Date of Award

9-2023

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Department of Engineering Physics

First Advisor

Daniel J. Emmons, PhD

Abstract

Solar energetic particles (SEPs) are especially high energy protons (≥ 5 or 10MeV kinetic energies) originating from the Sun that are known to have significant effects on space-based military operations. The Air Force’s current forecast model, the Proton Prediction System (PPS), forecasts less than half of SEP events that are observed, motivating the need for an update to the operational model. To develop the updated PPS model, PPS2.0, an expanded SEP catalog was produced, which includes 103 previously unidentified events and specification of the proton flux response above background levels or prior events. Event detection and source association were initially automated, but inspection revealed that only 49.5% of the automatically detected rises in the ≥ 10MeV channel were events and a mere 50.5% of those were correctly associated to flares., necessitating significant manual review. The updated SEP catalog includes a collection of smaller proton flux responses that have not been included in previous catalogs, which improves PPS2.0 performance and provides a novel dataset for model development, physical insight, and space weather forecasts. PPS2.0’s peak proton flux response predictions use power laws in the associated flare peak flux multiplied by rise time, with a mean absolute log10 error (MALE) of 0.61 on events in Solar Cycle 24, and the flare fluence, with a MALE of 0.64. A lookup table of threshold event probabilities, built from regression on the observed frequencies by flare class, has Receiver Operating Characteristics (ROC) Area Under The Curve (AUC) values of 0.79 and higher. For thorough validation of PPS2.0, its performance was compared with UMASEP, SEPSTER, SEPSTER2D, and the current PPS over the entire Solar Cycle 24. PPS2.0 has significantly improved the probability of detection (POD) up to 75%, though this is accompanied by a strikingly high false alarm rate (FAR) over 95%. Further, PPS2.0’s highest MALE of 0.64 is considerably decreased from the current PPS’s lowest of 1.25. UMASEP performs well across all considered metrics, suggesting that a time derivative- or time series-based approach may yield additional progress in the future. Overall, PPS2.0 shows significant improvement over the current PPS and is ready to use in operational Air Force forecasts.

AFIT Designator

AFIT-ENP-DS-23-S-009

Comments

An embargo was observed for posting this dissertation.
This work is marked Distribution A: Approved for public release, distribution unlimited.
PA case number 88ABW-2023-0749

Download

Share

COinS