2008 Joint Meeting of The Geological Society of America, Soil Science Society of America, American Society of Agronomy, Crop Science Society of America, Gulf Coast Association of Geological Societies with the Gulf Coast Section of SEPM

Paper No. 1
Presentation Time: 8:00 AM-6:00 PM

Composite Section Formation Via Simulated Annealing of Horizon Ordination

SHEETS, H. David1, WILLIS, John M.1, IZARD, Zachary1 and MITCHELL, Charles2, (1)Dept. of Physics, Canisius College, 2001 Main St, Buffalo, NY 14208, (2)Geology, University at Buffalo, 411 Cooke Hall, Buffalo, NY 14260, sheets@canisius.edu

The process of composite section formation based on biostratigraphic information produces a temporal ordering of events and collections, critical for any high-temporal resolution study of geological events, particularly patterns of biodiversity change. Shaw's graphic correlation method has been widely used, as have more recent numerical methods such as RASC/CASC and CONOP. We present a new method, horizon annealing (HA), which like CONOP uses simulated annealing methods to produce an optimal ordination, minimizing the range extension implied by a proposed solution. Unlike CONOP, which produces ordinations of species FADS and LADS and other marker events, horizon annealing directly orders collection horizons. We present example calculations of composites based on the Riley formation problem as discussed by Shaw and Sadler and also on graptolite collections from the Yangtze platform. Results from HA are similar to those produced by CONOP or graphic correlation. However, the horizon annealing approach may offer advantages over other numerical approaches in dealing with large numbers of sections, integrating new collections into existing composites, or in merging multiple existing composite sections. HA readily produces detailed records of presence/absence information encompassing collections from many discrete sections, which are desirable for detailed study of biodiversity patterns. Additionally, the approach used may be readily adapted to other numerical search methods such as genetic algorithms.