We present a study of coronal mass ejections (CMEs) which impacted one of the STEREO spacecraft between January 2008 and early 2010. We focus our study on 20 CMEs which were observed remotely by the Heliospheric Imagers (HIs) onboard the other STEREO spacecraft up to large heliocentric distances. We compare the predictions of the Fixed-Phi and Harmonic Mean (HM) fitting methods, which only differ by the assumed geometry of the CME. It is possible to use these techniques to determine from remote-sensing observations the CME direction of propagation, arrival time and final speed which are compared to in situ measurements. We find evidence that for large viewing angles, the HM fitting method predicts the CME direction better. However, this may be due to the fact that only wide CMEs can be successfully observed when the CME propagates more than 100 deg from the observing spacecraft. Overall eight CMEs, originating from behind the limb as seen by one of the STEREO spacecraft can be tracked and their arrival time at the other STEREO spacecraft can be successfully predicted. This includes CMEs, such as the events on 4 December 2009 and 9 April 2010, which were viewed 130 deg away from their direction of propagation. Therefore, we predict that some Earth-directed CMEs will be observed by the HIs until early 2013, when the separation between Earth and one of the STEREO spacecraft will be similar to the separation of the two STEREO spacecraft in 2009--2010.

The Fixed-\Phi (F\Phi) and Harmonic Mean (HM) fitting methods are two methods to determine the average direction and velocity of coronal mass ejections (CMEs) from time-elongation tracks produced by Heliospheric Imagers (HIs), such as the HIs onboard the STEREO spacecraft. Both methods assume a constant velocity in their descriptions of the time-elongation profiles of CMEs, which are used to fit the observed time-elongation data. Here, we analyze the effect of aerodynamic drag on CMEs propagating through interplanetary space, and how this drag affects the result of the F\Phi and HM fitting methods. A simple drag model is used to analytically construct time-elongation profiles which are then fitted with the two methods. It is found that higher angles and velocities give rise to greater error in both methods, reaching errors in the direction of propagation of up to 15 deg and 30 deg for the F\Phi and HM fitting methods, respectively. This is due to the physical accelerations of the CMEs being interpreted as geometrical accelerations by the fitting methods. Because of the geometrical definition of the HM fitting method, it is affected by the acceleration more greatly than the F\Phi fitting method. Overall, we find that both techniques overestimate the initial (and final) velocity and direction for fast CMEs propagating beyond 90 deg from the Sun-spacecraft line, meaning that arrival times at 1 AU would be predicted early (by up to 12 hours). We also find that the direction and arrival time of a wide and decelerating CME can be better reproduced by the F\Phi due to the cancellation of two errors: neglecting the CME width and neglecting the CME deceleration. Overall, the inaccuracies of the two fitting methods are expected to play an important role in the prediction of CME hit and arrival times as we head towards solar maximum and the STEREO spacecraft further move behind the Sun.

Rank-Ordered Multifractal Analysis (ROMA), a recently developed technique that combines the ideas of parametric rank ordering and one parameter scaling of monofractals, has the capabilities of deciphering the multifractal characteristics of intermittent fluctuations. The method allows one to understand the multifractal properties through rank-ordered scaling or non-scaling parametric variables. The idea of the ROMA technique is applied to analyze the multifractal characteristics of the auroral zone electric field fluctuations observed by SIERRA. The observed fluctuations span across contiguous multiple regimes of scales with different multifractal characteristics. We extend the ROMA technique such that it can take into account the crossover behavior -- with the possibility of collapsing probability distributions functions (PDFs) -- over these contiguous regimes.