Automated Detection and Predictive Modeling of Flux Transfer Events using CLUSTER Data

Details Automated Detection and Predictive Modeling of Flux Transfer Events using CLUSTER Data Automated Detection and Predictive Modeling of Flux Transfer Events using CLUSTER Data

Dec 2006

adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2006agufmsm51a1391s&link_type=abstract

American Geophysical Union, Fall Meeting 2006, abstract #SM51A-1391

Mathematics

Logic

0520 Data Analysis: Algorithms And Implementation, 0555 Neural Networks, Fuzzy Logic, Machine Learning, 2723 Magnetic Reconnection (7526, 7835), 7526 Magnetic Reconnection (2723, 7835), 7800 Space Plasma Physics

Scientific paper

Almost all statistical studies of flux ropes (FTEs) and traveling compression regions (TCRs) have been based on (i) visual inspection of data to compile a list of events and (ii) use of histograms and simple linear correlation analysis to study their properties and potential causes and dependencies. This approach has several major drawbacks including being highly subjective and inefficient. The traditional use of histograms and simple linear correlation analysis is also only useful for analysis of systems that show dominant dependencies on one or two variables at the most. However, if the system has complex dependencies, more sophisticated statistical techniques are required. For example, Wang et al. [2006] showed evidence that FTE occurrence rate are affected by IMF Bygsm, Bzgsm, and magnitude, and the IMF clock, tilt, spiral, and cone angles. If the initial findings were correct that FTEs occur only during periods of southward IMF, one could use the direction of IMF as a predictor of occurrence of FTEs. But in light of Wang et al. result, one cannot draw quantitative conclusions about conditions under which FTEs occur. It may be that a certain combination of these parameters is the true controlling parameter. To uncover this, one needs to deploy more sophisticated techniques. We have developed a new, sophisticated data mining tool called MineTool. MineTool is highly accurate, flexible and capable of handling difficult and even noisy datasets extremely well. It has the ability to outperform standard data mining tools such as artificial neural networks, decision/regression trees and support vector machines. Here we present preliminary results of the application of this tool to the CLUSTER data to perform two tasks: (i) automated detection of FTEs, and (ii) predictive modeling of occurrences of FTEs based on IMF and magnetospheric conditions.

Affiliated with

Also associated with

No associations

Rating

Automated Detection and Predictive Modeling of Flux Transfer Events using CLUSTER Data does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.

If you have personal experience with Automated Detection and Predictive Modeling of Flux Transfer Events using CLUSTER Data, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Automated Detection and Predictive Modeling of Flux Transfer Events using CLUSTER Data will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFWR-SCP-O-974913