HE INTERIOR OF THE EARTH HAS FASCINATED PHILOSOPHERS, PO-T ETS, and scientists throughout much of history. With the acceptance of the theory of plate tectonics in the last few decades, geologists now realize that some of the features we observe on the earth's surface have been formed by dynamic processes occurring deep within the planet's interior. Thus, understanding the inner workings of the earth has taken on new significance. Since we can't actually go down very deep into the earth, indirect observation combined with limited laboratory simulation of deep-earth conditions has had to suffice. These methods, in combination with increasingly sophisticated numerical models, tell us much of what we know about the inner dynamics of our planet.The dynamics of the earth's mantle-the layer that makes up the majority of the earth's volume-can be modeled using the equations that govern a creeping, viscous fluid. While on a human time scale the earth seems like a rigd, elastic body, over millions of years it responds to stresses by creeping flow. This behavior is similar to the more familiar movement of glaciers. Two of the problems faced by geophysicists in modeling the earth's interior are the variation in the material properties of the mantle and the range of length scales of the phenomenon they want to study. To study the problems currently of interest to geologists requires three-dimensional, spherical shell models that can resolve details on the order of several kilometers. This is beyond the CPU and memory limitations of almost all computers available today. In addition, the large variation in the material properties of mantle materials causes most numerical algorithms to become unstable.To model the dynamic behavior of the earth's interior, we must first know something about its composition. Our knowledge is considerable but far from perfect: many of the material parameters are only poorly constrained by laboratory and other measurements. It is useful to note in passing that numerical modeling can help to test the sensitivity of the earth's interior dynamics to the unknown parameters, guiding experimentalists toward better measurements of the key properties that control the dynamics. 1070-9924/95/$4 00 0 1995 IEEE IEEE COMPUTATIONAL SCIENCE h ENGINEERING BEST COW AVAILABLB , \