2-1-2 solar energetic particle prediction and alert · fig.2 relation between maximum proton...

1 Introduction Now as the sphere of human activity is expanding into space, space environmental disturbances more often than not threaten the security and safety of human life. Given this situation, the technique for space weather forecasting to gain a precise insight into space environmental disturbances, as well as to pre- dict them and issue alerts, is assuming greater importance. In particular, as space travel becomes a reality, such as the International Space Station (ISS) manned by astronauts for extended periods of time orbiting in space, possible exposure to cosmic radiation poses a very serious challenge to the health of astro- nauts and other crew members. The term “cosmic radiation” refers to the ionizing radiation present in space and can be broadly divided into three types according to the origin: galactic cosmic rays, solar ener- getic particles, and radiation belt particles. From the standpoint of astronauts being exposed to radiation hazards, galactic cosmic rays and solar energetic particles are impor- tant. Galactic cosmic rays arriving from out- side the solar system are generally believed to have constant intensity on a short time scale compared to the solar activity cycle of 11 years, with the radiation dose aboard ISS being estimated at about 0.5 mSv a day. This is roughly comparable to the radiation dose experienced in a chest CT scan test once every two weeks. In contrast, solar energetic parti- cles subject to sudden increases in intensity under the influence of solar flares or similar events are estimated to result in a radiation dose a dozen to several tens of times higher than normal [1] . This suggests that avoiding exposure to solar energetic particles is neces- sary for protection against radiation hazards. Astronauts on an extended period of stay aboard the ISS or traveling to the Moon, Mars or elsewhere onboard manned spacecraft should therefore benefit from being alerted to the occurrence of solar energetic particles, and thus notified of the risks of solar energetic particles. There are two broad conceptual approach- es to issuing alerts on the occurrence of solar energetic particles. One involves alerts regard- ing occurrence of the events themselves. The other involves alerts on time variations in the solar energetic particle flux. Figure 1 shows an example of extremely large solar energetic particle flux observed immediately after the events initially occurred. In this case, issuing an alert on occurrence of the events them- 17 KUBO Yûki et al. 2-1-2 Solar Energetic Particle Prediction and Alert KUBO Yûki, NAGATSUMA Tsutomu, and AKIOKA Maki Cosmic radiation is a commonly known source of exposure to natural radiation for astronauts and artificial satellites. The phenomenon of solar energetic particles suddenly increasing in intensity is one of the most hazardous events. This paper first introduces statistical and simula- tion approaches to predicting and alerting the occurrence of solar energetic particles. It then reviews the development of a space environment reporting system designed to detect solar energetic particles in the early stages of occurrence and disseminate related information. Keywords Solar energetic particles, Particles transport, Radiation hazard, Event reporting system

Upload: others

Post on 11-Jun-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: 2-1-2 Solar Energetic Particle Prediction and Alert · Fig.2 Relation between maximum proton intensity and proton flare fluence Fig.3 Relation between peak X-ray flux of proton flares

1 Introduction

Now as the sphere of human activity isexpanding into space, space environmentaldisturbances more often than not threaten thesecurity and safety of human life. Given thissituation, the technique for space weatherforecasting to gain a precise insight into spaceenvironmental disturbances, as well as to pre-dict them and issue alerts, is assuming greaterimportance. In particular, as space travelbecomes a reality, such as the InternationalSpace Station (ISS) manned by astronauts forextended periods of time orbiting in space,possible exposure to cosmic radiation poses avery serious challenge to the health of astro-nauts and other crew members.

The term “cosmic radiation” refers to theionizing radiation present in space and can bebroadly divided into three types according tothe origin: galactic cosmic rays, solar ener-getic particles, and radiation belt particles.From the standpoint of astronauts beingexposed to radiation hazards, galactic cosmicrays and solar energetic particles are impor-tant. Galactic cosmic rays arriving from out-side the solar system are generally believed tohave constant intensity on a short time scalecompared to the solar activity cycle of 11

years, with the radiation dose aboard ISSbeing estimated at about 0.5 mSv a day. Thisis roughly comparable to the radiation doseexperienced in a chest CT scan test once everytwo weeks. In contrast, solar energetic parti-cles subject to sudden increases in intensityunder the influence of solar flares or similarevents are estimated to result in a radiationdose a dozen to several tens of times higherthan normal[1]. This suggests that avoidingexposure to solar energetic particles is neces-sary for protection against radiation hazards.Astronauts on an extended period of stayaboard the ISS or traveling to the Moon, Marsor elsewhere onboard manned spacecraftshould therefore benefit from being alerted tothe occurrence of solar energetic particles, andthus notified of the risks of solar energeticparticles.

There are two broad conceptual approach-es to issuing alerts on the occurrence of solarenergetic particles. One involves alerts regard-ing occurrence of the events themselves. Theother involves alerts on time variations in thesolar energetic particle flux. Figure 1 showsan example of extremely large solar energeticparticle flux observed immediately after theevents initially occurred. In this case, issuingan alert on occurrence of the events them-

17KUBO Yûki et al.

