2D STOCHASTIC SIMULATION MODEL OF COSMIC RAY MODULATION: COMPARISON WITH EXPERIMENTAL DATA P.Bobik(1) (3), M.Gervasi(1) (4), D.Grandi(1) (4), P.G.Rancoita(1), I.G.Usoskin(2) (1) INFN sezione di Milano, P.zza delle Science 3, 20126 Milano, Italy, piergiorgio.rancoita@mib.infn.it (2) Sodankyla Geophysical Observatory, University of Oulu, Finland, ilya.usoskin@oulu.fi (3) Institute of Experimental Physics, Watsonova 47, 04053 Kosice, Slovak Republic, bobikp@kosice.upjs.sk (4) Universitá di Milano-Bicocca, P.zza della Scienza 3, 20126 Milano, Italy, massimo.gervasi@mib.infn.it ABSTRACT We developed a 2D stochastic simulation model of heliospheric propagation of galactic cosmic rays. The model solves numerically the transport equation of particles in the heliosphere. In the calculation we use also drift effects which are included through analytical effective drift velocities. We estimated the cosmic rays spectrum at 1AU using this model formalism. The calculated spectra are compared with other models (CREME96) and with experimental data (IMP8 and AMS) for positive (A>0) and negative (A<0) solar periods. 1. INTRODUCTION In the last years models of heliospheric propagation of galactic cosmic rays (GCR) are continuously developing. Models begin to use new structure of heliospheric magnetic field [1] which has been developed after Ulysses measurements of small latitudinal gradient of cosmic rays intensity in the inner heliosphere. One of most recent is a model with mixed Fisk and Parker fields, which is a pure Fisk heliospheric magnetic field at mid-latitudes but changes to a Parker field in the solar equatorial plane and above the solar poles [2]. These models also became time dependent due to the measurements connected to drift effects in the heliosphere [3]. Since a few years we have developed a 1D heliospheric model using the stochastic simulation approach [4] and [5]. Now we have developed a two dimensional (radius and latitude) drift model of GCR propagation in the heliosphere based on the same Monte Carlo approach. The model is dependent on the charge sign of particles due to curvature, gradient and current sheet drifts. It is time dependent too due to the variation of the measured values of the solar wind velocity in the ecliptic plane (V) and the tilt angle (). U 1 2 U 2 (r K rr ) t r r r 1 U (1) 2 ( K sin ) r sin 1 (r 2V ) 1 2 (TU ) 2 (r 2VU ) 3r r T r r where is the heliolatitude, U is the cosmic ray number density per unit interval of kinetic energy T (per nucleon), r is radial distance and V solar wind velocity. = (T+2T0)/(T+T0) and T0 is proton’s rest energy. The first two terms of equation (1) describe diffusion of GCR in the heliosphere, the third term adiabatic energy losses and the last one convection by outgoing solar wind. Using a stochastic simulation technique, particle co-ordinates variation in a small time step t results: 1 d (r 2 K rr ) t V t Rg 2 K rr t r2 dr d 2 K 2 K 1 2 t Rg 2.(1 ) 2 t d r r r (2) Where r is the radial variation, = cos is the latitudinal variation of the particle, Rg is the Gaussian distributed random number with unit variance, t is the time step of calculation. Radial diffusion coefficient is Krr K cos2 K .sin 2 (3) where is the angle between radial and magnetic field directions [6]. The latitudinal coefficient is K K (4) where parallel and perpendicular coefficients are 2. MONTE CARLO 2D MODEL WITH DRIFT EFFECTS Propagation of GCR across the heliosphere is described by Fokker–Planck equation. 2D heliospheric stochastic simulation is based on equation for GCR transport in the heliosphere (without drift terms) [6] and [7]: K K 0 K P ( P) K ( K )0 K B 3B (5) (6) where K0= 2 x 1022 cm2s-1 [8], β is the particle velocity (in unit of the light speed), Kp(P) take into accounts the dependence on rigidity (in GV), (K)0 is the ratio among parallel and perpendicular diffusion coefficient, B ( 5 nT) is the value of heliospheric magnetic field at the Earth orbit, and B is the Parker field [9]. We use the Parker model of heliospheric magnetic field because only for this model an analytical solution for drift velocities exists. For the Fisk model there are difficulties with the evaluation of diffusion tensor and with the analytical solution for drift velocity. Moreover solutions of the transport equation in the Fisk field model can become unstable easier than in the Parker field