In this study we compare the proton pitch angle distributions (PADs) in the ring current region (L ∼ 3–4) obtained from Combined Release and Radiation Effects Satellite (CRRES) observations during the large magnetic storm (minimum Dst = −170 nT) on August 19, 1991, with results of phase‐space mapping simulations in which we trace the bounce‐averaged drift of protons during storm‐associated enhancements in a model of the convection electric field. We map the phase‐space density ƒ according to Liouville's theorem except for attenuation by charge exchange, which we compute for both an empirical model [Rairden et al., 1986] and a theory‐based model [Hodges, 1994] of the neutral H density distribution. We compare simulated pitch angle distributions at 48 keV, 81 keV, and 140 keV at L = 3 and L = 4 directly with the CRRES distributions at the same energies and L values before and during the storm. A steady‐state application of our transport model, using the empirical neutral H density model of Rairden et al. [1986], reproduces the absolute intensities well except for E = 140 keV at L = 3 (M = 10 MeV/G) and E = 48 keV at L = 4 (M = 13 MeV/G). The anisotropies (A ∼ 0.2–0.8) of the CRRES and modeled pre‐storm pitch angle distributions agree within factors ≲ 2. Time‐dependent application of our transport model reproduces measured recovery phase anisotropies (t = 10–12 h after storm onset; A ∼ 0.4–1.2) similarly well at the selected energies and L values, but agreement between modeled and measured absolute intensities is energy‐dependent and not consistently good. Our model underpredicts the proton intensities found by CRRES for E > 80 keV at L = 4 in early recovery phase (t = 10–12 h). Perhaps the impulsive stormtime convection electric field was stronger during the main phase than we have assumed here. Comparisons were more difficult in late recovery phase (t = 20 h) because CRRES was too far off the magnetic equator. Proton life‐times inferred from the CRRES data during the recovery phase of this storm are considerably shorter than charge‐exchange lifetimes for either model, but the empirical neutral H density model of Rairden et al. [1986] leads to smaller discrepancies with the CRRES data at all the selected energies and L values than the theory‐based neutral H density model of Hodges [1994] for parameters that most closely represent the seasonal and solar maximum conditions of the August 19, 1991, storm. It appears that charge exchange alone is not enough to explain the observed rapid decay of the ring current proton intensities during the recovery phase of this storm.
Read full abstract