2-1-2 Solar Energetic Particle Prediction and Alert

KUBO Yûki, NAGATSUMA Tsutomu, and AKIOKA Maki

Cosmic radiation is a commonly known source of exposure to natural radiation for astronautsand artificial satellites. The phenomenon of solar energetic particles suddenly increasing inintensity is one of the most hazardous events. This paper first introduces statistical and simula-tion approaches to predicting and alerting the occurrence of solar energetic particles. It thenreviews the development of a space environment reporting system designed to detect solarenergetic particles in the early stages of occurrence and disseminate related information.

KeywordsSolar energetic particles, Particles transport, Radiation hazard, Event reporting system

Page 2: 2-1-2 Solar Energetic Particle Prediction and Alert · Fig.2 Relation between maximum proton intensity and proton flare fluence Fig.3 Relation between peak X-ray flux of proton flares

18 Journal of the National Institute of Information and Communications Technology Vol.56 Nos.1-4 2009

selves is very important. Since the presentstate of the art does not yet allow us to identi-fy the mechanism of particle acceleration orcomprehend the physics of solar flares andCMEs (coronal mass ejections) involved withthe particle acceleration, the work of predict-ing the occurrence of the events with physicaljustification is extremely difficult or impracti-cable to achieve. As a workaround, techniquesadhering to empirical rules based on observa-tional data have been proposed[2]–[7].

At the same time, the significance of timevariations in solar energetic particle flux fol-lowing occurrence of the events, particularlythe peaks in particle flux and total particle fluxalerts, has been pointed out[8]. This may bepredictable with some physical justificationgiven the process of solar energetic particlepropagation through the solar wind, and relat-ed attempts are now underway (see[9]).

Along with these attempts, a space envi-ronment event reporting system is beingdeveloped to disseminate information aboutoccurrence of the events in its early stages.

This paper discusses a study of alerts onthe occurrence of the solar energetic particleevents based on the empirical rules describedin Reference[2] in Chapter 2. Chapter 3describes a study of predicting time variationsin solar energetic particle flux by means ofsimulation, and Chapter 4 introduces the

development of a space environment eventreporting system.

2 Occurrence of the solar energeticparticle events

Since the present state of the art does notyet allow us to identify the mechanism of par-ticle acceleration or comprehend the physicsof solar flares and CMEs involved with theparticle acceleration, the work of predictingoccurrence of the events with physical justifi-cation is impracticable to achieve. Under thesecircumstances, deriving empirical rulesregarding occurrence of the events by analyz-ing observational data would provide apromising solution to alerts regarding occur-rence of the solar energetic particle events.

We analyzed X-ray flare data as observedby GOES (Geostationary Operational Envi-ronment Satellite) to delve into the propertiesof X-ray flares that generate the solar ener-getic particle events, in an attempt to deriveempirical rules regarding occurrence of theevents.

2.1 Data analysesTo issue alerts on the occurrence of solar

energetic particles, observational data madeavailable in real time must be analyzed. Tothis end, we used soft X-ray flux and solarenergetic particle intensity data as observed bythe GOES satellite. The term “solar energeticparticle events” lacks a strict definition since itmay occur on a variety of scales. For the sakeof convenience, particles having energy of atleast 10 MeV, with intensity exceeding 10 pro-tons cm-2sec-1sr-1 (as observed by the GOESsatellite), are defined as solar energetic parti-cles. The ISES (International Space Environ-ment Service) uses this definition as a stan-dard for space weather prediction and is arequirement imposed from the perspective ofengineering, not physics. Solar energetic parti-cle events matching this definition are called“proton events.” From the standpoint of spaceweather prediction, variations of the solarenergetic particle events not meeting this defi-

Fig.1 Solar energetic particle events(reprinted from Space WeatherPrediction Center, NOAA Website)

Page 3: 2-1-2 Solar Energetic Particle Prediction and Alert · Fig.2 Relation between maximum proton intensity and proton flare fluence Fig.3 Relation between peak X-ray flux of proton flares

19KUBO Yûki et al.

nition are excluded from the scope of our dis-cussions. The analysis period was from Janu-ary 1997 to mid-July 2002.

Proton events are considered to normallyoccur in connection with large solar flares.Although the scale of a solar flare is generallyexpressed as a peak X-ray flux (such as anM1.0 class flare), the time integral of an X-rayflux (also known as “fluence”), which is con-sidered relevant to the total energy emitted byflares, may provide a better indicator. In thiswork, fluence is therefore used as an indicatorof solar flare scale. Fluence is defined as atime integral of an X-ray flux, and can beexpressed as:

where, F1, F(t), ts and te denote the fluence,X-ray flux, flare start time, and flare end time,respectively. The background level is assignedas a flux value of the X-ray flare start time.The flare start time has been identified byNOAA's Space Weather Prediction Center(SWPC). The flare end time is defined as thetime when the X-ray flux has diminished tohalf of its maximum.

Fig. 2 shows the relation between the max-imum proton intensity and the fluence offlares that produce proton events (protonflares). The flare (C1.7 class) at the leftmostpoint in the diagram can be ignored from thescope of subsequent discussions for certainreasons. As shown in Fig. 2, some relationmay exist between the maximum proton inten-sity and the fluence of X-ray flares, but thecorrelation coefficient between both is onlyabout 0.43 and inadequate to confirm a corre-lation. Figure 2 also shows that proton flareshave a fluence value of at least 20 ergs cm-2.This finding is important, because it helps usconclude that no proton events will occur aslong as proton flares have a fluence valuelower than 20 ergs cm-2.

