Abstract
The modern ‘machine-learning models’ are a section of artificially intelligent machines used to implement complex models, which can learn and improve from experience with respect to certain class of jobs, without being specifically programmed. In the present analysis, a comparative study is made of the popular machine-learning techniques regarding the prediction of auroral activity as reflected by the auroral electrojet index (AE index) during geomagnetically disturbed periods. The study also explores the suitability of the online sequential version of the best machine-learning algorithm, which has the potential for real-time forecast of the AE index from short-time input datasets with extremely fast convergence than batch-training methods. The study discusses the need for the correct choice of the input dataset that can be used for predicting the AE index from several combinations of input datasets which include coupling functions, geomagnetic indices and solar wind parameters. The study reveals that extreme learning machine and its online sequential version are promising models which could predict the AE index extremely fast with a high degree of accuracy even during disturbance periods. The study also shows that the choice of the polar cap index (PC index) as an input parameter is extremely important for an accurate prediction of the AE index.
Similar content being viewed by others
References
Akasofu S I 1975 The roles of the north-south component of the interplanetary magnetic field on large-scale auroral dynamics observed by the DMSP satellite; Planet. Space Sci. 23 1349–1354. https://doi.org/10.1016/0032-0633(75)90030-6.
Akasofu S I 2004 Several ‘controversial’ issues on substorms; Space Sci. Rev. 113 1–40.
Alexeev I I, Belenkaya E S and Kalegaev V V et al. 1996 Magnetic storms and magnetotail currents; J. Geophys. Res. Phys. 101 7737–7747, https://doi.org/10.1029/95ja03509.
Amariutei O A and Ganushkina N Y 2012 On the prediction of the auroral westward electrojet index; Ann. Geophys. 30 841–847, https://doi.org/10.5194/angeo-30-841-2012.
Amata E, Consolini G, Pallocchia G and Marcucci M 2009 ANN forecast of hourly averaged AE index based on L1 IMF and plasma measurements; Acta Geophys. 57 185–196, https://doi.org/10.2478/s11600-008-0083-1.
Anastasiadis A D, Magoulas G D and Vrahatis M N 2005 New globally convergent training scheme based on the resilient propagation algorithm; Neurocomputing 64 253–270, https://doi.org/10.1016/j.neucom.2004.11.016.
Anderson J A, Rosenfeld E and Pellionisz A 1988 Neurocomputing; MIT Press, Cambridge, MA.
Baker D N, McPherron R L, Cayton T E and Klebesadel R W 1990 Linear prediction filter analysis of relativistic electron properties at 6.6 RE; J. Geophys. Res. 95 15,133–15,140, https://doi.org/10.1029/JA095iA09p15133.
Bargatze L F, Baker D N, Hones E W and Mcpherron R L 1985 Magnetospheric impulse response for many levels of geomagnetic activity; J. Geophys. Res. 90 6387–6394, https://doi.org/10.1029/JA090iA07p06387.
Boberg F, Wintoft P and Lundstedt H 2000 Real time Kp predictions from solar wind data using neural networks; Phys. Chem. Earth 25 275–280, https://doi.org/10.1016/S1464-1917(00)00016-7.
Breiman L 1996 Bagging predictors; Mach. Learn. 24 123–140, https://doi.org/10.1007/BF00058655.
Breiman L 2001 Random forests; Mach. Learn. 45 5–32, https://doi.org/10.1023/A:1010933404324.
Breiman L, Friedman J H, Olshen R A and Stone C J 1984 Classification and regression trees; Chapman & Hall/CRC, Boca Raton.
Cai L, Ma S Y and Zhou Y L 2010 Prediction of SYM-H index during large storms by NARX neural network from IMF and solar wind data; Ann. Geophys. 28 381–393, https://doi.org/10.5194/angeo-28-381-2010.
Clauer C R and Kamide Y 1985 DP 1 and DP 2 current systems for the March 22, 1979 substorms; J. Geophys. Res. 90 1343, https://doi.org/10.1029/JA090iA02p01343.
