Prediction of the geomagnetic storm associated Dst index using an artificial neural network algorithm

Kugblenu, S.; Taguchi, S.; Okuzawa, Takashi

doi:10.1186/bf03352234

Cited by 36 publications

(42 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Wu and Lundstedt (1997) made an Elman recurrent neural network model for the D st variations from IMF B z and the solar wind number density and speed. Although discrepancy between the model and observation sometimes occurs in the recovery phase, and for the accurate reproduction of this recovery phase some arrangement may be needed (Kugblenu et al, 1999), the model of Wu and Lundstedt (1997) showed a remarkably good performance; i.e., the correlation coefficient between the observed D st and the modeled D st is 0.91. Higher correlation coefficient has been reached by the neural network model by Kugblenu et al (1999) The AL index representing substorm activities has been also used as a prediction target, and reasonably good reproduction has been reported, for example, by McPherron (1997), and Gleisner and Lundstedt (1997).…”

Section: Introductionmentioning

confidence: 99%

“…Such accurate models have been enabled by recent advances in linear/non-linear prediction filter (e.g., McPherron, 1997;Klimas et al, 1998) and in artificial neural network (Wu and Lundstedt, 1997;Kugblenu et al, 1999). In the linear prediction analysis McPherron (1997) showed that 85% of the D st variance is accounted for by the solar wind dynamic pressure and a coupling function expressed by the solar wind speed, IMF B z , and B y .…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Consequences of the neural network investigation for Dst-AL relationship

2014

View full text Add to dashboard Cite

Several recent studies have suggested that most of the D st main phase variations and of the AL variations similarly respond to a certain type of solar wind condition although the processes are independent of each other. This similarity suggests that some consistency between the D st main phase development and AL variations exists, regardless of the existence of causality. In what situations this consistent relationship really exists or collapses has been examined with the technique of an Elman recurrent neural network. The network was trained with the D st and hourly averaged AL indices for 70 storm events from 1967 to 1981, and tested for nine storms that occurred in 1982. The result shows that the D st -AL relationship can be categorized into two types: high correlative mapping for which 80% and more of the D st peak in the main phase is reproduced by AL, and partially correlative mapping where only about a half of the D st peak is reproduced. It is found that whether the correlation is high or partial is determined by whether the D st main phase develops smoothly or with a discontinuity, i.e., for storms having a discontinuity in the main phase, the coherency collapses. The discontinuity in the D st main phase is associated with the rapid southward IMF change after the northward excursion. We suggest that it is this IMF variation to which storms and/or substorms respond in a highly complex manner and that such a complex response can be associated with about a half of the maximum ring current intensity.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Consequences of the neural network investigation for Dst-AL relationship

2014

View full text Add to dashboard Cite

show abstract

“…Many neural and neurofuzzy methods like Multi Layer Perceptron (MLP) neural network (Kugblenu et al, 1999) with back propagation learning algorithm (BP), Radial Basis Function (RBF) neural networks with orthogonal least square (OLS) for center learning of RBFs and recently Adaptive Network-Based Fuzzy Inference System (AN-FIS) are widely used for nonlinear system identification and also for prediction of chaotic time series. In this section we utilize locally linear neurofuzzy models with Lofunction of LLM that is split is change but the validity functions that are the normalized version of membership functions (Eqs.…”

Section: One-step Ahead Prediction Of Dst Indexmentioning

confidence: 99%

Multi-step prediction of Dst index using singular spectrum analysis and locally linear neurofuzzy modeling

2006

View full text Add to dashboard Cite

Even one-step prediction of natural time series without delay especially in main phase of storm is difficult for many complicated time series such as Dst index. In this study, with a new method based on singular spectrum analysis, we extract the main components of the time series, model each component with a locally linear neurofuzzy network, and utilize the trained networks for multi-step ahead prediction of a validation set of data, and finally combine the predicted patterns for construction of general prediction. Our methods are compared with several previous studies for Dst index prediction. Several solar geomagnetic extreme events are predicted well with our state-of-the-art method; such as extreme events in 14 March 1989 that led to power black-out in Quebec, as well as other extreme storms.

show abstract

“…For the input parameters, these authors used the solar wind density N and velocity V, the magnitude of the interplanetary magnetic field (IMF), and the By and Bz components of IMF. Kugblenu et al (1999) used a multi-layer feed-forward error back-propagation algorithm in their study and obtained good results considering that the training time series consisted of only 20 storms.…”

Section: Introductionmentioning

confidence: 99%

Prediction of the Dst index from solar wind parameters by a neural network method

Watanabe¹,

Sagawa²,

Ohtaka³

et al. 2014

Earth Planet Sp

View full text Add to dashboard Cite

Using the Elman-type neural network technique, operational models are constructed that predict the Dst index two hours in advance. The input data consist of real-time solar wind velocity, density, and magnetic field data obtained by the Advanced Composition Explorer (ACE) spacecraft since May 1998 (http://www2.crl.go.jp/uk/uk223/service/nnw/index.html). During the period from February to October 1998, eleven storms occurred with minimum Dst values below −80 nT. For ten of these storms the differences between the predicted minimum Dst and the minimum Dst calculated from ground-based magnetometer data were less than 23%. For the remaining one storm (beginning on 19 October 1998) the difference was 48%. The discrepancy is likely to stem from a imperfect correlation between the solar wind parameters near ACE and those near the earth. While the IMF Bz remains to be the most important parameter, other parameters do have their effects. For instance, Dst appears to be enhanced when the azimuthal direction of IMF is toward the sun. A trapezoid-shaped increase in the solar wind density enhances the main phase Dst by almost 10% compared with the case of no density increase. Velocity effects appear to be stronger than the density effects. Our operational models have, in principle, no limitations in applicability with respect to storm intensity.

show abstract

Prediction of the geomagnetic storm associated Dst index using an artificial neural network algorithm

Cited by 36 publications

References 22 publications

Consequences of the neural network investigation for Dst-AL relationship

Consequences of the neural network investigation for Dst-AL relationship

Multi-step prediction of Dst index using singular spectrum analysis and locally linear neurofuzzy modeling

Prediction of the Dst index from solar wind parameters by a neural network method

Contact Info

Product

Resources

About