Fluence used as an indicator of flare scaleis defined here as a time integral of flux asexpressed in Equation (1), but generally thepeak X-ray flux is used as described earlier.

Fig.2 Relation between maximum protonintensity and proton flare fluence

Fig.3 Relation between peak X-ray flux ofproton flares in solar cycle 23 (dur-ing the analysis period) and flareduration, and threshold line drawnby solving Equation (1)

Fig.4 Similar to Fig. 3, except that thedata points are those in solar cycle22.

Page 4: 2-1-2 Solar Energetic Particle Prediction and Alert · Fig.2 Relation between maximum proton intensity and proton flare fluence Fig.3 Relation between peak X-ray flux of proton flares

20 Journal of the National Institute of Information and Communications Technology Vol.56 Nos.1-4 2009

Accordingly, the results above can be madeeasier to understand intuitively and becomemore useful from a real-time alerting stand-point when expressed using the peak X-rayflux. Data analyses of the proton flares usedthis time draw on the existence of the follow-ing relation between fluence F1 and peak X-ray flux Fp expressed as:

Fp・D=1.0076 Ft 0.96633 (1)

where, D denotes the duration of the flaredefined as D = t c - tm, and tm the time at thepeak X-ray flux. Using this equation alongwith the threshold of proton event occurrenceFt = 20 ergs cm-2, a threshold line (D = 18.2Fp-1) can be drawn in the Fp-D plane asshown in Fig. 3.

Figures 3 and 4 plot the durations andthreshold lines of proton flares in solar cycles23 and 22, respectively, with respect to thepeak X-ray flux. As can be seen from thesediagrams, nearly all proton flares sit above thethreshold lines, indicating the presence ofthresholds during the flare duration withrespect to the peak X-ray flux. For example,only M1.0 class flares having a duration ofabout 30 minutes or longer can produce protonevents. In the past, it had been empiricallystated that solar flares of the M class or higherhaving a longer duration are more likely toproduce proton events, a statement now legiti-mately endorsed by data.

Figure 5 plots the fluence distributions ofall C class and higher flares (observed about10,000 times) that occurred during the analy-sis period. Evidently, a majority of these flareshave fluence values of 20 ergs cm-2 or lower,and flares that could produce proton eventswere rare. During the analysis period, flaresmeeting the proton flare requirement wereobserved about 400 times, as compared withproton events being observed only about 60times. This means that all flares meeting thefluence requirement of at least 20 ergs cm-2

did not produce proton events. From this find-ing, it is concluded flares failing to meet thefluence requirement of at least 20 ergs cm-2will

not produce proton events, but one should alsoremember that this fluence requirement is onlya necessary condition for a proton event tooccur.

Yet, knowing in advance whether protonevents will occur can be very useful informa-tion in making decisions on whether to call offextravehicular activities by astronauts aboarda space shuttle, ISS or other spacecraft.

3 Time variations in solar energeticparticle flux

Solar energetic particles are accelerated byboth the solar flare region and interplanetaryshock waves. Both factors must generally betaken into account to predict time variations inthe particle flux near the earth. While solarenergetic particles having energy of at least100 MeV are observed in the initial phase of aproton event, the acceleration of particles to ahigh energy by interplanetary shock waves isdifficult to achieve. Rather, such solar ener-getic particles have presumably propagatedthrough the solar wind upon being acceleratedin a solar flare region or solar corona. Hence,particles accelerated by interplanetary shockwaves may be set aside from the process ofpredicting time variations in such solar ener-getic particle flux. Therefore, only the propa-gation process of solar energetic particlesaccelerated near the sun through the solarwind need be heeded in considering the initial

Fig.5 Fluence distributions of all C classand higher flares occurring duringthe analysis period

Page 5: 2-1-2 Solar Energetic Particle Prediction and Alert · Fig.2 Relation between maximum proton intensity and proton flare fluence Fig.3 Relation between peak X-ray flux of proton flares

21KUBO Yûki et al.

phase of a proton event.Solar energetic particles emitted near the

sun arrive near the earth propagating throughthe solar wind magnetic field. Because theLarmor radius of solar energetic particles issufficiently small compared to the globalstructure of the solar wind magnetic field, par-ticles approaching the earth are aligned withmagnetic field lines in the solar wind. In themeantime, small-scale magnetic field distur-bances (turbulent magnetic fields) existthroughout the solar wind, thereby diffusingthe solar energetic particles propagatingthrough the solar wind. Consequently, thesolar energetic particles propagate along aglobal magnetic field upon being diffused inthe solar wind. Such a form of propagation hasgenerally been approached numerically in asolving the Fokker-Planck equation or throughsimulation[9]based on a stochastic differen-tial equation. Implementing such numeric sim-ulations essentially requires that the diffusioncoefficient tensor that provides a diffusioncoefficient and drift motion, and a pitch anglediffusion coefficient relevant to the mean freepath be defined in a quantitative manner.Although these coefficients are generallyassigned analytically under various conditionsof approximation, they are not necessarilyvalid and may deviate by a wide margin fromthe values evaluated by observations undervaried solar wind conditions[10]. Despitethose drawbacks, the calculation techniquesdescribed above are still being used since theyeliminate the need to track down the motion ofparticles, save calculation time, and thus elim-inate the need to employ supercomputers andhigh-speed parallel computing technique.

