Induced seismicity in the Castarthao reservoir, NE Brazil - Preliminary results

The Castanhao reservoir was built in the state of Ceara, a dry region in Northeastern Brazil, to regulate the flow of the Jaguaribe River, for irrigation, and for power generation. It is an earth-filled dam, 60 m high, with a water capacity of 4.5 x 10(9) m(3). The seismicity in the area has been monitored since 1998, with a few interruptions, using one analog or one digital station and, during a few periods, a three-station network. The first earthquakes likely to be induced events were detected in 2003, when the water level was about 20 in high. In early 2004 a very heavy rainfall season quickly filled the reservoir. Shortly after, an increase in the seismic activity occurred and many micro-earthquakes were recorded. We suggest that this activity resulted from an increase in pore pressure due to undrained response. Therefore, we may classify this cluster of microearthquakes as ""initial seismicity."" We deployed a network with four analog stations in the area, following this activity, to determine the epicentral zone. At least three epicentral areas under the reservoir were detected. The spatio-temporal analysis of the available data revealed that the seismicity occurs in clusters and that these were activated at different periods. We identified four sets of faults (N-S-, E-W-, NW-SE-, and NE-SW-oriented), some of which moved in shallow crustal levels and as recently as the Quaternary (1.8 Ma). Under the present-day stress regime, the last two sets moved as strike-slip structures. We suggest a possible correlation between dormant faults and the observed induced seismicity. (c) 2008 Elsevier B.V. All rights reserved.

A multi-GPU algorithm for large-scale neuronal networks

Large-scale simulations of parts of the brain using detailed neuronal models to improve our understanding of brain functions are becoming a reality with the usage of supercomputers and large clusters. However, the high acquisition and maintenance cost of these computers, including the physical space, air conditioning, and electrical power, limits the number of simulations of this kind that scientists can perform. Modern commodity graphical cards, based on the CUDA platform, contain graphical processing units (GPUs) composed of hundreds of processors that can simultaneously execute thousands of threads and thus constitute a low-cost solution for many high-performance computing applications. In this work, we present a CUDA algorithm that enables the execution, on multiple GPUs, of simulations of large-scale networks composed of biologically realistic Hodgkin-Huxley neurons. The algorithm represents each neuron as a CUDA thread, which solves the set of coupled differential equations that model each neuron. Communication among neurons located in different GPUs is coordinated by the CPU. We obtained speedups of 40 for the simulation of 200k neurons that received random external input and speedups of 9 for a network with 200k neurons and 20M neuronal connections, in a single computer with two graphic boards with two GPUs each, when compared with a modern quad-core CPU. Copyright (C) 2010 John Wiley & Sons, Ltd.