1. Study of the Effect of Auroral Substorms on the Low - Latitude Currents Inga Maslova 1, Piotr Kokoszka 1, Jan Sojka 2, Lie Zhu 2 1 Mathematics and Statistics Department, 2 Center for Atmospheric and Space Sciences, Utah State University AbstractMethods In recent years there has been some discussions on whether or not the auroral substorms have an effect on the equatorial and mid-latitude currents. Our work was motivated by this problem. We use a novel statistical technique that allows us to test at a specified significance level whether such an effect exists. Our approach uses the original 1-min data of H component measurements, rather than the derived indices as in some previous related studies. One of the shortcomings of the correlation analysis based on indices like AE, Kp or Dst is that the physical interpretation of the indices is not always obvious. Furthermore, interpretation of a correlation analysis of such derived data is not straightforward, as statistical uncertainty cannot be readily quantified. The approach we propose here is novel in several ways: 1) we work directly with the measurements of the magnetic field, rather than indices; 2) we view magnetometer records over one day as single functional observations; 3) we use a statistical test of significance, which by its very nature takes into account random variability not attributable to physical effects. We show that the substorms do affect low-latitude currents at all longitudes. It appears that this dependence is significant not only during the same day the substorm occurs, but also the next day. Our study shows that the effect normally fades out two days after the substorms. It decreases in time faster for the cases of weak and medium substorms than those of strong substorms. Objectives and Motivation Conclusions References Acknowledgements Contact Information Results Fig. 1 Horizontal intensities of the magnetic field measured at high-, mid- and low-latitude stations during a substorm (left column) and a quiet day (right column). Note the different vertical scales for high-latitude records.  Research supported by NSF grant DMS-0413653  Data provided by the global network of observatories – INTERMAGNET data  Kokoszka, P., Maslova, I., Sojka, J., Zhu, L. (2007) Testing for lack of dependence in the functional linear model. Canadian Journal of Statistics, forthcoming  Ramsay, J. O. and Silverman, B. W. (2005). Functional Data Analysis. Springer Verlag Test statistic: Under H 0  Test if auroral geomagnetic activity reflected in the high-latitude curves has an effect on the processes in the equatorial regions reflected by the mid- and low-latitude curves  Use one-day high-latitude records during the substorms as X n. Let Y n be the records during the same substorm days and 1, 2, 3 days later. Check how long the effects of substorms last  Table 1 provides test results for the substorm effects during: (A) January – March (40 substorms); (B )March – May (42 substorms); (C) June – August (42 substorms)  The selection of three month periods ensures that the data is identically distributed (distribution could be effected by the Earth rotation) Table 1. Results of the test for substorm days that occurred in 2001 from January to March (A), March to May (B), June to August (C). Mid- latitude ABCABCABCABC BOU0BOU1BOU2BOU3 111111?0?000000 FRD0FRD1FRD2FRD3 111111?0?000000 THY0THY1THY2THY3 111111000000 MMB0MMB1MMB2MMB3 111111?0? 00? 00 Low-latitude HON0HON1HON2HON3 111111?0?0?00000 SJG0SJG1SJG2SJG3 111111?000000 HER0HER1HER2HER3 111111000000 KAK0KAK1KAK2KAK3 1111111?0?000?00 For further information please contact Inga Maslova (Inga.Maslova@gmail.com)Inga.Maslova@gmail.com Manuscript on which the poster is based is available upon request. Poster is available at http://cc.usu.edu/~inga/research.htm Test hypothesis: versus Model: Here, p, q are the numbers of the most important principal (PC’s) components, of and respectively and are eigenvalues, and are the PC’s  New methodology allows to work directly with the measurements and is relatively easy to apply, useful for finely sampled data analysis  Substorm effect lasts for about 2 days. This dependence is weaker further we get from the longitude of CMO station. The effect decreases faster for moderate substorms, for which only the same day effect can be claimed  Only the effect of strong substorm extends to low latitudes on the second day  The interpretation of this result is not readily apparent  These statistical findings may imply some physical connections between the substorm electrodynamics and the physical processes in other regions of the M-I system that we are not aware of at the present time  Test statistic gives good approximation for sample sizes around 50  The empirical sizes are close to nominal sizes and are not affected by the choise of the number of PC’s  Test does not depend on the choise of the functional basis  Finely sampled records available (magnetometer data analysis)  Testing whether auroral currents have impact on equatorial and mid-latitude currents (Fig. 1)  Novel approach: o Use the measurements of the magnetic field, rather than indices o View magnetometer records over one day as single functional observations o Use a statistical test of significance which takes into account random variability not attributable to physical effects Substorms effect lasts for about 2 days “1” – significant effect, “0” – no effect, “1?” – inclined toward effect, “0?” – inclined toward no effect, “1?0?” – inconclusive.