In local earthquake tomography (LET) problems, the spatial resolution of ve
locity structure varies significantly in a model due to the naturally heter
ogeneous distribution of earthquake sources. We present a flexible gridding
strategy for tailoring the velocity model parameterization to the dataset
being analyzed. The strategy combines regular gridding with two forms of no
de 'linking' to allow for finer-scale resolution where warranted by the dat
a, to provide improved model stability where resolution is relatively poor
and to allow for the incorporation of a priori information on structure, fo
r example the location of significant discontinuities. Flexible gridding ha
s been incorporated into a widely-used LET inversion algorithm. The inversi
on approach is documented along with a discussion of other recent advances
and improvements to the algorithm. We discuss various situations in which f
lexible gridding can be useful. An example application of flexible gridding
to a dataset from the San Andreas fault in central California illustrates
how the approach can yield a sharper image of the fault while at the same t
ime improving the overall quality of the model and the fit to the data. (C)
1999 Elsevier Science Ltd. All rights reserved.