Extrapolating Beyond the Last Data Point

Kim Kastens
Author Profile
published Aug 19, 2009

In followup to our recent EOS paper "How Geoscientists Think and Learn" (Kastens, Manduca, et al, 2009), Michael D. Max, of Marine Desalination Systems, wrote:

"In most scientific disciplines data is produced, analyzed and interpreted. Extrapolation beyond a data set is more or less forbidden. Geologists, on the other hand, are trained from the outset to understand that they almost never will have enough actual data to arrive at a useful solution except in the most closely controlled three dimensional situations such as mine and reservoir mapping. But this usually involves a level of cost that is only rarely obtainable."

I think this is an interesting point, that extrapolation beyond the last data point is more or less forbidden in science, but geologists do it anyway.

Historically, geoscientists have been taught to think mostly about the past. We were taught to think about the past through fossils, through fluctuations in the earth's magnetic field, though stable and radiometric isotopes, course after course on how to think about the past via various ingenious lines of reasoning.

In recent years, though, society has been asking the geoscience community to answer questions about the future, rather than about the past: what will happen to the climate? when will the next earthquake be? etc. To think about the future, by definition, requires making interpretations outside the bounds of the available data, which as Dr. Max noted, is something that many scientists shy away from.

What do we need to do to teach the next generation of geoscientists to be (a) comfortable and (b) competent at thinking rigorously about the future?


Reference:




Extrapolating Beyond the Last Data Point --Discussion  

Kim: Isn't it the case that post-diction (i.e. telling the story of the past history of the Earth) opens the door for prediction? If the Earth was capable of operating in some well-documented way in the past, why couldn't it do so again? Aren't we better off by knowing something about frequency and magnitude of Earth events like earthquakes, floods, extreme weather, etc. as an appropriate means to make plans to live on, and plan for dealing with, a restless Earth? Perhaps the issue lies in schools of thought that are primarily deterministic compared with much of Earth science which is probablistic.

1176:3584

Share edittextuser=7 post_id=3584 initial_post_id=0 thread_id=1176

This post was editted by Kim Kastens on Jan, 2010
[further thoughts from author of post]
Today I took a teacher's question about the early days of plate tectonics to my Lamont-Doherty colleague Walter Pitman, one of the pioneers of the use of seafloor magnetic anomalies for determining rates and directions of plate motion. He got to reminiscing about the paper in which he and colleagues put ages onto magnetic anomalies, shaking his head about how lucky they had been when they took 4 million years worth of magnetic reversals dated from land sections plus a handful of dated seafloor samples and extrapolated boldly out to 80 million years worth of seafloor magnetic anomaly record. Now there is a case of extrapolating beyond the last data point! The extrapolation worked because it wasn't just a blind extrapolation, it was an extrapolation based on a model, and the model, seafloor spreading, turned out to be correct.

In the same conversation, Walter passed on an interesting quote from former Barnard geology professor John Sanders, paraphrased: In sciences like physics and chemistry you are always trying to design an experiment to prove or disprove an hypothesis. In geology you are always trying to figure out what the experiment was.

Ref: Heirtzler, J. R., G. O. Dickson, E. M. Herron, W. C. Pitman III, and X. Le Pichon (1968), Marine Magnetic Anomalies, Geomagnetic Field Reversals, and Motions of the Ocean Floor and Continents, J. Geophys. Res., 73(6), 2119-2136. Online at: http://www.agu.org/pubs/crossref/1968/JB073i006p02119.shtml

1176:3667

Share edittextuser=75 post_id=3667 initial_post_id=0 thread_id=1176

1176:3668

Share edittextuser=75 post_id=3668 initial_post_id=0 thread_id=1176

Join the Discussion


Log in to reply


« Science Education Should Not be Viewed as the Public Relations Arm of the Research Establishment       "Geologists are comfortable with uncertainty" »