case. In our model a solar wind speed V() as function of the heliolatitude has been used [10]: V = V0 (1 + sin ), km s-1, for 0o < < 60o V = 750 km s-1, for 60o < < 90o (7) Where V0 is the velocity of solar wind in ecliptic plane. The heliosphere is considered to be a sphere with radius equal 100AU. The model does not include an effect of the termination shock. Drift effects are included through analytical effective drift velocities [11]. The average drift velocity is vd = (/3B) (8) Where is the speed ratio, is the CR particle’s rigidity and B is the magnetic field. In the Parker spiral field, gradient, curvature and drift along the neutral sheet are added to the previous formulas (2) to calculate position variation of the particle during a time step t: rd r vg vdns t vq t d cos arccos( )+ arctg r (9) Where rd is the radial variation with drift effect, d is the latitudinal variation of the particle, vg is the velocity of gradient drift, vdns is the velocity of neutral sheet drift and vθ is the velocity of curvature drift. Both vg and vdns are directed along er, while vθ is directed along eθ in spherical coordinates. We use Burger’s model [12] as Local Interstellar Spectrum of protons (LIS hereafter). The heliospheric propagation model has been optimized by fine tuning the parameters Kp(P) and (K)0 of the diffusion tensor. We performed long time consuming simulations and compared results with the available experimental data. In particular we are interested to the flux spectrum at 1AU and as a reference data set we used both the IMP8 measurements and Creme96 model. Moreover we concentrated on periods with low-medium solar activity and opposite solar field polarity. Finally we get the best combination of these two parameters of the model producing spectra and comparing them with the data set at 1AU for both positive and negative solar field periods. Regarding the ratio between perpendicular and parallel diffusion coefficients (K)0 we obtain as best value (K)0 = 0.025. This is the same value reported by Giacalone et al. [13]. For the dependence of the diffusion tensor on rigidity we have found similar but not the same results obtained in [13]. Our simulations produce the best spectra at 1AU for Kp(P) P0.78, which corresponds to a slope 2 – q, where q is the slope of the high energy part of LIS. The heliospheric propagation model has been optimised mainly for periods of minimum solar activity and negative polarity. As a matter of fact negative polarity periods are more sensitive to parameters variation. Not only (K)0 and Kp(P), but also variations of other parameters, like the tilt angle, induce very important changes in the modulated spectrum at 1AU. 3. EXPERIMENTAL DATA BASE We selected two periods in 1987 and 1998 during solar minima in two opposite solar cycles. For these periods we have experimental data to compare to the model and to estimate the charge drift effect. In our model different periods of solar activity are uniquely characterized by the measured value of the tilt angles and solar wind velocities V in ecliptic plane. Experimental values of the tilt angle during years 1987 and 1998 are (see the web page http://quake.stanford.edu/~wso/wso.html): 1987 : A < 0, 3o – 21o 1998 : A > 0, 16o – 50o We considered average values for both these periods: for 1987 we choose the value = 10o; for 1998 = 30o - 40o. We also consider as solar wind speed in ecliptic plane at 1AU the value V = 400 kms-1 for both periods. We compared results of the simulations with measurements of IMP8-GME experiment and IMP8CRNC experiment and with Creme96, a model of GCR in heliosphere. We selected measurements from IMP8 satellite apparatus, in particular from the Goddard Medium Energy (GME) experiment (see the web page http://spdf.gsfc.nasa.gov/imp8_GME/GME_home.htm) and from the Cosmic Ray Nuclear Composition (CRNC) (see the web page http://ulysses.sr.unh.edu/ WWW/Simpson/imp8.htm). These experiments cover positive and negative solar periods from year 1973 to year 2002 and energy range from 0.5 MeV to 500 MeV for protons. We also used the model Creme96 [14] (see the web page http://crsp3.nrl.navy.mil/creme96/) as a smooth reference spectrum, even if it is not reproducing completely the experimental data. 4. RESULTS In Fig. 1 results from the present 2D drift model spectrum for the year 1987 ( = 10o) are compared with IMP8 measurements