Interpretation of seismic refraction data gives a better understanding of the spatial variations in strength dynamics of subsurface materials in slopes approachi ng failure. However, there are several potential sources of error which need to be understood when considering time-lapse seismic refraction data. These can be grouped into errors associated with: picking first-arrivals, repositioning geophones between surveys, and the changing surface topography of the landslide between surveys. Inclusion of these errors can result in data artifacts after the inversion of the data, and can cause artificial changes in velocity in time-lapse datasets. The seismic refraction dataset was acquired at the Hollin Hill Landslide Observatory, a slowly deforming natural landslide in North Yorkshire, UK. Over a two year period, we undertook 14 surveys to acquire P-and S-wave refraction data from the same profile at the site. We attempt to quantify the error produced in picking the first-arrivals between time-steps. Incorporation of GNSS positional data can provide landslide surface topography, and allows for calculation of true positions of, and distances between, geophones. However, the changes in these data between surveys means that a universal inversion mesh cannot be used on all the data. This can be overcome by sampling data at absolute positions between the inversion meshes of individual time-steps. Using these approaches reduces the errors that can arise when comparing different time-steps. We present some of these time-lapse seismic refraction results, and discuss the implications of this approach to high-resolution spatial monitoring of landslide processes.
