Recorded presentation5.3
Probability distributions of electron precipitation at high magnetic latitudes
D. M. Ober, Air Force Research Laboratory, Hanscom AFB, MA; and E. G. Holeman, D. A. Hardy, W. J. Burke, L. C. Gentile, and K. H. Bounar
Starting in the early 1980s investigators at the Air Force Research Laboratory have sought to exploit information contained in the massive database accumulated by the electron and ion spectrometers on spacecraft of the Defense Meteorological Satellite Program (DMSP). One objective has been to characterize particle precipitation at high latitudes by determining the average properties of precipitating electrons as functions of magnetic latitude (MLat), magnetic local time (MLT) and the level of geomagnetic activity specified by the Kp index. Parameters used in the present analysis, integral number flux, integral energy flux, and average energy of precipitating electrons are widely used in the space weather community to estimate auroral ionospheric conductances. In the original Hardy-85 model data from the SSJ3 sensor on DMSP F2 and F4 were used to calculate "average" quantities for given MLat-MLT-Kp bins, assuming that sampled populations were reasonably singular and normally distributed. The current investigation tests these assumptions using the much larger SSJ4 data set from DMSP F7 though F15. We find that at all local times and latitudes probability distributions for detecting any of the three parameters is always lognormal and often nonsingular. Lognormal distributions have inherent positive skews with the average value larger than the most probable value. Consequently, ionospheric conductance maps based on average values provide conductance values that are too high. Using the most probable value of each distribution would provide a better estimate of the conductivities but given the lognormal and nonsingular character of the distributions even the most probable value is an unlikely predictor of actual conditions. Recorded presentation
Session 5, New Developments with Physics-Based Forecast Models Part I
Tuesday, 22 January 2008, 8:30 AM-9:45 AM, 221
Previous paper Next paper