Forecasting well-connected Solar Energetic Proton storms

A Solar proton event occurs when protons emitted by the Sun become accelerated to very high energies during a solar flare accompanied by a coronal mass ejection or in interplanetary space by the shocks associated with coronal mass ejections. Protons are finally guided by the interplanetary magnetic field lines.

Energetic proton storms can electrically charge spacecraft to levels that can damage electronic components. Solid state memory can be altered.

Energetic solar protons are a significant radiation hazard to spacecraft and astronauts. They can cause spacecraft to lose their orientation. Flashes and streaks of light occur when energetic protons strike the sensitive optical electronics in spacecraft and can also destroy the efficiency of the solar panels. Significant proton radiation exposure can be experienced by astronauts who are outside a protective shield in the space.

Earth is largely protected by its magnetic field, or magnetosphere. However, the closer a spacecraft (or aircraft) approaches the polar regions, the greater the exposure to energetic proton radiation will be.

There are two kinds of SEPs with respect to the onset delay: Well-connected SEP events, also called prompt SEPs, that arrive to the near-Earth environment in minutes or a few hours after the major solar event; and, delayed SEP events, that arrive to the near-Earth environment in days, due to particles that are accelerated by shocks associated with coronal mass ejections.

Fig. 1. Solar energetic protons emitted by the Sun become accelerated to very high energies. Energetic protons and ions are finally guided by the interplanetary magnetic field lines.

Regarding well-connected SEPs, ¿ what is the most possible and frequent cause of the prompt component (if it exists) of a SEP? A lot of studies have been done on this issue. The most possible cause is a mixed particle acceleration process, flares and CME-driven shocks: Firstly, the flare accelerates protons/ions. These flare-accelerated ions escape through open field lines that connect the flare site to the shock front. Secondly, shock re-accelerates proton/ions. These shock reaccelerated ions travel throw magnetic lines that connect the front of the shock to the spacecraft at 1 AU. The prompt component of well-connected SEPs was largely studied by Cane et al (2003) and simulated by Li et al (2004).

The UMA SEP Forecaster

Magnetically well-connected SEP events, may be detected at 1AU several minutes or hours after a major solar event. We have developed a new signal processing method that proves that the incoming proton fluxes are correlated with the evolution of the associated flare in most of the well-connected SEP events. Our approach correlates soft X-rays with proton fluxes in several energy channels (ranging between 9MeV and 500MeV) and uses the correlation result as an empirical measure of magnetic connectivity. In Núñez et al (2006), we observed that a high magnetic connectivity is present several minutes or hours before the rise of the integral proton flux (E>10MeV). In Núñez (2009), we found that the intensity of the first hours of the prompt SEP is proportional to the peak intensity of the associated flare and the magnetic connectivity.

Based on these findings, we have developed the UMA SEP forecaster, that may predict in real-time the onset and the intensity of the prompt component of well-connected SEP. This real.time forecaster processes the current X-ray and proton data every 5 minutes. The forecaster output is updated automatically in the forecast panel. Figure 2 shows the main output of the UMA forecaster. The upper time series shows the integral proton flux (E>10MeV) and the predicted interval time of occurrence of the SEP onset and the expected peak intensity during the first hours, which is shown in red. The middle time series shows the X-ray flux. The lower time series of this figure shows the estimated magnetic connectivity, which is based on the correlation between the X-ray flux and the proton data detected at 1 AU.

The UMA SEP forecaster detects periods of high magnetic connectivity between the Sun and the analyzed satellite (GOES satelittes, by default), and forecasts well-connected phenomena, which could not only be SEPs, but also well-connected integral proton enhancements lower than 10 fpu (E> 10MeV). If the model forecasts a SEP, the graphical output shows in red the range of values in which the integral proton flux is expected to evolve, which allows to easily see the time interval in which a well-connected SEP onset is expected to occur. It also shows the proton flux peak of the predicted SEP prompt component during the first 8 hours of the storm. The textual output consists of some sentences in natural language describing the details of the forecast.

Fig. 2. This figure shows two SEP events occurred during October 26th and 28th of 2003. The upper figure shows when the first forecast was issued. Note that the integral proton flux has not risen yet. The lower figure shows when the second forecast is issued. Note that the first forecast was correct because the proton flux evolves within the forecasted proton flux depicted in red. The second forecast is also correct as shown in the small figure at the right lower corner. In can be observed that the intensity forecasts made with an anticipation of 42 minutes and 1 hour (for both SEP events) was accurate.

The maximum anticipation time (also called warning time) of the forecast of well-connected SEPs is 6:10 hours, in contrast to the current NASA approach, Posner (2007), for predicting this kind of SEPs with an anticipation time of up-to-one hour. Currently, there is no other method that may give real-time alerts of well-connected SEPs with so much time in advance. The large anticipation time of our model is due to the fact that it may empirically estimate the magnetic connectivity from the correlation of the X-ray flux of the beginning of the flare with the solar proton fluxes detected at 1 AU. When the X-ray flux is at maximum level, the forecaster has accumulated a lot of evidence about the magnetic connectivity, and it therefore may forecast a well-connected SEP with confidence. This model is the core of the UMA SEP forecaster. Figure 3 shows the output of the forecaster for several well-connected SEPs. Figure 3b and 3c show the prediction of two strong well-connected SEPs. Figure 4 shows the overall performance of the forecaster for solar cycles 22 and 23.