Recent advances in supercomputers andhigh-speed parallel computing technique havemade high-speed computing possible. The useof such high-speed computing technique indirectly tracking down the motion of solarenergetic particles in a turbulent magneticfield would allow the process of particle prop-agation involving the diffusion process insolar wind to be reproduced[11]. Because thistechnique directly tracks down the motion of

particles by solving an equation of motion, iteliminates the need to define a diffusion coef-ficient tensor or other geometric entity thatdescribes diffusion or drift motion, therebyoffering the advantage of simulating theprocess of solar energetic particle propagationgiven a degree of magnetic field turbulence.Subsequent discussions introduce a method ofsimulating the process of solar energetic parti-cle propagation through the solar wind basedon this technique.

3.1 Simulation of diffusion process Three-dimensional isotropic turbulent

magnetic fields are generally expressed as asuperposition of numerous waves. Each waveis expressed as a trigonometric function deter-mined by four random numbers—two in thewavenumber vector direction and one each forthe polarization and phase of the wave. Theturbulent magnetic field generated as a super-position of these waves may be added to back-ground magnetic field B0 to determine a gener-al magnetic field as follows:

γ is set to 11/3 to make G(kn) Kol-mogorov-type turbulence spectrum where knis large. Lc denotes the correlation length ofthe turbulence, and the unit vector in thesinusoidal wave direction. Settingensures that div . The intensity of theturbulence can be varied by changing thevalue of parameter (δB/Bo). Solving the equa-tion of motion of solar energetic particles inthe resultant magnetic field will track downthe motion of particles and reproduce theirprocess of diffusion.

The first step is to perform a test calcula-

Page 6: 2-1-2 Solar Energetic Particle Prediction and Alert · Fig.2 Relation between maximum proton intensity and proton flare fluence Fig.3 Relation between peak X-ray flux of proton flares

22 Journal of the National Institute of Information and Communications Technology Vol.56 Nos.1-4 2009

tion assuming a uniform background magneticfield to verify that the diffusion process isreproduced. Figure 6 plots the results of track-ing down the motion of 10-MeV particles in amagnetic field having a turbulence intensity of0.1. The tracks of eight sample particles arepictured in the diagram, though the simulationtraced the tracks of 20,000 particles. Thebackground magnetic field is oriented perpen-dicularly to the space from the opposite sideof that space, suggesting that the particles

have traversed the background magnetic fieldlines. To verify that these are a precise repro-duction of the diffusion process, the root meansquare of the distance from the Z-axisis calculated using the simulation resultsshown in Fig. 7. The root mean square isfound to increase in proportion to time t. Thisproportionate increase is characteristic of thediffusion process, thereby demonstrating thatthe simulation is a precise reproduction of thediffusion process. These calculations allow usto confirm that simulating particle motiontracking in a turbulent magnetic field canreproduce the diffusion process.

3.2 Propagation of solarenergetic particles

We can simulate the process of solar ener-getic particle propagation through the solarwind by using the method described in preced-ing section. The solar wind magnetic field is asuperposition of the Parker field with a turbu-lent magnetic field, and the simulation repro-duces the propagation of solar energetic parti-cles accelerated near the sun. The effect of acurrent sheet on the propagation of solar ener-getic particles is examined below. A currentsheet is a structure in which the direction ofthe magnetic field is reversed on both sides ofan extremely thin layer.

Figure 8 plots the positions of simulated10-MeV solar energetic particles having beenreverted toward the sun along the Parker fieldafter arriving at the earth's orbit. This revert-ing of particles toward the sun along the Park-er field makes it possible to cancel outchanges in their latitude and longitude result-ing from motion in parallel with the magneticfield lines. Hence, the diagram only showschanges in the latitude and longitude of parti-cles resulting from their motion traversing theParker field, without and with a current sheet(black line) as shown in the left and right pan-els, respectively. Since the particles are inject-ed at the position of 0 degrees heliographiclatitude and -1 degrees heliographic longitudein the solar disk, the particles should be con-centrated in the vicinity of location (0, -1) in

Fig.7 Time variations in <r 2>, root meansquare of the distance from the ori-gin calculated from 20,000 simulat-ed particles

Fig.6 Simulated motion of 10-MeV parti-cles in a turbulent magnetic field

Page 7: 2-1-2 Solar Energetic Particle Prediction and Alert · Fig.2 Relation between maximum proton intensity and proton flare fluence Fig.3 Relation between peak X-ray flux of proton flares

23KUBO Yûki et al.

the diagram were it not for a turbulent mag-netic field, but both latitudinal and longitudi-nal distributions are evident in the diagram.This is suggestive of the motion of particles inthe direction in which they traverse the mag-netic field under the influence of a turbulentmagnetic field or current sheet. Without a cur-rent sheet, the particles may spread 12 to 13degrees by diffusion caused by a turbulentmagnetic field. Conversely, where a currentsheet exists, the travel of particles on bothsides of the current sheet would result in driftmotion, causing the particles to evidentlymove along the current sheet. The results ofthis simulation suggest that, if an active regionexists near a current sheet, solar energetic par-ticles accelerated near the sun may beobserved even from an observation point(such as the earth) not linked to the activeregion by a magnetic field line. This mightexplain the events whereby solar energeticparticles accelerated by a flare occurring onthe eastern side of the solar disk are observednear the earth.

