SEC proton prediction model: verification and analysis

This paper describes a model that has been used at the NOAA Space Environment Center since the early 1970s as a guide for the prediction of solar energetic particle events. The algorithms for proton event probability, peak flux, and rise time are described. The predictions are compared with observations. The current model shows some ability to distinguish between proton event associated flares and flares that are not associated with proton events. The comparisons of predicted and observed peak flux show considerable scatter, with an rms error of almost an order of magnitude. Rise time comparisons also show scatter, with an rms error of similar to 28 h. The model algorithms are analyzed using historical data and improvements are suggested. Implementation of the algorithm modifications reduces the rms error in the log(10) of the flux prediction by 21%, and the rise time rms error by 31%. Improvements are also realized in the probability prediction by deriving the conditional climatology for proton event occurrence given flare characteristics. (C) 1999 Published by Elsevier Science Ltd. All rights reserved.