Abstract: Based on the stochastic Poisson Model and the Gutenberg-Richestlaw, we simulated a set of earthquake catalog ues, from 300 to 5000 events in each, and calculated how the load /unload response ratio (LU RR), namely Y, varies with time under different sample conditions. The results show that the difference between upper magnitude threshold M_t used in the calculation and the maximum magnitude extrapolated from the linear relation of log N~M has an influence on the distribution of Y. The effect decreases rapidly when the sample size of the earthquake events increases. When the number of events in the sliding time window for calculating Y increases to a quantity of several dozens, and M_t is about M_T-1. 5 (or to 2.0), Y gets a much lower probabi lity to be a high value greater than 2. 0. When the above number of events reaches an order of mag nitude of hundreds, and even M_t is about M_T -0.0 (or to 0.5), Y also gets a very low probability to be a high value greater than 2.0. This research shows that if the upper magnitude threshold M_T is considered when calculating Y from real earthquake catalog ues, the stability of Y and the reliability of its anomalies will be much improved.

Key words: Sample condition, Upper magnitude, Simulating earthquake, LURR