Limitations on near-surface correction for multicomponent offset VSP

Multicomponent data are degraded due to near-surface scattering and Ion-ideal or unexpected source behaviour. These effects cannot be neglected when interpreting relative wavefield attributes derived from compressional and shear waves. They confuse analyses based on standard scalar procedures and a prima facia interpretation of the vector wavefield properties. Here, we highlight two unique polar matrix decompositions 'or near-surface correction in offset VSPs, consider their inherent mathematical constraints and how they impact on subsurface n terpretation. T h e subsurface response M (CO) may be assumed complex symmetric (M_A = M_A^T ) and the near-surface response to be complex orthogonal (N^T (to)N(w) = I), or M A (w) can be unitary (M A^T M, = I) and N(w) hermitean (N = N*^T ) . Neither assumption is suitable for the near-surface operator, although the former is more appropriate for a general anelastic subsurface. The first method is applied to a four component subset of a six component field data from a configuration of three concentric rings and walkaway source positions forming offset VSPs in the Cymric field, California. The correction appears successfull in automatically converting the wavefield into its ideal form, and the qS1 polarizations scatter around N15°E in agreement with the layer stripping of Winterstein and Meadows (199 1).