Clauer C R, McPherron R L, Searls C and Kivelson M G 1981 Solar wind control of auroral zone geomagnetic activity; Geophys. Res. Lett. 8 915–918, https://doi.org/10.1029/GL008i008p00915.
Dietterich T G 2000 Experimental comparison of three methods for constructing ensembles of decision trees: Bagging, boosting, and randomization; Mach. Learn. 40 139–157, https://doi.org/10.1023/A:1007607513941.
Drucker H, Burges C J C and Kaufman L et al. 1997 Support vector regression machines; Adv. Neural. Inf. Process. Syst. 9 155–161.
Friedrich E, Rostoker G, Connors M G and McPherron R L 1999 Influence of the substorm current wedge on the Dst index; J. Geophys. Res. Phys. 104 4567–4575, https://doi.org/10.1029/1998ja900096.
Gannon J L and Love J J 2011 USGS 1-min dst index; J. Atmos. Sol.-Terr. Phys. 73 323–334, https://doi.org/10.1016/j.jastp.2010.02.013.
Gavrishchaka V V and Ganguli S B 2001 Support vector machine as an efficient tool for high-dimensional data processing: Application to substorm forecasting; J. Geophys. Res. Sp. Phys. 106 29,911–29,914, https://doi.org/10.1029/2001JA900118.
Geurts P, Ernst D and Wehenkel L 2006 Extremely randomized trees; Mach. Learn. 63 3–42, https://doi.org/10.1007/s10994-006-6226-1.
Gleisner H and Lundstedt H 1997 Response of the auroral electrojets to the solar wind modeled with neural networks; J. Geophys. Res. 102 14269, https://doi.org/10.1029/96JA03068.
Gleisner H and Lundstedt H 1999 Ring current influence on auroral electrojet predictions; Ann. Geophys. 17 1268.
Gleisner H and Lundstedt H 2001 Auroral electrojet predictions with dynamic neural networks; J. Geophys. Res. 106 24,541–24,549, https://doi.org/10.1029/2001ja900046.
Gopinath S, Suji K J and Prince P R 2015 Information measures based analysis of complex solar wind-magnetosphere interaction dynamics during geomagnetic storms; Indian J. Phys. 89(8) 759–772, https://doi.org/10.1007/s12648-015-0655-2.
Hammerstrom D 1993 Working with neural networks; IEEE Spectr. 30 46–53, https://doi.org/10.1109/6.222230.
Hernandez J V, Tajima T and Horton W 1993 Neural net forecasting for geomagnetic activity; Geophys. Res. Lett. 20 2707–2710, https://doi.org/10.1029/93GL02848.
Huang G-B, Zhu Q and Siew C et al. 2006 Extreme learning machine: Theory and applications; Neurocomputing 70 489–501, https://doi.org/10.1016/j.neucom.2005.12.126.
Huang G B, Wang D H and Lan Y 2011 Extreme learning machines: A survey; Int. J. Mach. Learn. Cybern. 2 107–122, https://doi.org/10.1007/s13042-011-0019-y.
Huang G, Huang G B, Song S and You K 2015 Trends in extreme learning machines: A review; Neural Netw. 61 32–48.
Iyemori T, Maeda H and Kamei T 1979 Impulse response of geomagnetic indices to interplanetary magnetic field; J. Geomagn. Geoelectr. 31 1–9.
Jia X, Wang R, Liu J and Powers D M W 2016 A semi-supervised online sequential extreme learning machine method; Neurocomputing 174 168–178, https://doi.org/10.1016/j.neucom.2015.04.102.
King J H and Papitashvili N E 2005 Solar wind spatial scales in and comparisons of hourly wind and ACE plasma and magnetic field data; J. Geophys. Res. 110 A02104, https://doi.org/10.1029/2004JA010649.
Klimas A J, Baker D N and Roberts D A et al. 1992 A nonlinear dynamical analogue model of geomagnetic activity; J. Geophys. Res. 97 12253, https://doi.org/10.1029/92JA00794.
