Insights into an ultraslow volcanic spreading centre by an unusual LET: Seismometers mounted on ice floes

• Main Authors: Korger, Edith, Schlindwein, Vera
• Format: Conference Object
• Language: unknown
• Published: 2013
• Subjects: Arctic Ocean
• Online Access: https://epic.awi.de/id/eprint/32206/; https://hdl.handle.net/10013/epic.40845

In 1999 an unusually strong, teleseismically registered earthquake swarm at the volcanic complex at 85°E/85°N marked the onset of a spreading episode at Gakkel ridge with exceptional, deep submarine explosive volcanism. At this site, the rift valley centre hosts several distinctive volcanic features which cluster on ridge-parallel faults or fissures. The rift valley itself is well-defined by 1-2 km high bounding walls. To gain a better understanding of the seismological characteristics of the region, we investigate here three-dimensional P- and S-wave velocity structure in the crust and upper mantle. Three arrays of four seismometers each which were mounted on ice floes, drifted 21 days with a drift speed of up to 1 km/h over a total region of 60 km x 70 km. We recorded over 300 local earthquakes. Of these, we used 128 earthquakes for the local tomography which were recorded by more than two arrays and yielded more than 1200 P- and S-phases. Because of the unusual set up of seismometers on drifting ice floes, each phase was defined by a set of unique station coordinates. Additionally, the unusually variable bathymetry in the survey region had to be taken into account. We therefore defined a first layer comprising vp=1500 m/s with a variable thickness of more than 1100 m in 3D. Due to the remote location and the perennial ice cover of the Arctic ocean, no previous information on crustal structure or local seismicity, either with OBS or active reflection/refraction studies, were available. We adressed the lack of prior information on the velocity structure by localizing a set of well recorded events with HYPOSAT, using a suite of more than 90 randomly selected velocity models. The respective traveltimes were corrected for the varying thickness of the water layer by compensating to a depth of 3900 m. The overall best performing velocity model was then chosen as our 1D starting background model, and was also used to relocate our whole dataset for input source locations. Here we present the results of this seismic ...