Inversion for multiple parameter classes

✍ Scribed by B. L. N. Kennett; M. Sambridge

• Book ID: 104460806
• Publisher: John Wiley and Sons
• Year: 1998
• Tongue: English
• Weight: 73 KB
• Volume: 135
• Category: Article
• ISSN: 0956-540X
• DOI: 10.1046/j.1365-246x.1998.00657.x

✦ Synopsis

Summary

Many geophysical data, such as the frequencies of the free oscillations of the Earth, depend on more than one type of model parameter. For inverse problems depending on multiple parameter classes, an iterative solution procedure is introduced in which each parameter class can be treated in the same way. This approach has considerable advantages where a large number of parameters are employed, but can still be useful for smaller systems.

The iteration by parameter class commences by solving for the direct dependence on a particular parameter class, and at subsequent iterations the cross-dependences between classes are introduced. The update affects only the right-hand side of the equations, and, because the same sets of equations have to be solved at each iteration, an efficient computational implementation can be made. The largest set of equations that has to be solved at a time corresponds to the number of variables in an individual parameter class rather than the full set of parameters, which confers substantial computational benefits for very large problems.