| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 25.31 KB | Adobe PDF |
Advisor(s)
Abstract(s)
A new method, based on linear correlation and phase diagrams was successfully developed for
processes like the sedimentary process, where the deposition phase can have different time
duration - represented by repeated values in a series - and where the erosion can play an
important rule deleting values of a series. The sampling process itself can be the cause of
repeated values - large strata twice sampled - or deleted values: tiny strata fitted between two
consecutive samples.
What we developed was a mathematical procedure which, based upon the depth chemical
composition evolution, allows the establishment of frontiers as well as the periodicity of
different sedimentary environments. The basic tool isn't more than a linear correlation analysis
which allow us to detect the existence of eventual evolution rules, connected with cyclical
phenomena within time series (considering the space assimilated to time), with the final
objective of prevision.
A very interesting discovery was the phenomenon of repeated sliding windows that represent
quasi-cycles of a series of quasi-periods. An accurate forecast can be obtained if we are inside
a quasi-cycle (it is possible to predict the other elements of the cycle with the probability
related with the number of repeated and deleted points).
We deal with an innovator methodology, reason why it's efficiency is being tested in some case
studies, with remarkable results that shows it's efficacy.
Keywords: sedimentary environments, sequence stratigraphy, data analysis, time-series,
conditional probability.