Geology Fail: The Problem with Proxies

Using an observable data set as a stand-in for a theoretical model can be misleading, as several new geology papers illustrate.

An ice core is pulled up from Greenland or Antarctica. Scientists measure oxygen isotopes at various levels. Do the measurements tell a story about past climate? Perhaps, but the data are only a stand-in, or proxy, for a history that cannot be observed directly. Historical geology is full of the use of proxies (indirect indications) to tell stories about continental and atmospheric changes over millions of years. How reliable are they? Let’s look at some confessions in this month’s leading earth science journal, Geology.

Snowball Earth: One of the leading proxies for the hypothesized “Snowball Earth” has been carbon and boron. When carbon-13 and boron-11 take “excursions” from their normal levels, geologists interpret “to represent considerable perturbations of the carbon cycle and the accompanying reduction in global ocean pH,” Stewart et al. say in Geology. “Yet this interpretation is predicated on these isotopic signals being primary in origin.” They call into question the primary origin of these signals, and with it, the Snowball Earth interpretation. “Our results clearly demonstrate that the boron system is impacted by meteoric diagenesis, implying that a rigorous assessment of the diagenetic history of all ancient carbonates is required to ensure any paleoceanographic interpretation” based on carbon and boron.

Mountain building: A proxy for the rate of mountain uplift has been the height and shape of river terraces. Now, Gallen et al. call that into question in Geology.  They think mistakes can be corrected, but if nobody thought of this correction before, what other unknowns are biasing the interpretation?

Incision rates derived from river terraces are commonly used to infer rock uplift rates; however, an apparent dependence of incision rate on measured time interval may confound directly relating incision to uplift. The time-dependent incision rates are a Sadler effect that have been argued to result from a stochastic distribution of hiatal intervals in river incision, potentially reducing the utility of incision records for interpreting unsteadiness in tectonic processes. Here we show that time-dependent incision rates can arise from a simple systematic bias in the distance measurement used to calculate incision rate, and thus stochastic causes are not required…. Because it is challenging to reconstruct the full elevation history for a river channel, most researchers use the modern streambed elevation as a reference datum, but we demonstrate that doing so imposes a bias that manifests as an apparent dependence of rate on measured time interval. Fortunately, correction of this bias is straightforward, and allows river incision data to be used in studies of tectonic or climatic unsteadiness.

 A man, a plan, Panama: The isthmus of Panama has a complex history, Goldsmith et al. say in their paper in Geology. They point out another long-assumed proxy that needs a correction. Their measurements of carbon dioxide consumption in andesite varies by an order of magnitude, “leading us to suggest that andesite terrains should be considered separately when calculating removal of CO₂ from the atmosphere via silicate weathering.”

Cave confidence: Bourne et al., in their Geology paper, feel confident that they can use magnetic material in stalactites as a proxy for long-term changes in precipitation. But can measurements from one cave in West Virginia speak for the climate history of the globe? Watch those hidden assumptions. First, they rely on U-Th dating. Then, “We interpret the changes in magnetite concentration as reflecting variations in local pedogenic processes, controlled by changes in regional precipitation.” How confident is everyone that this “can constrain interpretations of speleothem climate proxies” for global climate models?

Reassessing Lyell’s Dictum

The most revealing paper in Geology this month asks a startling question after 180 years of Lyellian dogma (i.e., uniformitarianism: “the present is the key to the past”). The title by Nyberg and Howell asks, “Is the present the key to the past? A global characterization of modern sedimentary basins.” Watch as they claim that Lyell was arguing out of thin air, not from sound evidence:

The stratigraphic record is heavily biased because it is uniquely composed of sediments that were laid down in basins, whereas the majority of the present and historic land surface of the planet is composed of areas that are in net long-term erosion with no preservation potential. Global mapping and quantification of the distribution of currently active sedimentary basins suggest that only 16% of Earth’s terrestrial land surface is within sedimentary basins; the remainder of the land is in upland areas that will not be represented in the future rock record. Furthermore, 60% of the modern basin area has an arid climate, as opposed to 27% of the land surface. Tectonic classification of modern basins indicates that intracratonic and foreland basins cover the greatest area, whereas forearc, extensional, and strike-slip basins are the least represented by area. While this modern snapshot does not account for differences in subsidence rate or basin longevity, the mapping and quantification of modern basins highlight the incompleteness of the stratigraphic record, and the importance of caution when assuming “the present is the key to the past.”

As an example, we reported back on 4/10/08 that some interpretations of the Grand Canyon require imagining previous canyons up in the air, eroded away long ago before the current canyon became visible.

A geologist holds a rock in his hand; if he’s a good storyteller, he has all the terms and jargon to tell how it got that way. Some of it may be true, but we think you deserve to know the amount of assumption, guesswork, speculation and interpretation that goes into the tales. These papers are just the latest in a long series of doubts, corrections and reassessments in historical geology. Proxies can often have multiple interpretations, because the known unknowns do not reveal all the unknown unknowns. And nothing is more misleading than a cute slogan, like “the present is the key to the past,” to oversimplify a complex world and send a whole army of indoctrinees on a primrose path to error.