Science News 

Week of Feb. 2, 2002; Vol. 161, No. 5 

Return to Gallery

It's a Rough World
Fractals help model vexing problems in earth science

Sid Perkins

In the realm of mathematics, perfection abounds. Lines stretch 
straight to infinity, planes are flawlessly flat, and spheres are 
impeccably round. The real world, however, is almost always 
irregular—the jagged spear of a lightning bolt, the rough face of a 
broken rock, and the ragged profile of a mountain range are just a 
few examples. Although people have always been surrounded by texture, 
until recently they could not describe it in anything other than 
qualitative terms such as smooth or rugged or lumpy. Long after 
scientists developed ways to measure physical properties such as 
temperature, weight, and time, the techniques needed to quantify 
roughness eluded them. 

Then came fractals.

Before these mathematical tools were developed during the 1960s, 
scientists usually represented the physical world using only three 
dimensions. Lines occupy one dimension; planes, two; and cubes, 
three. But fractals freed scientists from the tyranny of integer 
dimensions and enabled them to describe objects using fractional 
dimensions.

Not only do such representations of objects show roughness and 
irregularity, they do so across the entire range of scale. A close-up 
look at a jagged nonfractal line would reveal that it’s made up of 
small, straight segments. But for a fractal line—say, a 
1.3-dimensional line—the view would be the same from afar as from 
close-up. Each segment would have the same degree of roughness no 
matter the scale at which it’s viewed. Likewise, a microscopic look 
at a rock with a 2.7-dimensional surface would display the same 
texture as the telescopic view of a cliff face with the same 
fractional dimension. 

Far from being merely mathematical abstractions, fractals pop up 
throughout the natural and the humanmade world. They appear in the 
swirling patterns on a flowing liquid’s surface (SN: 1/23/93, p. 53) 
and the patchy mosaics of urban sprawl (SN: 1/6/96, p. 8). 

Scientists have used fractals to analyze all sorts of irregular 
objects or phenomena, from swings in the stock market to the 
frequency of natural disasters, says Donald L. Turcotte, a 
geophysicist at Cornell University. One of the first uses of fractals 
in geosciences was a simple one, describing the roughness of 
Britain’s coastline. Today, researchers are applying fractals to 
far-ranging topics—the probability of wildfire, as well as the spread 
of toxic fluids through rocks and soil.

Natural disasters 

Earthquakes, wildfires, and other large-scale phenomena are the focus 
of many fractal analyses. The relationship between the frequency and 
the magnitude of earthquakes is fractal, Turcotte notes. When 
researchers plot these two quantities on a logarithmic graph, the 
data points fall on a line. In such a correspondence between the two 
quantities, called a power-law relationship, the slope of that line 
is an exponential power that can be employed as a fractal dimension. 
Fractal equations allow scientists to more easily develop computer 
models to simulate how often earthquakes occur. 

Scientists first noted the power-law relationship between the 
frequency and magnitude of earthquakes in the 1950s—well before 
fractals were invented, says Bruce D. Malamud, a mathematician at 
King’s College in London. The relation holds for earthquakes of 
almost all sizes. 

More recently, Malamud and his colleagues showed that a power-law 
relationship exists between the frequency and size of wildfires. The 
researchers discovered that the fractal relationship held true for 
four different sets of wildfire data, even though the sets were 
compiled for areas of the world radically different in terrain and 
vegetation. Malamud says that the consistent relationship between the 
size and the frequency of wildfires—even for time periods as short as 
2 years—gives hazard-management personnel an opportunity to use 
limited amounts of data to predict the likelihood of wildfires of 
various sizes in an area. 

Scientists are now subdividing the four wildfire data sets to see if 
intervention—either extinguishing fires before they die out naturally 
or setting preventative fires to remove dried brush—changes the 
relationship between the size and frequency of blazes. The new 
analyses could reveal whether certain firefighting policies affect 
the location and intensity of wildfires in unexpected ways, Malamud 
says.

Fractal fractures

Many phenomena related to the fracture of rocks and minerals, from 
tiny bits of volcanic ash to the large blocks of Earth’s crust that 
scrape past each other during earthquakes, can be described or 
modeled using fractals. North of Los Angeles, the two sides of the 
San Gabriel Fault sandwich a 6-meter-thick layer of pulverized 
material. This crushed rock, called gouge, can be found between the 
surfaces of many faults, says Charles G. Sammis, a geophysicist at 
the University of Southern California in Los Angeles. 