Fig. 3. This figure shows several successful forecasts for several SEP events. The left chart shows the forecast of a SEP occurred during November 8th, 1987, in which the forecast anticipation time was 4.01 hours. The middle chart shows forecast of a SEP occurred during October 30th, 1992. In this case the forecast anticipation time was 58 minutes. In the right chart, the forecast anticipation time was 45 minutes for the SEP of September 30th, 1998.

Fig. 4. This figure shows forecast accuracy data. The UMA SEP forecaster predicted the 84% of all prompt SEP events, which are approximately 46% of all SEP events. Every prediction consists of an expected time interval and an expected peak intensity of the first hours. The first version of the UMA SEP forecaster started to work in real-time in January 2007. Recently, we have installed the second version which reduced the false alarm rate to 45% taking into account the historical data of the solar cycles 22 and 23. We also provide a web service that may process user-provided historical data to estimate magnetic connectivity time intervals and to predict SEP events for validating our results or for research purposes.

Evaluation of the UMA SEP forecaster

In order to evaluate the UMA SEP forecaster, we used 5-min data of solar cycles 22 and 23 with information about X-ray, and all solar proton flux data as input of two evaluation tools: one for measuring the statistical forecasting performance and another tool for measuring the level of operativeness on a daily-basis.

The statistical evaluation tool makes the forecaster model issue a prediction every 5-minutes step of solar cycles 22 and 23. After performing parallel processing, the tool summarizes all these forecasts and generates several statistical performance indices: the percentage of well-connected SEPs whose onsets were successfully predicted; the percentage of all SEPs that were successfully predicted; and, the false alarm rate. This statistical performance tool yields that the UMA SEP forecaster successfully predicted 84% of well-connected SEPs. This evaluation tool also found that the 45% of all forecasts anticipated a SEP that never occurred; this is a false alarm rate of 45%. The well-connected SEPs are nearly a half of all SEP events, as it is shown in fugure 6a. Figure 5 shows the performance of the forecasted prompt component instensity.

Fig. 5. This figure shows the performance of the predicted intensity of the first 8 hours of well-connected SEPs compared with real SEP intensity values.

In order to evaluate the level of operativeness of this real-time forecaster, we developed another tool, available as an public online service at spaceweather.uma.es/uma_sep_tool.htm that allows the user to go to any specific time step of cycles 22 and 23 and make the forecaster process the solar data, so he/she may evaluate the graphical, numerical and textual outputs of the forecasting model for that specific time. By default, this tool analyzes real historic data of the GOES satellite network of cycles 22 and 23, however the user may also upload to our server several days of a situation of X-ray and proton flux data from the sensors of other satellite and make our model process the data. The output of this tool is the similar to the output of the real-time forecaster (see figures 2 and 3).

Figure 6a shows that the well-connected SEP are nearly a half of all SEPs. Figure 6b shows the accuracy of the UMA SEP forecaster, based on the time to the peak of causing solar flare, called Well Connection Time (WCT). It shows that the UMA SEP forecaster successfully predicted the 93% of the fastest and relativistic well-connected SEPs of the last solar cycle, that is, those SEPs with WCT<=2 hours.

Fig. 6. The left chart shows that well-connected SEPs are approximately a half of all SEPs taking into account a Well Connection Time (WCT) ~ 8 hours. We think that if the WCT ≤ 9 hours and there is a high magnetic connectivity, the SEP is well connected. The right chart shows that the our system predict slightly better the relativistic SEPs with WCT ≤2 hours.

Conclusions

Well-connected SEPs are very fast phenomena. For this reason the UMA SEP forecaster needs to anticipate to the worst scenarios, tracking the beginning of every flare looking for correlated proton behaviours at 1 AU in real-time. When the correlation is very high a forecast is issued. In order to correlate these complex signals, a new correlation method was designed. The statistical performance of this forecast, based on the solar data of cycles 22 and 23, was that the forecaster predicted the 84% of all well-connected SEP, had a false alarm rate of 45%, and a maximum anticipation time of 6 hours 10 mins.

References

Cane, H. V., T. T. von Rosenvinge, C. M. S. Cohen, and R. A. Mewaldt (2003), Two components in major solar particle events, Geophysical Research Letters, 30(12), 8017, doi:10.1029/2002GL016580 (2003)
Li G. and Zank, G. P. (2005) Some aspects of particle acceleration and transport at CME-driven shocks, Proceedings IAU Symposium, No. 226, 2005.
Nuñez M., Morales R. (2006), Early Warning of Solar Proton Events, Third European Space Weather Week, Brussels, 2006.
Nuñez M. Núñez-Montañez D. A, (2009), A Model for Predicting the Onset and Intensity of Well-Connected SEPs and Its Evaluation for Cycles 22 and 23, European Space Weather Week, Belgium, 2009.
Posner, A. (2007), Up to 1-hour forecasting of radiation hazards from solar energetic ion events with relativistic electrons, Space Weather, 5, S05001, doi:10.1029/2006SW000268.