| Popis: |
In the last few years we started a project of building computer programs for “exact” synthetic seismograms, i.e. that can be calculated with a given accuracy for all the wanted time values. In other words, we don’t make any approximations based on physical considerations as to expected phases and their characteristics, but let the complete solution to the problem indicate what phases appear indeed on the seismograms. One basic motivation of our approach is that when one approximates the solution according to additional assumptions and build a computer program that is easy to use, fast and attractive intuitivelly, there must be a way to determine if that particular algorithm is sound and the program works. There is also another good reason why we should get accurate results based solely on the solution of the initial-boundary value problem at hand without additional assumptions beyound the physical conditions used when formulating the problem. Namely, we don’t know in advance what we will see on the seismograms, sometimes we find that phases that should be small theoretically have an important contribution to the disturbance. This was the case with the so called S* wave discovered using a program to compute a complete seismogram (Hron and Mikhailenko, 1981). It turned out that characterizing this phase as “evanescent” is somehow misleading, since as one can see on the examples given in the above mentioned paper and others (e.g. F. Abramovici, 1992), this phase is one of the most proeminent feature of the seismogram. Our aim is therefore to build programs based on the exact mathematical solution and calculate this solution using numerical methods that give accurate results in an acceptable cpu time. |
