The objective of my research is to provide a better understanding of the relationship between the micro-seismicity, tectonics and crustal structure in southwest Yukon in order to improve seismic hazard assessments in this region. I used a combination of single event and multiple event location techniques to determine earthquake locations and depths. As well, frequency-magnitude statistics were calculated to analyze rates of seismicity and possible changes in the rates of seismicity. The addition of the YUK array in August 2010 has enabled location of smaller events and detection of a systematic northeast trend of earthquakes. Seismicity is concentrated in four main areas: 1) Yaktutat Block-Fairweather Fault, 2) Duke River Fault, 3) Denali Fault, and 4) a NE-trend. There was relatively little seismic activity during this period along the northern Denali Fault segment and only a small amount of activity along the southern portion of the Denali Fault. There is significantly more seismic activity along the Duke River Fault and NE-trend and a clear region of seismicity just west and parallel to the Alaska-Yukon border between the Duke River Fault and northern Denali Fault. Frequency-magnitude statistics and seismic hazard analyses for southwest Yukon were improved by decreasing the minimum magnitude of completeness from M3.0 to M1.0. Between September 2010 and November 2011, event magnitudes ranged from 0.2 to 4.7 and depths from 0 to 35 km. To address how the YUK array has improved single event locations and depths, we use a single-event location technique to monitor seismic activity. Only 37 of the 106 events detected for the Duke River Fault and NE-trend could potentially be located without the YUK array. When the Alaska Earthquake Information Center (AEIC) network was combined with the Canadian National Seismograph Network (CNSN), events within the NE-trend shift on average 6.6 km to the northeast and the depth increased on average 2.6 km. Within the Duke River and NE-trending clusters, there is an average maximum horizontal error of ±0.9 km and an average error in depth of ±3.2 km. Free depths in the Duke River and NE-trending clusters range from 0 to 20 km. These depths are not well-constrained as the closest station is more than 20 km away. Two events within the southern Denali Fault cluster have well-constrained depths of 4.8 km and 8.2 km at distance less than ~8 km from station YUK6, consistent with upper crust (2-10 km) focal depths. A Progressive Multiple Event Location technique (PMEL) was used to identify and better constrain spatial patterns along the Duke River Fault and NE-trend. Results clearly shows that events fall along the Duke River Fault and that the NE-trend events are located on a previously unidentified active fault. To determine rates of seismicity and possible changes in the rates of seismicity, I examine b-values from frequency-magnitude statistics for each cluster of earthquakes before and after the 2002 M7.9 Denali Fault earthquake. b-values increased from 0.81 ± 0.14 to 1.05 ± 0.22 , suggesting higher Coulomb stress and more frequent smaller earthquakes. / Graduate

<p>The potential of variations of <i>b-</i>values in the G-R relation, log<i>N=a-bM</i> as earthquake precursors for small events (rockbursts) in Zinkgruvan mine, Sweden and for tectonic (large) earthquakes in the Andaman-Sumatra region were investigated. </p><p>The temporal frequency-magnitude distribution, <i>b(t)</i>, of rockbursts in Zinkgruvan mine was examined using high quality data recorded during the period November 1996 to April 2004 with magnitude ranges from <i>M</i><i>w</i>= -2.4 to 2.6. A sliding time-window was applied to compute <i>b</i>-values. The windows contain 50 events and were shifted with steps of 5 events. The results indicated that <i>b</i>-values significantly drop preceding rockbursts of magnitude <i>M</i><i>w</i>≥1.6.</p><p>Temporal and spatial variations of <i>b</i>-values were also examined for tectonic earthquakes, magnitude <i>M</i><i>w</i>≥4.1, in the Andaman-Sumatra region. Earthquake data from the ISC, IDC, NEIC and HVRD earthquake catalogs for a period from 01/01/1995 to 12/26/2004 were used for analysis. Spatial variations of <i>b</i> were calculated from circular areas containing 50 events, with nodes on a 0.5° x 0.5° grid. The analysis shows that <i>b(t)</i> estimates using data from different catalogs<i> </i>are comparable and that large earthquakes are preceded by a drop in <i>b(t)</i> of about 0.3~1. The distribution of stress deduced from <i>b</i>-value mapping shows that large earthquakes occurred in the high stress, i.e. low <i>b</i>-value, areas.</p><p>Aftershock sequences of the <i>M</i><i>w</i>=9, December 26, 2004 and the <i>M</i><i>w</i>=8.7, March 28, 2005 shocks were investigated by using the same methods. Results from aftershock sequences show similar behaviour as for the large and presumed independent main events.</p><p>The observed variations of <i>b</i>-values with time and in space support the hypothesis that <i>b</i>-values have a precursory potential. The method can be used for a wide range of earthquake magnitude, from microearthquakes (<i>M</i><i>w</i><3) to giant tectonic shocks (<i>M</i><i>w</i>~9) and for both of independent shocks and aftershocks.</p>