The power-law relationship between the number and the size of the 
particles, from large rocks to fine mineral grit, has a fractal 
dimension of 1.6, Sammis notes. In other words, chunks of rock with a 
cross-sectional area of 1.6 units are about one-tenth as common as 
those with a cross-sectional area of 1 unit. This fractal 
relationship holds across a wide range of particle sizes, he adds. 

Gouge forms in the early life of a fault, says Sammis. Stress 
transfers from one side of the fault to the other through bridges of 
large rocks. When the stress along the fault is high enough to break 
the gouge particles, it fractures only those that are bearing the 
load. As the load shifts to new bridges, those rocks fracture, and 
the debris is gradually ground into finer and finer grains, all the 
while maintaining a size distribution with a fractal dimension of 
1.6. 

Computer models that simulate particles of rock trapped in a moving 
fault, as well as laboratory tests with pellets of crushed granite, 
also generate a distribution of particle sizes with a fractal 
dimension of 1.6. This specific value is interesting, Sammis 
explains, because it seems to mark a transition point for the 
behavior of faults. For values more or less than 1.6, the gouge does 
not transfer stress across the fault because bridges don’t form. In 
those cases, the two faces of the fault slide past one another 
without triggering an earthquake. When the value equals 1.6, however, 
the rock bridges lock the fault and stress builds up until a temblor 
occurs. 

Many other characteristics of fractured rock are fractals, notes 
Christopher C. Barton, a geophysicist with the U.S. Geological Survey 
in St. Petersburg, Fla. For example, the relation between the number 
and the lengths of fractures in many types of rock follow a fractal, 
power-law relationship on scales that range from long fissures in 
Earth’s crust down to microscopic cracks in small stones.

Flowing through the rock 

Scientists have developed computer models that use fractals to 
simulate the movement of fluids through permeable rock, such as 
limestone. Rina Schumer, a hydrologist at the Desert Research 
Institute in Reno, Nev., and her colleagues model the spread of fluid 
through fractured and porous rock. A rock’s porosity—the percentage 
of the material’s volume that’s occupied by space between solid 
grains—can easily be expressed as a fractal dimension. 

By taking just a few measurements of rock and soil characteristics 
near the site of a toxic spill such as a fuel leak, for example, 
researchers can use fractals to estimate the directions and speeds at 
which the fluid will travel through the surrounding earth. This 
information can be useful to emergency-response personnel, and it may 
eventually help government officials develop plans to protect 
aquifers from spills.

Models that use fractals to simulate the presence of fractures in 
porous rock may provide better answers than those that assume the 
rock is uncracked. For example, Department of Energy scientists 
modeling the potential seepage of fluids from Nevada’s Yucca Mountain 
Project—an underground repository being designed to hold radioactive 
waste from nuclear plants (SN: 1/19/02, p. 39: 
http://www.sciencenews.org/20020119/fob7.asp)—estimate that if a leak 
were to occur, it would take about 1,000 years for the first traces 
of tainted fluid to reach 5 kilometers from the site. However, that 
analysis assumes that waste simply seeps through unfractured rock, 
says David T. Purvance, a hydrogeologist at hydrOhm Environmental 
Geophysics in Reno.

A fractal analysis that incorporates the effect of fractures in the 
rock predicts a much faster travel of liquid radioactive waste. A 
model that uses a previously measured distribution of crack sizes in 
the region’s so-called fractured welded tuff—a cracked, porous rock 
made up of volcanic ash particles that have been fused together under 
extreme heat and high pressure—estimates that spilled waste could 
travel a distance of 5 km in only 200 years. 

Volcanic dimensions 

Although the roughness of the glassy particles spewed out by volcanic 
activity doesn’t keep the same fractal dimension across all scales, 
measuring that parameter of the ash bits nevertheless may help 
scientists analyze the pattern of a volcano’s past eruptions. Anton 
H. Maria and Stephen N. Carey, both volcanologists at the University 
of Rhode Island in Narragansett, studied ash ejected from five 
different eruptions, including some deep undersea.

