MEMPSEP-I. Forecasting the Probability of Solar Energetic Particle Event Occurrence Using a Multivariate Ensemble of Convolutional Neural Networks

The Sun continuously affects the interplanetary environment through a host of interconnected and dynamic physical processes. Solar flares, Coronal Mass Ejections (CMEs), and Solar Energetic Particles (SEPs) are among the key drivers of space weather in the near-Earth environment and beyond. While some CMEs and flares are associated with intense SEPs, some show little to no SEP association. To date, robust long-term (hours-days) forecasting of SEP occurrence and associated properties (e.g., onset, peak intensities) does not effectively exist and the search for such development continues. Through an Operations-2-Research support, we developed a self-contained model that utilizes a comprehensive data set and provides a probabilistic forecast for SEP event occurrence and its properties. The model is named Multivariate Ensemble of Models for Probabilistic Forecast of Solar Energetic Particles (MEMPSEP). MEMPSEP workhorse is an ensemble of Convolutional Neural Networks that ingests a comprehensive data set (MEMPSEP-III by Moreland et al. (2024, )) of full-disc magnetogram-sequences and in situ data from different sources to forecast the occurrence (MEMPSEP-I-this work) and properties (MEMPSEP-II by Dayeh et al. (2024, )) of a SEP event. This work focuses on estimating true SEP occurrence probabilities achieving a 2.5% improvement in reliability and a Brier score of 0.14. The outcome provides flexibility for the end-users to determine their own acceptable level of risk, rather than imposing a detection threshold that optimizes an arbitrary binary classification metric. Furthermore, the model-ensemble, trained to utilize the large class-imbalance between events and non-events, provides a clear measure of uncertainty in our forecast. Solar Energetic Particles (SEPs) play an important role in modulating near-Earth space weather environments affecting communication satellites and posing radiation hazards to astronauts. Because of the complex processes involved reliable early forecast of SEP events still remains a challenge. In this paper, we present an ensemble of models that ingest both solar images and parameters measured by in situ instruments to reliably forecast the chance of SEP event occurrence. Our approach transforms a "yes"/"no" forecast to the probability of event occurrence and provides the users additional flexibility in prioritizing their forecasting goals. Our ensemble of models utilizes the fact that SEP events are much rarer than non-events and generate uncertainties on predicted probabilities. These uncertainties add value in providing a measure of trust to users in forecasting outcomes. End-to-End Deep Neural Network model-ensemble trained on remote sensing + in situ data set for forecast of SEP occurrence Usage of model ensemble maximizes the use of an imbalanced data set and brings forecasting confidence with uncertainty estimates Calibration of model outcome to true probability optimizes the forecast reliability acting as a means to step away from the binary forecast