HR: 17:00h
AN: H44C-05 INVITED     [Abstracts]
TI: Joint inversion of crosshole tomographic data using the cross-gradient function and stochastic regularization operators
AU: * Linde, N
EM: niklas.linde@geo.uu.se
AF: Department of Earth Sciences/Geophysics, Uppsala University, Villav. 16, Uppsala, 75236 Sweden
AU: Binley, A
EM: a.binley@lancaster.ac.uk
AF: Department of Environmental Science, Lancaster University, Lancaster, LA1 4YQ United Kingdom
AU: Tryggvason, A
EM: ari.tryggvason@geo.uu.se
AF: Department of Earth Sciences/Geophysics, Uppsala University, Villav. 16, Uppsala, 75236 Sweden
AU: Pedersen, L B
EM: laust.pedersen@geo.uu.se
AF: Department of Earth Sciences/Geophysics, Uppsala University, Villav. 16, Uppsala, 75236 Sweden
AB: Hydrogeophysical investigations are often characterized by a multitude of data types. Ideally, models based on inversion of geophysical data should not only adequately fit the data used in the inversion; they should also be consistent with information derived from other data sources. This objective can, in theory, be achieved by joint inversion of all data. However, it is often practical to incorporate certain data types (e.g., geophysical logs) in the inversion by estimating a priori models and geostatistical models used to calculate regularization operators. We introduce (1) an efficient method to calculate regularization operators based on stationary geostatistical models by utilizing the block-circulant structure of model covariance matrices and (2) a structural approach to joint inversion of crosshole geophysical data. Crosshole ground penetrating radar (GPR) traveltime and electrical resistance tomography (ERT) data, collected in unsaturated sandstone close to Eggborough, UK, were jointly inverted in three-dimensions by assuming that the gradients of the two models have the same or opposite directions. This was achieved by including the cross-gradient function in the objective function; the cross-gradient function was recently defined and successfully applied to surface based data. The regularization operators used in the inversions were based on geostatistical models derived from EM conductivity logs and zero-offset profile GPR traveltime data. For a given data fit, models based on individual inversions that used the new regularization operators were more layered compared with models obtained by using traditional smoothness constraints; a pronounced layering is evident in the EM conductivity and gamma logs. The resulting models based on the joint inversion fit the data within the estimated error levels and they are structurally similar making them useful to infer geological units, clay content, or saturation. Finally, we argue that structural approaches are suitable for joint inversion of hydrogeological and geophysical data.
DE: 0925 Magnetic and electrical methods (5109)
DE: 1875 Vadose zone
DE: 3260 Inverse theory
DE: 3285 Wave propagation (0689, 2487, 4275, 4455, 6934)
SC: Hydrology [H]
MN: Fall Meeting 2005