Earthquakes track subduction fluids from slab source to mantle wedge sink.

Subducting plates release fluids as they plunge into Earth's mantle and occasionally rupture to produce intraslab earthquakes. It is debated whether fluids and earthquakes are directly related. By combining seismic observations and geodynamic models from western Greece, and comparing across other subduction zones, we find that earthquakes effectively track the flow of fluids from their slab source at >80 km depth to their sink at shallow (<40 km) depth. Between source and sink, the fluids flow updip under a sealed plate interface, facilitating intraslab earthquakes. In some locations, the seal breaks and fluids escape through vents into the mantle wedge, thereby reducing the fluid supply and seismicity updip in the slab. The vents themselves may represent nucleation sites for larger damaging earthquakes.

Ambient noise levels and detection threshold in Norway.

Ambient seismic noise is caused by a number of sources in specific frequency bands. The quantification of ambient noise makes it possible to evaluate station and network performance. We evaluate noise levels in Norway from the 2013 data set of the Norwegian National Seismic Network as well as two temporary deployments. Apart from the station performance, we studied the geographical and temporal variations, and developed a local noise model for Norway. The microseism peaks related to the ocean are significant in Norway. We, therefore, investigated the relationship between oceanic weather conditions and noise levels. We find a correlation of low-frequency noise (0.125-0.25 Hz) with wave heights up to 900 km offshore. High (2-10 Hz) and intermediate (0.5-5 Hz) frequency noise correlates only up to 450 km offshore with wave heights. From a geographic perspective, stations in southern Norway show lower noise levels for low frequencies due to a larger distance to the dominant noise sources in the North Atlantic. Finally, we studied the influence of high-frequency noise levels on earthquake detectability and found that a noise level increase of 10 dB decreases the detectability by 0.5 magnitude units. This method provides a practical way to consider noise variations in detection maps.