@phdthesis{Liebs2014,
  author    = {Liebs, G{\"o}ran},
  title     = {Ground penetration radar wave velocities and their uncertainties},
  doi       = {10.25932/publishup-43680},
  url       = {http://nbn-resolving.de/urn:nbn:de:kobv:517-opus4-436807},
  school      = {Universit{\"a}t Potsdam},
  pages     = {ii, 106},
  year      = {2014},
  abstract  = {We develop three new approaches for ground penetration wave velocity calcultaions. The first is based on linear moveout spectra to find the optimum ground wave velocity including uncertainties from multi-offset data gathers. We used synthetic data to illustrate the principles of the method and to investigate uncertainties in ground wave velocity estimates. To demonstrate the applicability of the approach to real data, we analyzed GPR data sets recorded at field sites in Canada over an annual cycle from Steelman \& Endres [2010]. The results obtained by this efficient and largely automated procedure agree well with the manual achieved results of Steelman \& Endres [2010], derived by a more laborious largely manual analysis strategy. Then we develop a second methodology to global invert reflection traveltimes with a particle swarm optimization approach more precise then conventional spectral NMO-based velocity analysis (e.g., Greaves et al. [1996]). For global optimization, we use particle swarm optimization (PSO; Kennedy \& Eberhart [1995]) in the combination with a fast eikonal solver as forward solver (Sethian [1996]; Fomel [1997a]; Sethian \& Popovici [1999]). This methodology allows us to generate reliability CMP derived models of subsurface velocities and water content including uncertainties. We test this method with synthetic data to study the behavior of the PSO algorithm. Afterward, We use this method to analyze our field data from a well constrained test site in Horstwalde, Germany. The achieved velocity models from field data showed good agreement to borehole logging and direct-push data (Schmelzbach et al. [2011]) at the same site position. For the third method we implement a global optimization approach also based on PSO to invert direct-arrival traveltimes of VRP data to obtain high resolution 1D velocity models including quantitative estimates of uncertainty. Our intensive tests with several traveltime data sets helped to understand the behavior of PSO algorithm for inversion. Integration of the velocity model to VRP reflection imaging and attenuation model improved the potential of VRP surveying. Using field data, we examine this novel analysis strategy for the development of petrophysical models and the linking between GPR borehole and other logging data to surface GPR reflection data.},
  language  = {de}
}