The potential of variations of b-values in the G-R relation, logN=a-bM as earthquake precursors for small events (rockbursts) in Zinkgruvan mine, Sweden and for tectonic (large) earthquakes in the Andaman-Sumatra region were investigated. The temporal frequency-magnitude distribution, b(t), of rockbursts in Zinkgruvan mine was examined using high quality data recorded during the period November 1996 to April 2004 with magnitude ranges from Mw= -2.4 to 2.6. A sliding time-window was applied to compute b-values. The windows contain 50 events and were shifted with steps of 5 events. The results indicated that b-values significantly drop preceding rockbursts of magnitude Mw≥1.6. Temporal and spatial variations of b-values were also examined for tectonic earthquakes, magnitude Mw≥4.1, in the Andaman-Sumatra region. Earthquake data from the ISC, IDC, NEIC and HVRD earthquake catalogs for a period from 01/01/1995 to 12/26/2004 were used for analysis. Spatial variations of b were calculated from circular areas containing 50 events, with nodes on a 0.5° x 0.5° grid. The analysis shows that b(t) estimates using data from different catalogs are comparable and that large earthquakes are preceded by a drop in b(t) of about 0.3~1. The distribution of stress deduced from b-value mapping shows that large earthquakes occurred in the high stress, i.e. low b-value, areas. Aftershock sequences of the Mw=9, December 26, 2004 and the Mw=8.7, March 28, 2005 shocks were investigated by using the same methods. Results from aftershock sequences show similar behaviour as for the large and presumed independent main events. The observed variations of b-values with time and in space support the hypothesis that b-values have a precursory potential. The method can be used for a wide range of earthquake magnitude, from microearthquakes (Mw&lt;3) to giant tectonic shocks (Mw~9) and for both of independent shocks and aftershocks.

This thesis examines the advantages to nonlinear least-squares (NLS) fitting of diffusion-weighted MRI data over the commonly used linear least-squares (LLS) approach. A modified fitting algorithm is proposed which accounts for the positive bias experienced in magnitude images at low SNR. For b-values in the clinical range (~1000 s/mm2), the increase in precision of FA and fibre orientation estimates is almost negligible, except at very high anisotropy. The optimal b-value for estimating tensor parameters was slightly higher for NLS. The primary advantage to NLS was improved performance at high b-values, for which complex fibre architectures were more easily resolved. This was demonstrated using a model-selection classifier based on higher-order diffusion models. Using a b-value of 3000 s/mm2 and magnitude-corrected NLS fitting, at least 10% of voxels in the brain exhibited diffusion profiles which could not be represented by the tensor model.

This thesis examines the advantages to nonlinear least-squares (NLS) fitting of diffusion-weighted MRI data over the commonly used linear least-squares (LLS) approach. A modified fitting algorithm is proposed which accounts for the positive bias experienced in magnitude images at low SNR. For b-values in the clinical range (~1000 s/mm2), the increase in precision of FA and fibre orientation estimates is almost negligible, except at very high anisotropy. The optimal b-value for estimating tensor parameters was slightly higher for NLS. The primary advantage to NLS was improved performance at high b-values, for which complex fibre architectures were more easily resolved. This was demonstrated using a model-selection classifier based on higher-order diffusion models. Using a b-value of 3000 s/mm2 and magnitude-corrected NLS fitting, at least 10% of voxels in the brain exhibited diffusion profiles which could not be represented by the tensor model.