[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

Re: The Emperor's New Papers

Delurking for a moment.

I'm sorry, I am generally not a very negative peer reviewer but........ Having done some stats over time, I appreciate their usefulness in certain things however....... IMHO, there will never be a substitute for an experienced pair of eyes (or in my case eye) in the field. Statistical methods in this matter could lead you to a fossil-less gully while ignoring the grassy hillside nearby with vertebra sitting on the surface. Chance (pure luck) is only one way statistical probability can account for the gauntlet of preservation, position relative to the surface, and crossing of the two time lines (the fossils and yours). You have to go into the field to gather the data to put into the stats which means you have already picked up the neon signs that say dig here anyway. Remote sensing will tell you nicely that this 14 mile wide stretch of earth consists of Hell Creek (for instance) and you may know that upper Hell Creek may have more fossils than lower (around here anyway) but there is very limited application to stats in this particular part of the section. I haven't read the paper and I am sure it is/was well done technically but field work has to be done to acquire enough data to make it useful. If I mark down every occurrence of gar pike (fish) scales around here, I will notice that lots of gar scales at a site means that I will find other fossils too. So I then go back to the computer and put that in, out comes the result of the computer saying dig there. I could have saved the trip back to the lab and just dug there in the first place.

Ok, gullies have more fossil exposures than hillsides generally. The uppermost Cretaceous Hell Creek Formation has more dinosaur fossils than the Paleocene Formations above generally. I guess the statistical GIS system would tell me that. Any GIS statistical system would just as likely lead you on a wild fossil avian theropod chase. This kind of system is only as good as the field work that gathered the data to create the layers of numbers. In my experience, the third standard deviation (or for that fact the second) doesn't apply to fossil hunting very well. Giving GIS it's due, the interpolation of bed spatial geometry mixed with topographic data may be (marginally) useful in pin pointing presence or absence of a particular unit on the surface. (Aerial photos do that too). But I can't imagine how GIS stats can interpolate preservational bias, exposure, degrees of weathering, vegetative cover, paleoenvironmental specific et al in that particular bed without the field work that would walk you to the fossils in the first place. Heck, if the Angus bulls are out in that pasture, the GIS system might just get you into a foot race.

Fossil hunters of the world are not in danger of loosing their job to a GIS system. Being able to recognize color and texture of fossils and being able to separate them from the visual noise of the background sediments and vegetation will continue to be the way it is done. Covering ever square inch of terrain is called "paying your dues and looking for clues".

Maybe I missed the point.

Frank (Rooster) Bliss
MS Biostratigraphy
Weston, Wyoming

On Jul 23, 2007, at 8:07 PM, Andy Grass wrote:

Could someone send me a PDF of this, if anyone has it?

Oheim, K.B. 2007. Fossil site prediction using geographic information
systems (GIS) and suitability analysis: The Two Medicine Formation, MT, a
test case. Palaeogeography, Palaeoclimatology, Palaeoecology
251(3-4):354-365. doi: 10.1016/j.palaeo.2007.04.005.

ABSTRACT: Fossil site discovery has traditionally been the result of
educated guesswork followed by systematic searching of terrain. This study
approached the issues of fossil site identification by looking at key
variables in a GIS setting. The data were analyzed to create a predictive
model for finding fossils, thus facilitating the process of fossil discovery
and saving time and money. Geospatial variables believed to be most useful
for finding fossils were examined and ranked on a scale from 1 to 4, with 4
being the most advantageous score for finding fossils. Weighted sum addition
combined the layers to create a suitability surface. Field testing and
subsequent analysis showed the model accurately predicted areas of high,
medium, and low fossil likelihood. Field observations and additional site
data led to model refinements and increased resolution of fossil density
distribution. The final model explained a statistically significant 90% of
fossil density variation in the Two Medicine Formation.