Multifractality in Seismicity Spatial Distributions: Significance and Possible Precursory Applications as Found for Two Cases in Different Tectonic Environments

Abstract

We explore fractal properties of two observed seismicity distributions prior to the 2003 M w 7.4 Colima, Mexico and 1992 M w 7.3 Landers, USA earthquakes, together with several mathematical fractal distributions and two non-fractal ones, in order to estimate minimum reliable sample sizes, determine whether fractality for observed seismicity is essentially different from random uniform distributions, and explore the possibility of extracting premonitory information from fractal characteristics of seismicity before large earthquakes. Sample sizes above 800 events for whole catalogs appear to be sufficient to maintain ordered multifractality and to yield dimension estimates that vary smoothly and reliably. Fractal estimates appear to be best for whole catalogs that include aftershocks. The fractal characteristics of spatial distributions of seismicity are essentially different from those of the uniform random distribution, which is the null hypothesis of a non-fractal distribution with minimum information. The fractal dimensions and afractality measures of seismicity distributions change with time and show distinctive behaviors associated with foreshocks and main events, although these behaviors are different for each example. Results suggest the possibility of a priori identification of foreshocks to large earthquakes. A combination of fractal dimension and afractality measures over time may be helpful in large earthquake premonitory studies.