Introduction to Earth Sciences I
Topic 5
Prediction and Predictability of Earth Process
Up until now we have been developing a basis for understanding the Earth's physical systems by the development of simple theories that explain common observations, and we have learned how to deduce many aspects of the Earth's internal properties and behavior from the external observables. We would now like to apply what we have learned to address the question of predictability of the Earth's processes. The importance of this comes in our desire to forecast phenomena that can gave harmful consequences - large earthquakes, floods, hurricanes, landslides and volcanic eruptions. These events turn out to be very difficult to predict as individual events. We have yet to predict an earthquake. Volcanoes generally let us know when they are about to erupt by swelling and other indications, but before these things start to happen we have no idea that they will, in fact, begin. Why ? The answer is in the nature of the systems.
In general our ability to predict relies on recognizing some
sort of pattern, either in time or in space. The more regular a pattern is the
more readily one can predict in unknown areas. So for instance, if you had never
before heard a clock ticking regularly out the seconds you could listen to the
ticking for a while and be pretty confident that the clock will keep ticking
the same way. In other words, you could readily predict the future behavior
of the clock because all of its behavior in the past has been regular also.
I can't be absolutely sure of this, but I have gained intuition about the clock's
behavior by observing it closely and over a fairly long time.
However, if the clock was malfunctioning and was ticking sometimes on the second,
sometimes a little early and other times a little late and there was no order
to that behavior it would be very hard for you to say when the next tick would
come. That is, the past behavior of the system wouldn't give you sufficient
guidance to allow you to predict the future behavior of the system.
The same is true for special patterns. Were I to walk north in Manhattan for
the first time noting the names of cross streets as I go I would pretty soon
figure out that they are consecutively numbered in increasing order from south
to north. Standing at 50th street you could be pretty certain that the next
street is 51st - a spatial prediction. We all know that this regularity runs
out after a while without warning, so the prediction isn't perfect; it works
for a while then breaks down.
Unless you know how a clock mechanism works you cannot be completely sure why
the ticking should continue, and unless you know the city planners Master Plan
you can't be sure of the way the street patterns will go. In other words, you
really need a theory that explains the previous observations to be sure what
the next observations will be. The search for theories that explain past observations
well enough to be able to project into the future is one of the major quests
of modern Earth science, particularly because much of what the Earth is capable
of doing is fairly destructive to human life.
Theories can be very difficult to come by and often we are left we attempting
to find patterns of regularity where there appears to be none. The regular ordering
of sand pile landslide sizes in relation to their frequency of occurrence is
one example. Another is described below. It began when a scientist named Benoit
Mandelbrot asked the disarmingly simple question "how long is the coastline
of England?"