Solar particle event dose prediction using kernel regression

摘要

Solar energetic particle events (SEPs) are a naturally occurring phenomenon in which subatomic particles are radiated from the sun. Astronauts in deep space could be subjected to significant radiation exposures due to these events. The dose is usually received over approximately a 20–40 hour interval. The effects of these particles could be mitigated by an early detection and warning system, which provides reliable estimates of the magnitude of the expected radiation dose from the event. For this purpose we have created a new software package designed to forecast the final total dose for the event, from doses measured early in the event cycle. This package implements a sliding scale kernel regression model to calculate an estimate for the projected total dose. The software displays a plot of both the current dose forecast and a history of previous predictions. These predictions are accompanied by uncertainty estimates. Sample results are presented.