A seismic study of the Krafla volcanic system, Iceland
Following a major crustal rifting episode centred on the Krafla volcanic system in Northeast Iceland, the local seismicity was monitored for three months using a dense network of single component seismometers. Initial earthquake locations were computed for 489 local events using a one-dimensional velocity model derived from seismic refraction data. Activity was concentrated in clusters beneath a geothermal area within the Krafla caldera and below the Bjarnarflag geothermal well field to the south, and in a narrow linear zone coincident with a site of recent dyke injection. Events of magnitudes -2.4 to 2.1 were located. The b-value for the entire dataset is 0.77±0.10, and is lower for events at Bjarnarflag than for events in the dyke injection zone. Seismicity was continuous. A simultaneous inversion of 1771 P-wave arrival times was carried out to calculate the 3-d velocity structure and refine hypocentral locations. The derivative weight sum was used to identify the well-resolved volume. High velocity bodies at depth beneath the rim of the caldera are high density intrusives, probably gabbros. Volumes of low velocity coincide with zones of geothermal exploitation within the caldera and beneath Bjarnarflag, and result from intense fracturing and hydrothermal alteration. After relocation through the three-dimensional velocity structure, the spatial distribution of hypocentres is more focussed and considerably shallower than before. This led to a closer correlation between zones of seismicity and detailed features within the source volumes, such as geothermal reservoirs, fault surfaces and zones of known recent magmatic intrusion. Focal mechanisms were determined using P-wave polarity data for 153 of the best located earth quakes, of which 139 have double couple solutions. Fourteen events were non-double couple, 4 of which could be solved as either opening or closing tensile cracks assuming small circle nodal lines. The biasing effect of using incorrect hypocentres and an over-simplified velocity structure were investigated by 3-d ray tracing. The effects on ray angles are large and variable. The impact of these effects on double couple solutions is generally small, but can be critical for the confident identification of non-double couple events. Seismicity in geothermal areas results from cooling and fracturing of hot intrusives at depth, and fault surfaces are marked by enhanced seismicity when they provide efficient migratory paths for geothermal fluids. Seismicity may be induced partially by geothermal mining. The stress field in the Krafla volcanic system at the time of the survey was variable along its length, as was the mode of fracturing. The stress field at Bjarnarflag was chaotic. In the geothermal area within the caldera σ(_3) was perpendicular to the plate boundary, and in the dyke injection zone σ(_1) was perpendicular to the plate boundary. Such heterogeneity in the stress field is contrary to what might be expected at a spreading axis, and is attributed to variation in regional extensional stress release during the recent rifting episode.