Figure 9 shows time variations in the parti-cle flux on the earth as calculated from the

simulation results, without and with a currentsheet as shown in left and right panels, respec-tively. Of the five lines appearing in eachpanel of the diagram, the one marked “+0deg.” points to variations observed where,upon the occurrence of a flare, the flare region(particle injection point) is connected to theearth by a magnetic field line; the line marked“-1 deg.” points to variations observed wherethe magnetic field line connected to the earthupon the occurrence of a flare is located 1degree east of the flare region. Likewise, thelines marked “+1,” “+2” and “+3 deg.” desig-nate variations observed with magnetic fieldlines located 1, 2 and 3 degrees to the west,respectively. For the sake of better recogni-tion, each line is offset 0.01 in the y-axisdirection. In both panels, the flux rises fast incase of +0 deg., but slower as the earth movesaway from the flare region. The flux is some-what smaller in the presence of a currentsheet. Figure 10 also shows time variations inthe particle flux on the earth as calculatedfrom the simulation results. The blue line inthe diagram marks time variations in the fluxon the earth's orbit as observed when the earth

Fig.8 Positions of particles being reverted toward the sun along the Parker field after arrival atthe earth's orbit

Page 8: 2-1-2 Solar Energetic Particle Prediction and Alert · Fig.2 Relation between maximum proton intensity and proton flare fluence Fig.3 Relation between peak X-ray flux of proton flares

24 Journal of the National Institute of Information and Communications Technology Vol.56 Nos.1-4 2009

was located on the same side (left panel) asthe flare region with respect to the currentsheet, and on the opposite side at the time offlare occurrence (right panel) (see Fig. 11).The red line designates flux in the absence of

a current sheet at the same position. Evidentlyfrom the left panel of the diagram, the fluxexhibits similar variations, regardless of thepresence or absence of a current sheet, as longas the flare region and the earth are located on

Fig.9 Time variations in particle flux on the earth

Fig.10 Time variations in particle flux on the earth

Page 9: 2-1-2 Solar Energetic Particle Prediction and Alert · Fig.2 Relation between maximum proton intensity and proton flare fluence Fig.3 Relation between peak X-ray flux of proton flares

25KUBO Yûki et al.

the same side. When the earth is located onthe side opposite the flare region, however, thetime variations of flux exhibit noticeable dif-ferences. When the current sheet comes in-between in reference to the right panel, theflux observed from the earth is considerablysmaller than that observed in the absence of acurrent sheet. This is probably due to a dimin-ishing number of particles traversing the cur-rent sheet as the particles moved along thecurrent sheet, resulting in less flux.

The presence of various structures, such asa current sheet in the solar wind, could thushave a significant bearing on the process ofsolar energetic particle propagation and ontime variations in the solar energetic particleflux on the earth, thereby stressing the needfor a model capable of reproducing solar windstructures in detail for space weather forecast-ing by means of simulation. Simulations capa-ble of reproducing the structures of solarwinds are detailed in References[12]and[13].

4 Space environment eventreporting system

One key goal of space weather forecastingis realizing a quantitative forecasting of suchspace environmental disturbances as solarflares and proton events far in advance ofactual occurrence, and then disseminating the

relevant information. At the moment, howev-er, much remains yet to be discovered orsolved regarding the physical processes ofsolar flare occurrence and solar energetic par-ticle generation. Moreover, given the fewmeans available for observing the internalconditions of the sun from which these activi-ties originate, we have a very long way to gotoward achieving this goal. Perhaps the secondbest way to disseminate information is collect-ing observational data in real time in order tospot space environment events sequentially, sothat solar flares and increases in solar ener-getic particles can be identified for immediate-ly reporting events that exceed a warning levelas these events occur to alert users.

Since about 10 years ago, information andcommunications technology has advancedrapidly, shaping up a framework for collect-ing, processing and reporting various sourcesof space environment data in quasi real time.As a result, the ability to detect the occurrenceof, and changes in, space environment eventsfrom observational data in real time and dis-seminate relevant information promptly hasreached a stage of practical usefulness. Real-time dissemination of alarm informationshould not only help bolster the safety andsecurity of artificial satellite operations,broadcasting and wireless communications,but should also aid in expediting the analysisof failure factors. Moreover, the computeralgorithm-based workflow of detecting eventsand making decisions on whether to reportsuch events is of objective value and, whenautomated or made autonomous, eases the taskof building a scheme of monitoring the spaceenvironment around the clock with relativeease. In addition, by setting the address of acellular phone as a reporting contact address,the owner of the cellular phone can be notifiedof the occurrence of anomalies by e-mail atany time (provided that the owner is withinthe cellular phone service area), withoutresorting to the Internet. For the reasons statedabove, we have developed a space environ-ment event reporting system and commis-sioned it into operation as one of many infor-

Fig.11 Positional relations among flareregion, current sheet, and theearth

Page 10: 2-1-2 Solar Energetic Particle Prediction and Alert · Fig.2 Relation between maximum proton intensity and proton flare fluence Fig.3 Relation between peak X-ray flux of proton flares

26 Journal of the National Institute of Information and Communications Technology Vol.56 Nos.1-4 2009

mation services.

