Our investigation fulfills a vital component of the Space Weather requirements of ESA’s Space Situational Awareness program by contributing to the capability to protect Agency assets from solar activity space radiation. We propose to develop a prototype and various scientific aspects of a physics-based, operational heliospheric solar energetic particle (SEP) forecasting system. It will allow for producing predictions of SEP fluxes at multiple locations in the inner heliosphere, by modelling their acceleration at Coronal Mass Ejections (CMEs) near the Sun, and their subsequent interplanetary transport. The proposed prototype will incorporate results from our scientific investigations, the modification and linking of existing open source scientific software, and its adaptation to the goals of the proposed work. The system will incorporate a chain of data-driven analytic and numerical models, some of which have been developed by us, for estimating: coronal magnetic field from Potential Field Source Surface (PFSS) and Magnetohydrodynamics (MHD); dynamics of large-scale coronal (CME-driven) shock waves; energetic particle acceleration; scatter-based (not simple ballistic), time-dependent SEP propagation in the heliosphere to specific time-dependent positions (deriving coordinates from NASA’s SPICE toolkit), including spacecraft such as Solar Orbiter and Parker Solar Probe. We will develop and test the capability of the linked system to produce synoptic predictions of heliospheric SEP spectra based on tables of typical values of the different parameters along the process chain, which we have developed.

REQUIREMENTS:

To be able to achieve the above-stated objectives, there are several technical requirements. First, a core software platform must exist, which will control the forecasting system. We will use the recently developed CASHeW framework1 as a basis, on which to build the prototype. CASHeW is briefly described later on in the proposal; its main advantage is the capability to automatically characterise large-scale coronal shock waves from freely available EUV observations and data-driven models on an event-by-event basis, with minimal user interaction. However, it has been developed with generality and expected future addition of observations and models in mind, which makes it perfect to use as the basis of the proposed system. The CASHeW framework will coordinate the download of observational data and generation of input model data.

The next main requirement is for the forecasts to be based on realistic, physics-based modelling. To meet this requirement, we will incorporate existing data-driven models and model results of the coronal environment and shock dynamics into the system. Specifically, we will use code to create at regular intervals (at least daily) 2D plasma maps derived directly from observations (developed by P. Zucca and described below) between 1-6 Rʘ. In addition, we will use 3D coronal and heliospheric plasma maps from synoptic MHD model runs, freely available through a web interface, produced by Predictive Science, Inc. daily.

The third main requirement of the proposed system to produce meaningful prediction is to include the proper modelling parameters in the model chain. This will require the development of parameter tables for the particle acceleration and transport models, and derivation of meaningful correlations, based on machine learning techniques. We will carry out a targeted multi-event study on actual events to develop these tables, and use the results for an in-sample validation of the system. An out-of-sample validation will be performed on a separate set of events, which will not be used for parameter determination. These studies will allow to fine-tune the modelling parameter space.

The last technical requirement is to be able to automate the system execution, and bring it to a level of minimal human interaction. This will be achieved by developing the appropriate interfaces between the models, common intermediate data products, and scripting tools for transferring data and streamlining operation.

The forecasting model chain prototype will contribute to addressing the needs of satellite operators and space agencies by providing physics-based short-term forecasting of SEP arrival times and maximum intensities in the early stages of SEP events, thereby giving lead times for mitigating the impact on sensitive satellite electronics and humans in extravehicular activities with only nominal space suit protection. In the future, the system may be linked with flare forecasting tools, improving the warning times significantly.

Fig. 1 – Some aspects of the elements in the proposed prototype system. A. Base difference image of a coronal EUV wave originating close to the solar limb. B. Geometrical model of the wave front and its interaction (green points) with open (blue) and closed (orange) coronal PFSS field lines. C. Top-down view of the projection in the equatorial plane of the PFSS lines, with the expected Earth connecting points (red arc). D. 2D map of coronal density (1-6 Rʘ), constrained by AIA and LASCO observations. E. Example of the 3D EPREM particle model computational grid, for a Parker-type solar wind description. F. A map of radial solar wind speed, from a MAS MHD model. G. and H. Modelled SEP fluxes near the Sun from an EPREM simulation with an MHD CME, at two locations near 10 Rʘ. I. Comparison of the EPREM model fluxes from various field lines with observations near 1 AU (black dots).

IMPLEMENTATION ASPECTS

Fig. 2 – Workflow diagram of the proposed prototype system. The flow is from left to right. Lighter elements represent data and products, while darker ones show software modules.

Figure 1 shows the final workflow of the proposed prototype system, as we envision it.

  1. The first element of the system is a module with the sole purpose to identify the locations of active regions on the solar disk in near-real time. It will probe existing online catalogs, such as SPoCA, HEK, and the ESA SSA SWE monitoring systems. We will also test using synoptic AIA data for providing faster monitoring.
  2. The Data Loader module will load available data (AIA, LASCO) at regular intervals, and feed them to the 2D Plasma Maps Module. It will also load the relevant parameter tables for the coronal shock dynamics.
  3. The AIA data and parameter database, together with the results from the AR monitoring tool will be used to drive an instance of the Synthetic Synoptic Shock Module (S3M), which will launch a 3D geometric shock model into the coronal environment, described by either the 2D coronal plasma maps (for events close to the limb), or the 3D MAS MHD maps (if away from the limb).
  4. Based on the interaction of the shock model with the coronal environment, the relevant parameters for diffusive shock acceleration will be extracted, and fed into the DSA time-dependent solver model. It will also take input parameters such as typical suprathermal input spectra and diffusion coefficients, and calculate the time-dependent resulting ion spectra.
  5. These spectra will be fed into the Heliospheric propagator module (consisting of a solar wind description model and the EPREM model), which will calculate the SEP fluxes near 1 AU (can be done for multiple locations, such as Mars, L1, STEREO spacecraft, Solar Orbiter, Parker Solar Probe, etc.).
  6. The entire process is controlled by a management infrastructure.

1 Kozarev K. et al., 2017, JSWSC, https://doi.org/10.1051/swsc/2017028

2 https://naif.jpl.nasa.gov/naif/toolkit.html