Parallel Computation of Spatio-Temporal Variations of Geomagnetic Field Using One-Minute Time Resolution Dataset

Full Text PDF PDF
Author(s) Felix Ale | Oye Ibidapo-Obe | Theophilus A. Fashanu | Olufemi A. Agboola
Pages 1842-1853
Volume 2
Issue 11
Date November, 2012
Keywords Parallel Computing, Multi-core technology, Geomagnetic Field, Performance, Benchmark, Space Weather

Abstract

This work presents a novel method for analyzing spatio-temporal variations of geomagnetic field using parallel computing. One-minute resolution geomagnetic dataset of 1996 was obtained from INTERMAGNET global network of 64 observatory stations. In effect, large and three dimensional arrays of geomagnetic observations are to be processed. Thus, sequential and parallel algorithms were developed using MATLAB 2012a, interfaced with the well-known kriging method for efficient geostatistical data gridding and mapping of solar quiet (Sq) daily variations. The runtime of the sequential algorithm on a processor took 18.5 minutes while the corresponding parallel algorithm took 2.95 minutes using eight Intel Xeon E5410 2.33GHz processors in parallel. The efficiency profile of the model is logarithmic in nature. This was further optimized using quadratic polynomial interpolation. The results show that 13 processors will process the Sq in less than one minute, thus providing effective near real-time observation of Space weather. In addition, the foci of the generated finer Sq(H) plots revealed temporal variability that is consistently maximized at local noon at every location on this geomagnetic environment as demonstrated in the 2D visual display.

< Back to November Issue