4.1 System configurationThe system is modularized for each space

environment event, such as a solar flare orproton event. Each module consists of anevent detection process and an event reportingprocess. Figure 12 is a schematic diagram ofthe system configuration. The followingexplains why each module is divided into anevent detection process and an event reportingprocess.

To understand the properties of spaceenvironment events and enhance the technolo-gy for predicting the occurrence of suchevents, it is important to extract and analyzeall space environment environments, regard-less of scale. The frequency at which spaceenvironment events occur, however, is general-ly known to increase exponentially as thescale of these events narrows. Therefore, if allspace environments were reported upon detec-tion, a vast number of notifications-mostinvolving small-scale events-would result.Space environment events that could affectouter space utilization, social infrastructures,etc. are limited to larger-scale events, therebyrequiring an additional decision-makingprocess to determine whether to report onlythose events that have reached a given alertlevel, in addition to extracting all event infor-mation generated. The event extraction algo-rithm and reporting criteria can also be tuned

by editing the parameters coded in the config-uration file.4.1.1 Data flow

NOAA’s GOES satellite is equipped with aSpace Environment Monitor (SEM) to mea-sure solar X-ray flux, solar energetic particles,and radiation belt particles. Data collected bythe SEM is stacked on the SWPC data serverin quasi real time. We access this server everyfive minutes to collect the latest space envi-ronment data.

The Advanced Composition Explorer(ACE) satellite is a spacecraft that measuressolar wind at the L1 point. ACE supportsapparatus designed to measure solar windplasma (using SWEPAM), magnetic fields(using MAG), solar energetic particles (usingEPAM, SIS) and more. ACE is fulfilling anobservation mission as part of a global cooper-ation program to monitor solar winds aroundthe clock. NICT participates in this program toreceive data in real time[14]. ACE data isessentially collected in real time as well.4.1.2 Event detection process

The event detection process is initiated byreading a status file to collect informationabout the status of the last session of process-ing (e.g., the extent of processing conducted,whether an event occurred). This informationenables the process to determine from whendata is processed and in what sequence thedetection algorithm is run. There are two basic

Fig.12 Space environment reporting sys-tem schematic diagram

Fig.13 X3.4 class solar flare observed onDecember 13, 2006 (reprintedfrom the SWPC Website)

Page 11: 2-1-2 Solar Energetic Particle Prediction and Alert · Fig.2 Relation between maximum proton intensity and proton flare fluence Fig.3 Relation between peak X-ray flux of proton flares

27KUBO Yûki et al.

methods of event detection as follows:1) Determine the start of an event using a

variation (ΔX) in its physical quantity asa threshold.

2) Determine the start of an event using thesize (|X|) of its physical quantity as athreshold.

Method 1) is best suited for phenomenainvolving sharp variations in a physical quan-tity, such as a solar flare (Fig. 13); method 2)is suitable for gently varying phenomena, suchas proton events (Fig. 14). (For more informa-tion about the solar flare and proton event

detection algorithms, see Appendix.)4.1.3 Event reporting process

Space environment event informationextracted by the event detection process isstacked on the server in the form of an eventlist. By monitoring the updated status of thislist to meet a defined reporting requirement,the event reporting process can e-mail eventinformation to registered users as a prelimi-nary report.

Two event reporting requirements are pre-defined as follows:1) The scale of an event has reached a speci-

fied alert level.2) The duration of an event has exceeded a

specified length.Requirement 1) is common to all space

environment events, whereas requirement 2)applies to long-lasting variations of a protonevent or radiation belt particles. Its purpose isalert users to the status of an event now occur-ring.

The reporting thresholds are describedbelow. This information is coded in the con-figuration file and can be edited as needed.

[Solar flare preliminary reporting condition]An M class or higher (10-5Wm-2 or higher)flare has occurred.

Fig.14 Proton event (up to 700 PFU)observed on December 13, 2006(reprinted from the SWPC Web-site)

Table 1 Solar flare reporting example (for PCs)

Page 12: 2-1-2 Solar Energetic Particle Prediction and Alert · Fig.2 Relation between maximum proton intensity and proton flare fluence Fig.3 Relation between peak X-ray flux of proton flares

28 Journal of the National Institute of Information and Communications Technology Vol.56 Nos.1-4 2009

[Proton event preliminary reporting condition]An event with its 10 MeV or higher protonflux exceeding 10 PFU has started.A 10 MeV or higher proton flux has exceeded100,1000,10000 PFU.A 10 MeV or higher proton flux has fallenbelow 100,1000,10000 PFU.A 10 MeV or higher proton flux in excess of10 PFU has lasted for at least one day.A 10 MeV or higher variation of the protonevent has ended.

