Use of Bayesian Search Theory in geological interpretation I was having a look round a few things yesturday and came across Bayesian Search Theory. Thinking about this theory led me to think about a problem I was working on a few years ago regarding geological interpretation. 
We were looking at the geology at one specific site and it was essentially made up from two different types of rocks. Boreholes had been drilled at different locations and showed differing amounts of the two different types of rocks at different levels in the ground along with different amounts of weathering of the rock. A number of geologists looked at the available data and all came up with different interpretations. It seems to me that Bayesian Search Theory could have been used in this case, particualrly where extra data was gathered with time, to give some indication of how likely the different interpretations were. 
Has anyone encountered a case where Bayesian Search Theory has been used in this case. Is there a standard frameowrk for doing this? I would have thought this may be something that the oil industry may have a lot of research on because it would be applicable to the search for oil. 
 A: Though it is not generally labeled as Bayesian search theory, such methods are pretty widely used in oil exploration.  There are, however, important differences in the standard examples that drive different features of their respective modeling problems.  
In the case of lost vessel exploration (in Bayesian search theory), we are looking for a specific point on the sea floor (one elevation), with a  distributions modeling the likelihood of its resting location, and another distribution modeling the likelihood of finding the boat were it at that depth.  These distributions are then guide search, and are continuously updated through the results of the guided search.
Though similar, oil exploration is fraught with complicating features (multiple sampling depths, high sampling costs, variable yields, multiple geological indicators, drilling cost, etc.) that necessitate methods that go beyond what is considered in the prior example.  See Learning through Oil and Gas Exploration for an overview of these complicating factors and a way to model them.
So, yes, it may be said that the oil exploration problem is different in magnitude, but not kind from lost vessel exploration, and thus similar methods may be fruitfully applied. Finally, a quick literature search reveals many different modeling approaches, which is not too surprising, given the complicated nature of the problem.
A: There is a free book on Geostatistical Mapping with R here it might help your problem.
