Paper#: 137
Poster #: 1
Session Name: Poster Session 1
Room: East Ballroom
Day: Tuesday
Time: 10:10 a.m.-Noon
Abstract Title: New Generation of Models for Substorm Forecasting from Multi-Scale and High-Dimensional Data.
PresentSurname: Ganguli, S.B.
All Authors: S.B. Ganguli, V.V. Gavrishchaka
Abstract : One of the most challenging problems in space weather forecasting is substorm prediction. Capabilities of the data-driven substorm models depends on the quality and amount of the available data and on the algorithms used to extract generalized mappings. Availability of the real-time high-resolution solar wind, magnetospheric and ionospheric data constantly increases. However, the majority of advanced nonlinear algorithms, including neural networks (NN), can encounter a set of problems called dimensionality curse when applied to high-dimensional data. Nonstationarity of the system can also impose significant limitations on the size of training set which leads to poor generalization ability of the model. A very promising algorithm that combines the power of the best nonlinear techniques and tolerance to high-dimensional and incomplete data is support vector machine (SVM). We have summarized and demonstrated advantages of the SVM by applying it to substorm forecasting from solar wind data. We have shown that performance of the SVM model for substorm prediction can be comparable to or be superior to that of the best existing models including NNs. The advantages of the SVM-based techniques are expected to be much more pronounced in future space-weather forecasting models which will incorporate many types of high-dimensional and multi-scale input data once real-time availability of this information becomes technologically feasible.