and Creme 96 spectrum. In this period, i.e. minimum solar activity and negative polarity, a good agreement of the model results with experimental data is obtained. Besides both experimental data and the 2D model are different from Creme96 model, but difference is still inside the 25% error bar of the Creme96 specrtum. In Fig. 2 calculations based on the present 2D drift model spectrum for June 1998 ( = 30o) are compared with IMP8 measurements and Creme 96 spectrum. In this period, i.e. minimum solar activity and positive polarity, a good agreement among the model results, experimental data and Creme96 model is obtained for energies higher than 100MeV. For lower energies the measured spectrum and Creme96 model are systematically higher respect to our model. This is probably due to an additional component in the measured flux, which can be ascribed to anomalous cosmic rays: both measurements and Creme96 model take into account this extra component. Besides we need to remember that this period is not a really quite solar activity period, as it is confirmed by the measured value of the tilt angle ( = 30o). Fig. 1. Proton differential flux at 1AU. Comparison of our 2D drift model with IMP8 measurements and Creme96 model; year 1987, A 0 and = 10o. Fig. 3. Proton differential flux at 1AU. Comparison of our 2D drift model with AMS-01 measurements and Creme96 model; year 1998, A 0, =30o and 40o. Fig. 2. Proton differential flux at 1AU. Comparison of our 2D drift model with IMP8 measurements and Creme96 model; year 1998, A 0 and = 30o. In Fig. 3 we compare our simulated spectra with AMS01 data. As we can see the best value of the tilt angle is intermediate between 30o and 40o. The dependence of simulated spectra at 1AU on the tilt angle during positive polarity periods is shown in Fig. 4. It is evident the depletion of galactic protons at low energies by increasing the tilt angle. This trend is still more evident for negative polarity periods. the model is able to produce spectra at 1AU which reproduce experimental measurements quite well. Agreement is better for the negative solar period explored (1987) when the influence of drift effect is stronger and the solar activity is lower than in the opposite solar cycle. As a result we obtain that proton fluxes at low energy strongly depend on heliospheric current sheet position, i.e. on the tilt angle. REFERENCES 1. Fisk, L.A., J. Geophys. Res., Vol.101, 15547-15553, 1996. 2. Burger R. A. and Hitge M., American Geophysical Union, SH71A-04, Fall Meeting 2002. 3. Wibberenz G., Ferreira S. E. S., Potgieter M. S., Cane H. V., Space Science Reviews, Vol. 97, Issue 1/4, 373376, 2001. 4. Gervasi M. et al., Nuclear Phys. B (proc. suppl.), Vol. 78, 26-31, 1999. Fig. 4. Proton flux at 1AU for several tilt angles during positive polarity periods (A>0). The LIS is also shown. 5. Gervasi M. et al., Proceedings of the 26th ICRC – SH 3.1.18, 69-72, Salt Lake City – Utah, August 17-25, 1999. 6. Potgieter M. S., Le Roux J. A., Burlaga L. F., McDonald, F. B., Astrophysical Journal, Part 1 (ISSN 0004-637X), Vol. 403, no. 2, p. 760-768, 1993. 7. Fisk L. A., J. Geophys. Res., Vol. 81, 4646-4650, 1976. 8. Potgieter M. S., Le Roux, J. A., Astrophysical Journal, Vol. 423, 817, 1994. 9. Burger R.A., and Potgieter M.S., The Astrophys. J., Vol. 339, 501-511, 1989. Fig. 5. Proton flux at 1AU vs tilt angle for particles at energy equal to 5.0, 3.1, 2.0, 1.2, 0.8, and 0.5GeV during a positive polarity period (A 0). The flux depletion of galactic protons is still more evident if we plot the intensity vs the tilt angle for several energy bins. From Fig. 5 we can se as the modulation is more effective at lower energy while it disappears at higher energy. CONCLUSION We developed a 2D stochastic simulation model of heliospheric propagation which includes charge drifts effects. For both, positive and negative solar periods, 10. Reinecke J. P. L., Steenberg C. D., Moraal H., McDonald F. B., Advances in Space Research, Vol. 19, Issue 6, 901-904., 1997. 11. Hatting M. , and Burger R.A., Adv. Space. Res., Vol. 16, No. 9, 213-216, 1995. 12. Burger R. A., Potgieter M. S., Heber B., J. Geophys. Res., Vol. 105, Issue A12, 27447-27456, 2000. 13. Giacalone J., Jokipii J. R., The Astrophysical Journal, Vol. 520, Issue 1, 204-214., 1999. 14. Tylka A.J. et al., IEEE Trans. Nuclear Sci., Vol. 44, 2150-2160, No. 6, 1997.