Klimas A J, Vassiliadis D and Baker D N 1997 Data-derived analogues of the magnetospheric dynamics; J. Geophys. Res. 102 26993.
Kugblenu S, Taguchi S and Okuzawa T 1999 Prediction of the geomagnetic storm associated Dst index using an artificial neural network algorithm; Earth Planet. Space 51 307–313.
Lazzús J A, López-Caraballo C H, Rojas P, Salfate I, Rivera M and Palma-Chilla L 2016 Forecasting of Dst index from auroral electrojet indices using time-delay neural network + particle swarm optimization; J. Phys. Conf. Ser. 720 012001.
Liang N-Y, Huang G-B, Saratchandran P and Sundararajan N 2006 A fast and accurate online sequential learning algorithm for feedforward networks; IEEE Trans. Neural Netw. 17 1411–1423, https://doi.org/10.1109/TNN.2006.880583.
Liemohn M W, Kozyra J U and Thomsen M F et al. 2001 Dominant role of the asymmetric ring current in producing the stormtime Dst; J. Geophys. Res. Sp. Phys. 106 10,883–10,904, https://doi.org/10.1029/2000JA000326.
Liu H L H 2010 On the Levenberg–Marquardt training method for feed-forward neural networks; In: Natural computation (ICNC), 2010 sixth international conference 1, https://doi.org/10.1109/ICNC.2010.5583151.
Lundstedt H and Wintoft P 1994 Prediction of geomagnetic storms from solar wind data with the use of a neural network; Ann. Geophys. 12 19–24, https://doi.org/10.1007/s00585-994-0019-2.
McPherron R L, Baker D N and Bargatze L F et al. 1988 IMF control of geomagnetic activity; Adv. Sp. Res. 8 71–86, https://doi.org/10.1016/0273-1177(88)90114-7.
Munsami V 2000 Determination of the effects of substorms on the storm-time ring current using neural networks; J. Geophys. Res. 105, https://doi.org/10.1029/2000ja000041.
Pallocchia G, Amata E and Consolini G et al. 2008 AE index forecast at different time scales through an ANN algorithm based on L1 IMF and plasma measurements; J. Atmos. Solar-Terr. Phys. 70 663–668, https://doi.org/10.1016/j.jastp.2007.08.038.
Pavlov Y L 1997 Random forests; In: Probabilistic methods in discrete mathematics, (eds) Kolchin V F, Kozlov V Y, Pavlov Y L and Prokhorov Y V, Proc. 4th International Petrozavodsk Conf. 1996, VSP Publishers, Utrecht, The Netherlands, pp. 11–18.
Price C P and Prichard D 1993 The non-linear response of the magnetosphere: 30 October 1978; Geophys. Res. Lett. 20 771–774, https://doi.org/10.1029/93GL00844.
Revallo M, Valach F, Hejda P and Bochníček J 2014 A neural network Dst index model driven by input time histories of the solar wind-magnetosphere interaction; J Atmos. Sol.-Terr. Phys. 110–111 9–14, https://doi.org/10.1016/j.jastp.2014.01.011.
Riedmiller M 1994 Advanced supervised learning in multi-layer perceptrons – From backpropagation to adaptive learning algorithms; Comput. Stand. Interfaces 16 265–278, https://doi.org/10.1016/0920-5489(94)90017-5.
Riedmiller M and Braun H 1993 A direct adaptive method for faster backpropagation learning: The RPROP algorithm; In: IEEE international conference on neural networks – Conference proceedings, pp. 586–591.
Rostoker G, Akasofu S I, Baumjohann W, Kamide Y and McPherron R L 1987 The roles of direct input of energy from the solar wind and unloading of stored magnetotail energy in driving magnetospheric substorms; Space. Sci. Rev. 46 93.
Rostoker G, Akasofu S I and Baumjohann W et al. 1988 The roles of direct input of energy from the solar wind and unloading of stored magnetotail energy in driving magnetospheric substorms; Space. Sci. Rev. 4 93–111, https://doi.org/10.1007/BF00173876.
