Probabilistic Forecasting of Radiation Exposure for Spaceflight

Abstract

Extended human presence beyond low-Earth orbit (BLEO) during missions to the Moon and Mars will pose significant challenges in the near future. A primary health risk associated with these missions is radiation exposure, primarily from galatic cosmic rays (GCRs) and solar proton events (SPEs). While GCRs present a more consistent, albeit modulated threat, SPEs are harder to predict and can deliver acute doses over short periods. Currently NASA utilizes analytical tools for monitoring the space radiation environment in order to make decisions of immediate action to shelter astronauts. However this reactive approach could be significantly enhanced by predictive models that can forecast radiation exposure in advance, ideally hours ahead of major events, while providing estimates of prediction uncertainty to improve decision-making. In this work we present a machine learning approach for forecasting radiation exposure in BLEO using multimodal time-series data including direct solar imagery from Solar Dynamics Observatory, X-ray flux measurements from GOES missions, and radiation dose measurements from the BioSentinel satellite that was launched as part of Artemis~1 mission. To our knowledge, this is the first time full-disk solar imagery has been used to forecast radiation exposure. We demonstrate that our model can predict the onset of increased radiation due to an SPE event, as well as the radiation decay profile after an event has occurred.

Figures (5)

• Figure 1: The overall model architecture and the construction of the context and prediction windows. $x^{sdo}$ and $e^{sdo}$ denote the SDO images and their embeddings through a CNN architecture; $x^{1D}$ denote the vectors of univariate time series data, including radiation dose rate and X-ray flux. Note: The prediction window is shown for the autoregressive inference mode functionality.
• Figure 2: Model predictions (red with their mean in black) versus the ground truth (blue) in a hold-out interval around an SEP event occurring in May 2024 from the test set. The solid vertical red line indicates the "now" time after which the prediction is performed. The dashed vertical red line to the left of the "now" line indicates the beginning of the context window. The dashed vertical red line to the right of the "now" line indicates the end of the prediction window size used during training, but this window can be extended arbitrarily at test time due to the autoregressive nature of the model.
• Figure 3: Top row: SDO imagery ($x^{sdo}_{t}$) at t = 2024-05-10 18:45 before the onset of the SEP event. Middle & bottom rows: Radiation dose and X-ray flux ($x^{1D}$) from t = 2024-05-10 07:00 to 2024-05-13 03:00.
• Figure 4: Top row: SDO imagery ($x^{sdo}_{t}$) at t = 2024-05-11 01:45. Middle & bottom rows: Radiation dose and X-ray flux ($x^{1D}$) from t = 2024-05-10 07:00 to 2024-05-13 03:00. Notice the flare occurring at the south-east region of the solar disk prominently visible in the aia_0131 channel (top row, second column) coinciding with a prominent peak in the X-ray flux right before the radiation peak onset.
• Figure 5: Top row: SDO imagery ($x^{sdo}_{t}$) at t = 2024-05-12 12:00 as the radiation decays. Middle & bottom rows: Radiation dose and X-ray flux ($x^{1D}$) from t = 2024-05-10 07:00 to 2024-05-13 03:00.