Application of linear statistical models of earthquake magnitude versus fault length in estimating maximum expectable earthquakes

Abstract

Correlation or linear regression estimates of earthquake magnitude from historic magnitude and length of surface rupture data should be based upon the correct regression. For example, the regression of magnitude on log(length) can be used to estimate magnitude, but the regression of log(length) on magnitude cannot. Regression estimates are most probable values and estimates of maximum values require consideration of one-sided confidence limits.