Cellular Automata (CA), a paradigm of parallel computing, represent an alte
rnative to differential equations and are used for modelling and simulating
very complex phenomena; CA models have been developed by our research grou
p for the simulation of landslides. We present SCIDDICA-3, our most efficie
nt model, a two-dimensional CA model together with the simulation results o
f the Mount Ontake (Japan) debris avalanche which occurred in 1984. Landsli
des are viewed as a dynamic system based exclusively on local interactions
with discrete time and space, where space is represented by square cells, w
hose specifications (states) describe physical and chemical characteristics
(friction, viscosity, altitude, debris thickness, etc.) of the correspondi
ng portion of space. At the time t=0, cells are in stales which describe in
itial conditions; the CA evolves then changing the state of all cells simul
taneously at discrete times. Input for each cell is given by the states in
the adjacent cells; the outflow computation from the cells gives the evolut
ion of the phenomenon. The comparison between the real and simulated event
is satisfying within limits to forecast the surface covered by debris. (C)
1999 Elsevier Science Ltd. All rights reserved.