Prediction in geology versus prediction in engineering

As humans become increasingly dominant agents of geologic change, prediction of the reaction of natural systems to human intervention and of the performance of geoengineered structures assumes increasing importance. To help clarify the role of geological prediction in an anthropic world, we examine the end-member cases of prediction in natural geologic systems and engineered systems. The behavior of natural geologic systems tends to be less reliably predictable than the behavior of engineered systems. Engineered systems are designed, and their behavior is predictable in terms of the function and interaction of their parts. Geologic systems, although undesigned, also have parts. Natural analogs of engineered parts are the emergent structures arising from nonlinear interactions between small-scale constituents. The behavior of natural systems at a given scale follows directly from the dynamics of their parts as defi ned at similar scales; it is argued that application of same-scale dynamics, or scale matching, provides the best basis for prediction in such systems. Mathematical models of natural systems are also likely to be most effective for prediction when applied at scales matched to the scales of the phenomenon of interest. There are also systems that are intermediate between natural systems and engineered systems. A "peri-engineering" transition zone governed by large-scale interactions is always present between the parts of the engineered structure and the parts of the surrounding natural environment. This contact region between engineered structure and the natural environment is often partially engineered to improve predictability. The peri-engineering halo is only partly subject to human design, and consequently it is often a region of reduced predictability and increased probability of malfunction or failure compared to the engineered system itself. © 2013 Geological Society of America.

Duke Scholars

Author Peter K. Haff Earth and Climate Sciences