The researchers found that the combination of fractal dimensions of 
the ash particles measured at two different scales enabled them to 
discriminate among the eruptions. The surface roughness of the ash on 
the larger scale was linked to the abundance of large bubbles in the 
molten rock that had erupted from the volcano. The explosive power of 
the eruption shattered the rock along these bubbles, leaving small 
particles with ragged edges. At smaller scales, the ash particles’ 
roughness was linked to the abundance of small bubbles trapped within 
the material as it cooled. These two factors explained more than 90 
percent of the differences among the ash particles from the five 
eruptions, says Maria.

The size of the bubbles in the volcanic ash particles seems to be 
influenced by several factors, including the type and amount of gases 
that accompanied the eruption and whether the eruption was explosive 
or relatively calm. Ash particles from eruptions on land, where the 
ash cloud spewed directly into the air, typically have different 
fractal dimensions from those produced by underwater volcanic 
activity.

These distinctions are particularly important because they may help 
scientists track a volcano’s long-term history in areas where there 
are few if any written records, says Maria. By analyzing particles 
from different ash layers in a long sequence of eruptions, 
researchers might be able to tell whether all of a volcano’s 
eruptions have been explosive or its strong eruptions were 
interspersed with calmer belchings of ash. 

Return to Gallery

References and Sources

References:

Barton, C.C. 2001. Scaling of rock fractures: An overview (Abstract 
NG32A-05). American Geophysical Union 2001 Fall Meeting. Dec. 10-14. 
San Francisco. 

Maria, A.H., and S.N. Carey. 2001. Fractal spectrum technique for 
quantitative analysis of volcanic particle shapes (Abstract 
V42D-1044). American Geophysical Union 2001 Fall Meeting. Dec. 10-14. 
San Francisco.

Sammis, C.G. 2001. Fractal fragmentation, friction, and 
faulting(Abstract NG32A-07). American Geophysical Union 2001 Fall 
Meeting. Dec. 10-14. San Francisco.

Schumer, R., D.A. Benson, and M.M. Meerschaert. 2001. Modeling 
multidimensional contaminant plume growth with fractal scaling 
(Abstract NG31B-0369). American Geophysical Union 2001 Fall Meeting. 
Dec. 10-14. San Francisco.

Turcotte, D.L. 2001. The legacy of Benoit Mandelbrot in geophysics 
(Abstract NG32A-02). American Geophysical Union 2001 Fall Meeting. 
Dec. 10-14. San Francisco.

Ventosa, I.P., and B.D. Malamud. 2001. Cellular automata models: A 
useful modelling tool to define environmental policy for forest fires 
(Abstract NG51B-0464). American Geophysical Union 2001 Fall Meeting. 
Dec. 10-14. San Francisco.

Further Readings:

Burroughs, S.M., and S.F. Tebbens. 2001. Upper-truncated power laws 
and limits to scale invariance in natural systems (Abstract 
NG31B-0376). American Geophysical Union 2001 Fall Meeting. Dec. 
10-14. San Francisco.

Malamud, B.D., G. Morein, and D.L. Turcotte. 1998. Forest fires: An 
example of self-organized critical behavior. Science 281(Sept. 
18):1840-1842. Abstract available at 
http://www.sciencemag.org/cgi/content/abstract/281/5384/1840.

Mandelbrot, B.B. 2001. Fractals as the measure of roughness in the 
earth sciences (Abstract NG12B-01). American Geophysical Union 2001 
Fall Meeting. Dec. 10-14. San Francisco.

______. 2001. Problems of geophysics that inspired fractal geometry. 
American Geophysical Union 2001 Fall Meeting. Dec. 10-14. San 
Francisco.

Peterson, I. 1996. The shapes of cities. Science News 149(Jan. 6):8-9.

______. 1993. From surface scum to fractal swirls. Science News 
143(Jan. 23):53.

______. 1983. Fractals for modeling ecosystems. Science News 123(June 
11):381.

Purvance, D.T. 2001. Travel times of nonlocal dispersion and their 
geoelectric approximation in Nevada's fractured welded tuffs. Water 
Resources Research 37(December):2915.


http://www.sciencenews.org/20020202/bob10.asp
From Science News, Vol. 161, No. 5, Feb. 2, 2002, p. 75.
Copyright (c) 2002 Science Service. All rights reserved.

Return to Gallery