Event information is delivered in the for-mat described in Table 1. For the sake of userconvenience, the reporting format is availablein two versions: one intended for PCs withless text length constraints, and one for

portable terminals with a limited displayabletext length. In both versions, event-specificinformation, such as time and physical quanti-ties, is read from an event list into a fixed-phrase template file as variables. For portableterminals, an event graph viewing function isimplemented (Figs. 15, 16).

4.2 System operating statusSince its prototype was completed in

March 2000, the system has been used as atool to assist forecasters at NICT. Since Febru-ary 2004, the system had only been test-runfor reporting solar flares and proton eventswith a view toward verifying its working per-formance and identifying any problems thatmight impede steady operation. The test rununcovered delays in e-mail delivery and aneed for reporting the start of an event. Thesystem has subsequently gone through modifi-cations and, with an additional repertoire ofservices being set, has migrated to a produc-tion run since April 2005. It now supports agroup of just under 700 registered users.

5 Conclusions

This paper discussed the possibilities of astatistical approach and a simulation approachto predicting and issuing alerts on the occur-rence of solar energetic particles. Both tech-niques are still evolving and should benefitfrom ongoing research. Research on predict-ing and issuing alerts on solar flares and coro-nal mass ejections (CMEs), which are the ori-gins of solar energetic particles, is also neededbut remains in an evolutionary stage, with sta-tistical probability forecasting being the bestmethod attained thus far[15][16]. As a break-through, a space environment event reportingsystem that automatically detects and issuesalerts on the occurrence of solar flares andproton events has been developed. The systemprovides a useful means of detecting events inthe early stages of occurrence and disseminat-ing related information, and has alreadyreached a state of practical usefulness. Thus,research on predicting and issuing alerts on

Fig.16 Example of event plot (X-rayflare)

Fig.15 Links to the event plot destination(for cellular phones)

Page 13: 2-1-2 Solar Energetic Particle Prediction and Alert · Fig.2 Relation between maximum proton intensity and proton flare fluence Fig.3 Relation between peak X-ray flux of proton flares

29KUBO Yûki et al.

the occurrence solar energetic particles hasbeen actively pursued from various perspec-tives and promises a further leap.

Appendix

A.1 Solar flare detection algorithmThe solar flare detection algorithm is

described below. Assume that the physicalquantity handled (X-ray intensity: Wm-2) andthe time series (1-minute value) of timinginformation are X (t) and T (t), respectively. 1) Checking the status of event detection

Use an event list to check the detectionstatus of a solar flare (no event, event occur-ring (event start, maximum value)). The sub-sequent course of processing varies dependingon the status of event detection.2) Detecting “start of event” (1)

If “no event” applies in 1), define the timeT (k-3) at which the time-series data at fourpoints [X (k-3), X (k-2), X (k-1), X (k)] ini-tially meets all the following three require-ments as a solar flare start time, and assume“start of event” detection:A)X(k-3), X(k-2), X(k-1), X(k)>1.0× 10-7

B)X(k-3)<X(k-2)<X(k-1)<X(k)C)1.4×X(k-3)≦ X(k)3) Detecting “start of event” (2)

If the “start of event” has not been detect-ed in 2), the following algorithm is used todetect “start of event”:

Define the time T (k-9) at which the time-series data at 10 points [X (k-9), X (k-8), X(k-7), ....,X (k)] initially meets all of the fol-lowing three requirements as a solar flare starttime, and assume "start of event" detection:A)X(k-9), X(k-8), X(k-7),……., X(k)>1.0×10-7

B)X(k-9)<X(k-8)<X(k-7)<…….< X(k)C)1.2×X(k-9)≦ X(k)4) Detecting “maximum value and maxi-

mum time”Detect maximum value X (k) and its time

(maximum time) T (k) from the time-seriesdata following “start of event.” If the samemaximum value spans multiple times, consid-er the earliest time as the maximum time.

5) Detecting “end of event”Define the time T (k) at which the time-

series data following “start of event” initiallymeets all of the following two requirements asa solar flare end time, and assume “end ofevent” detection:A)X(k)<(Xstart+Xmax)/2(value at “start of

event,” Xmax: “maximum value”)B)T(k)>Tmax( “maximum time”)The same algorithm as used at the SWPC isused for detecting “start of event” (1). Thisalgorithm, however, is not capable of detect-ing a moderately starting event, such as aLong-Duration Event (LDE), and thereforehas been used in conjunction with the algo-rithm for detecting “start of event” (2) todetect an LDE.

