From Protons to Neutrons (and back again): Changepoint Detection, Spallation Modeling, and Spectral Inversion for SEP Events

We present a three-part framework for extracting more information from Solar Energetic Particle (SEP) events and neutron observations. First, we develop an adaptive, stochastic fitting approach that continuously matches a double Band function to incoming Geostationary Operational Environmental Satellite (GOES) proton data, using a changepoint algorithm to identify SEP onset and cessation independently across multiple energy thresholds. Second, we combine these inferred proton inputs with physics simulations to estimate neutron production from both SEP-related processes and proton-on-satellite spallation, giving an estimate on total (environmental and spallation-induced) neutron signal. Finally, we introduce a new inverse model that uses Bayesian inference to estimate the incident proton spectrum with uncertainties directly from neutron count rates. This model is trained and evaluated using the past decade of SEP events observed by the GOES series, learning behavior across events by adaptively aligning them on the same time axis. Together, these methods allow for improved SEP fitting and spectrum reconstruction from neutron measurements, and will be used to obtain SEP characteristics from the satellite-based neutron instruments.