Specification of >2 MeV geosynchronous electrons based on solar wind measurements

[ 1] Relativistic electron flux measurements from geosynchronous satellites show a local time dependence. This local time dependence is due to the radial profile of the electron fluxes and the dayside/nightside asymmetry of the Earth's magnetosphere and is also affected by geomagnetic activity, which is in turn affected by the solar wind. Statistical asynchronous regression (SAR), a statistical method recently adapted for magnetospheric studies, was used to determine the relationship between electron fluxes measured at different local times, as a function of the Kp index (O'Brien et al., 2001). In this study, we use measurements directly from the solar wind, instead of the Kp index, and the SAR method as the basis for determining the local time dependence of geosynchronous energetic electron fluxes. We use solar wind parameters as input in our model to map GOES 10 > 2 MeV electron measurements to other local times and compare them with electron measurements from five widely spaced LANL geosynchronous satellites when they pass through these local times. We cross calibrate the electron measurements from the five LANL satellites and find that the averaged electron flux from individual satellites can differ by up to 50% even though the particle detectors were identically designed. In this study, we normalizemeasurements from each LANL spacecraft to the average value of all five LANL spacecraft. We also cross calibrate the electron measurements from the LANL satellites and from GOES 10 and find that the energy spectrum is best described by a power law index which is a function of the current average LANL 1.1 - 1.5 MeV electron flux level. We explore the effects of solar wind velocity, dynamic pressure, and density on the local time dependence of geosynchronous electron fluxes. We find that for the given 4.3 year data set, using only solar wind velocity gives rise to the best results. We check the efficacy of the model by mapping GOES 10 > 2 MeV electron measurements at other local times to local noon and comparing with electron measurements from LANL satellites when they pass through local noon: We achieve an out-of-sample prediction efficiency ( PE) of 0.83 and a linear correlation coefficient (LC) of 0.93 for January 2000, whereas we achieve a PE of 0.81 and LC of 0.92 for the year 2000. The PE and LC are different when mapping GOES 10 > 2 MeV electron measurements at one local time to other local times, with the highest PE around prenoon and afternoon and the lowest near midnight.