An Independent Test of the Load/Unload Response Ratio Proposed Method of Earthquake Prediction

It has often been hypothesized that earthquakes should occur more frequently during times of increasing tidal stress, yet the most careful studies suggest that the effect is very small and difficult to detect. However, as the tectonic stresses rise and approach failure, tidal stress variations might be more likely to raise the total stress to the failure level. Thus, the sensitivity of small earthquakes to the tidal stress before an impending large shock might be increased. This idea has been developed into the Load/Unload Response Ratio (LURR) proposed method of earthquake prediction (Yin et al., 1995). The LURR method involves looking at the ratio of the number or some measure of the size of small earthquakes that occur during loading of the tidal cycle to that measure of earthquakes that occur during unloading of the tidal cycle, and suggests this ratio should increase before an impending large shock. Smith and Sammis (2001, 2003) concluded that there was no predictive significance to the LURR method, and to our knowledge, this method has not been validated by any independent researchers. We calculated LURR values for several large earthquakes in southern California, using the methods and parameters described by Yin et al. (2000) and estimates of the other necessary methods and parameters that were not published. We employed tidal stresses, both from the solid earth tides alone and together with tidal ocean loading, using programs provided by Agnew (1996). Increases in the LURR value do occur, but they do not occur more often prior to large earthquakes than at any other time. We also calculated LURR values for altered datasets in which we randomized the times of each small earthquake before calculating the tidal stresses, theoretically eliminating any real influence of the tides on LURR. There is no statistically significant difference between preliminary results from the calculations utilizing the actual earthquake times and calculations utilizing the randomized times. After considering our results and those of Smith and Sammis (2001, 2003), we conclude that the LURR method is not a useful tool in earthquake prediction.