Re: Collaboration on a facies training images database
On this issue. Training images are often used for fluvial or clastic environments (like in the petrel manual for this module). They are far more valuable though and this article shows their use in carbonate reef complexes. http://www.searchanddiscovery.com/docum ... /index.htm
We are currently engaged in such a modeling project and are getting some impressive results. The output of the modeling is just as good as our input and assumptions though, like in any model. We acknowledge this and basically our outcome is what we would expect and what we would also have created by hand, based on our understanding of facies distribution. Therefore I feel that what you propose - to get more robust and elaborate training image databases - is very valuable. I give it a couple of years and this collaborative project will be kicked off...
great idea and very interesting forums!