Rumelhart D E and McClelland J L 1986 Parallel distributed processing; MIT Press, Cambridge, MA.
Stepanova M V and Pérez P 2000 Autoprediction of Dst index using neural network techniques and relationship to the auroral geomagnetic indices; Geofís. Int. 39 143–146.
Takalo J and Timonen J 1997 Neural network prediction of AE data; Geophys. Res. Lett. 24 2403–2406.
Takalo J and Timonen J 1999 Neural network prediction of the AE index from the PC index; Phys. Chem. Earth, C, Sol. Terr. Planet Sci. 24 89–92, https://doi.org/10.1016/S1464-1917(98)00013-0.
Troshichev O and Janzhura A 2012 Space weather monitoring by ground-based means: PC index; Springer-Verlag, Berlin, Heidelberg, 288p. https://doi.org/10.1007/978-3-642-16803-1.
Unnikrishnan K 2014 Prediction of horizontal component of earth’s magnetic field over Indian sector using neural network model; J. Atmos. Sol.-Terr. Phys. 121 206–220, https://doi.org/10.1016/j.jastp.2014.06.014.
Vapnik V 1998 The support vector method of function estimation; In: Nonlinear modeling: Advanced Black-Box Techniques (eds) Suykens J A K and Vandewalle J, Kluwer Academic Publishers, Boston, pp. 55–85.
Vapnik V and Chapelle O 2000 Bounds on error expectation for support vector machines; Neural. Comput. 12 2013–2036, https://doi.org/10.1162/089976600300015042.
Vassiliadis D, Klimas A J, Baker D N and Roberts D A 1995 A description of the solar wind-magnetosphere coupling based on nonlinear filters; J. Geophys. Res. 100 3495, https://doi.org/10.1029/94ja02725, ISSN 0148-0227.
Weigel R S, Horton W, Tajima T and Detman T 1999 Forecasting auroral electrojet activity from solar wind input with neural networks; Geophys. Res. Lett. 26 1353–1356, https://doi.org/10.1029/1999gl900280.
Weygand J M and McPherron R L 2006 Dependence of ring current asymmetry on storm phase; J. Geophys. Res. Sp. Phys. 111 A11221 2156–2202, https://doi.org/10.1029/2006JA011808.
Acknowledgements
The authors are thankful to the World Data Center for Geomagnetism, Kyoto, Japan for providing AE, SYM and ASY data (http://wdc.kugi.kyoto-u.ac.jp/) and to USGS for providing 1-min Dst index data (http://geomag.usgs.gov/). The authors thank the World Data Center for Geomagnetism, Copenhagen (maintained by Technical University of Denmark) for providing 1-min PC index (ftp://ftp.space.dtu.dk/WDC/). The authors thank the OMNIWeb for providing the solar wind data (http://omniweb.gsfc.nasa.gov/). The authors thank C C Chang and C J Lin for providing the LibSVM package (https://www.csie.ntu.edu.tw/~cjlin/libsvm/), A Liaw and M Wiener for the Random Forest package (http://cran.r-project.org/web/packages/randomForest/), G B Huang for providing the ELM and OS-ELM codes (http://www3.ntu.edu.sg/home/egbhuang/elm_codes.html) and D Atabay for the pyrenn package (https://pyrenn.readthedocs.io). One of the authors, SG, extends his regards to Dr Madhu S Nair, assistant professor, Department of Computer Science, University of Kerala and to Dr K Satheesh Kumar, associate professor, Department of Future Studies, University of Kerala for their help and support. He also acknowledges a junior research fellowship (Ac.EVI(4)/17465/JRF/2017) from the University of Kerala, Trivandrum.
Author information
Authors and Affiliations
Corresponding author
Additional information
Corresponding Editor: Amit Kumar Patra
Rights and permissions
About this article
Cite this article
Gopinath, S., Prince, P.R. A comparison of machine-learning techniques for the prediction of the auroral electrojet index. J Earth Syst Sci 128, 172 (2019). https://doi.org/10.1007/s12040-019-1194-6
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s12040-019-1194-6