaller impact). We further show that a lack of data coverage around the fault further increases the amount of artificial SSD. Moreover, we find that correlated data noise with maximum amplitude as low as 5% of the maximum surface displacement can also produce significant artificial SSD. We conclude that simple elastic forward modeling in a homogeneous medium may be inappropriate for resolving slip-variability with depth, even in areas of limited topography or lateral variation in elastic properties. Finally, we argue that the estimation of model uncertainty due to data error should be systematic in inversion studies to avoid any misinterpretation of the slip distribution.