Mining-induced seismic events and the occurrence of large seismic events are critical concerns for many underground operations. With the trend in the mining industry heading towards deeper operations and larger production rates, this concern will only increase. This thesis presents a methodology for assessing highly stressed regions and identifying periods with increased likelihood to the occurrence of large seismic events.

A 67 month analysis, from January 2006 to July 2011, has been conducted at Coleman Mine focusing on interpreting their seismic data. The Apparent Stress Time History (ASTH) analysis has been selected to analyse this data since it is capable of displaying stress increases and instability in a rockmass. This ASTH technique creates a connection between blasting periods and the seismic response to mining. The current ASTH model has been tested against Coleman's data and has produced good results between temporal increases in mining-induced stress and periods of increase hazard, with 44% of the large events in the considered areas being identified by this method. The remaining 46% of the events either occurred "out of the blue", meaning that there was no precursor evidence from this model to suggest the occurrence of large events, or during the blasting window (10%).

This thesis also proposed alternative models for the ASTH to be developed and investigated. These alternative models focused on normalizing the Apparent Stress Frequency (ASF) and finding a relation between the rate of High Apparent Stress (HAS) events and the rate of all events. The alternative model of normalization placed too much weight on the time frame making the trends in the chart volatile to the timeframe selected. The second alternative was developed to remove the reliance on an arbitrary number of events to identify a period of high Apparent Stress and increased hazard. It was tested on both a highly stressed location and a major feature within Coleman Mine. This second alternative identified that there was no relation the existed between the event rate and HAS. In addition, this approach was sensitive to low event rates making interpretation ambiguous.

An analysis of the Lunch Room Fault was explored due to the occurrence of many large events in this region. The ASTH model was unsuccessful in the identification of periods of increased hazard and the occurrence of large seismic events near the fault. It appears that the fault-slip nature of this location was the driving force behind the large events and not the accumulation of stress. The Apparent Stress model displayed this result by having very few time periods which were considered to have an increase in Apparent Stress.

The ASTH model was investigated as an "alarm tool" to identify periods with an increase in seismic hazard and proved to be practical and successful. The regions of Coleman which were highly stressed and free of large geological structure produced a low number of false alarms at roughly one per month. For these regions, the ratio of false alarms to the identification of large seismic events was 2:​1, which is a relatively low ratio of false alarms. Due to the low number of false alarms and decent success rates, the use of Apparent Stress as an "alarm tool" is practical and useful.

Furthermore, the ASTH model was modified to create the ASTH Index to provide additional quantitative interpretation of the Apparent Stress Frequency. It was determined that as the ASTH index increased, the relation between the "occurrence of large events to false alarms increased. This relation increases the confidence of using Apparent Stress to identify periods of increased seismic hazard. This approach, just like the current ASTH model though, relies heavily on the knowledge and experience of the user.