A.2 Proton event detection algorithmThe proton event detection algorithm is

described below. Assume that the physicalquantity handled (10 MeV or higher protonflux: PFU) and the time series (5-minutevalue) of timing information are X (t) and T(t), respectively. 1) Checking the status of event detection

Use an event list to check the detectionstatus (no event, event occurring (event start,rising level, maximum value, falling level)). 2) Detecting “start of event” (rising level 1)

Define the time T (k) at which the time-series data at points num (X (0) to X (num))initially meets all of the following require-ments as an event start time, and assume “startof event” detection:X(k)> xlimwhere, xlim = 103) Detecting “rising level n”

Define the time T (k) at which the time-series data following “start of event” (includ-ing start time) initially meets the followingrequirement as a rising level n time, andassume “rising level n” detection. This timecan even be the start time:X(k)>xlv(n)where, xlv (2) = 102 (level 2), xlv (3) = 103

(level 3), xlv (4) = 104 (level 4), xlv (5) = 105

(level 5)

Page 14: 2-1-2 Solar Energetic Particle Prediction and Alert · Fig.2 Relation between maximum proton intensity and proton flare fluence Fig.3 Relation between peak X-ray flux of proton flares

30 Journal of the National Institute of Information and Communications Technology Vol.56 Nos.1-4 2009

References01 Fujitaka, K., Fukuda, S., and Yasuda, H., ISBN4–87639–407–5, in Japanese, 2004.

02 Kubo, Y. and Akioka, M., “Existence of thresholds in proton flares and application to solar energetic par-

ticle alerts,” Space Weather, Vol. 2, S01002, 2004.

03 Garcia, H. A., “Forecasting methods for occurrence and magnitude of proton storms with solar soft X

rays,” Space Weather, Vol. 2, S02002, 2004a.

04 Garcia, H. A., “Forecasting methods for occurrence and magnitude of proton storms with solar hard X

rays,” Space Weather, Vol. 2, S06003, 2004b.

05 Posner, A., “Up to 1–hour forecasting of radiation hazards from solar energetic ion events with relativis-

tic electrons,” Space Weather, Vol. 5, S05001, 2007.

06 Balch, C. C., “Updated verification of the Space Weather Prediction Center’s solar energetic particle

prediction model,” Space Weather, Vol. 6, S01001, 2008.

07 Laurenza, M., Cliver, E. W., Hewitt, J., Storini, M., Ling, A. G., Balch, C. C., and Kaiser, M. L., “A tech-

nique for short-term warning of solar energetic particle events based on flare location, flare size, and

evidence of particle escape,” Space Weather, Vol. 7, S04008, 2009.

08 Reames, D. V., “Solar energetic particles: Is there time to hide?,” Radiat. Meas., Vol. 30, pp. 297, 1999.

09 Zhang, M., Qin, G., and Rassoul, H., “Propagation of Solar Energetic Particles in Three-Dimensional

Interplanetary Magnetic Fields,” Astrophys. J., Vol. 692, pp. 109, 2009.

10 Dröge, W. and Kartavykh, Y. Y., “Testing Transport Theories with Solar Energetic Particles,” Astrophys.

J., Vol. 693, pp. 69, 2009.

11 Qin, G., Zhang, M., Dwyer, J. R., and Rassoul, H. K., “Interplanetary Transport Mechanisms of Solar

Energetic Particles,” Astrophys. J., Vol. 609, pp. 1076, 2004.

12 Nakamizo, A., Tanaka, T., Kubo, Y., Kamei, S., Shimazu, H., and Shinagawa, H., “Development of the 3-

D MHD model of the solar corona-solar wind combining system,” J. Geophys. Res., Vol. 114, A07109,

2009.

13 Nakamizo, A., Kubo, Y., and Tanaka, T., “A 3–D MHD Simulation Model of the Solar Corona-Solar Wind

System,” Special Issue of This NICT Journal, 2–3–2, 2009.

14 Maruyama, T., Watanabe, S., Ohtaka, K., and Shimazu, H., “Real Time Solar Wind Monitor Using ACE

Satellite,” J. Comm. Res. Lab., Vol. 43, No. 2, pp. 285, 1997.

15 Wheatland, M. S., “A statistical solar flare forecast method,” Space Weather, Vol. 3, S07003, 2005.

16 Kubo, Y., “Statistical Models for the Solar Flare Interval Distribution in Individual Active Regions,” Sol.

Phys., Vol. 248, pp. 85, 2008.

4) Detecting “maximum value and maxi-mum time”

Detect maximum value X (k) and its time(maximum time) T (k) from the time-seriesdata following “start of event.” 5) Detecting “falling level”Define the time T (k-2) at which the time-series data following the rising level initiallymeets the following requirement as a fallinglevel time, and assume “falling level” detec-tion:

X(k-2)<xlv & X(k-1)<xlv & X(k)<xlv[where, 2 < k < num]

6) Detecting “end of event”Define the time T (k-2) at which the time-series data following “start of event” initiallymeets the following requirement as an eventend time, and assume “end of event” detec-tion:X(k-2)< xlim & X(k-1)< xlim & X(k)<xlim[where, 2 < k < num]

Page 15: 2-1-2 Solar Energetic Particle Prediction and Alert · Fig.2 Relation between maximum proton intensity and proton flare fluence Fig.3 Relation between peak X-ray flux of proton flares

31KUBO Yûki et al.

KUBO Yûki

Senior Researcher, Space EnvironmentGroup, Applied ElectromagneticResearch Center

Solar Particle Physics

NAGATSUMA Tsutomu, Dr. Sci.

Research Manager, Space EnvironmentGroup, Applied ElectromagneticResearch Center

Solar-Terrestrial Physics

AKIOKA Maki, Dr. Sci.

Senior Researcher, Project PromotionOffice, New Generation NetworkResearch Center

Solar Physics, Optical System, SpaceWeather