Presentation is loading. Please wait.

Presentation is loading. Please wait.

Numerical MHD modelling of waves in solar coronal loops Petr Jelínek 1,2 and Marian Karlický 2 1 University of South Bohemia, Department of Physics České.

Similar presentations


Presentation on theme: "Numerical MHD modelling of waves in solar coronal loops Petr Jelínek 1,2 and Marian Karlický 2 1 University of South Bohemia, Department of Physics České."— Presentation transcript:

1 Numerical MHD modelling of waves in solar coronal loops Petr Jelínek 1,2 and Marian Karlický 2 1 University of South Bohemia, Department of Physics České Budějovice, Czech Republic 2 Astronomical Institute, Academy of Sciences of the Czech Republic Ondřejov, Czech Republic

2 České Budějovice Small university town in the South of Bohemia with inhabi- tants Approximately 200 km from Wien and 100 km from Linz University of South Bohemia – 7 faculties and 2 research institutes, about students

3 Ondřejov Small village outlying 30 km from Prague – Astronomical Institute, Academy of Sciences of the Czech Republic Founded in 1898, 4 main scientific departments

4 Outline Motivation of numerical studies Equations of magnetohydrodynamics (MHD) Numerical methods & solutions Results  1D model – impulsively generated acoustic waves  1D model – gravitational stratification  2D model – impulsively generated acoustic waves  2D model – modelling of wavetrains Conclusions

5 Motivation of numerical studies – I. Oscillations in solar coronal loops have been observed for a few decades The importance of such oscillations lies in their potential for the diagnostics of solar coronal structure (magnetic field, gas density, etc.) The various oscillation modes in coronal loops were observed with highly sensitive instruments such as SUMER (SoHO), TRACE The observed oscillations include propagating and slow magnetosonic waves. There are also observations of fast magnetosonic waves, kink and sausage modes of waves

6 Motivation of numerical studies – II. Coronal loop oscillations were studied analytically but these studies are unfortunately applicable only onto highly idealised situations The numerical simulations are often used for solutions of more complex problems – these studies are based on numerical solution of the full set of MHD equations Mentioned studies of coronal loop oscilla- tions are very important in connection with the problem of coronal heating, solar wind acceleration and many unsolved problems in solar physics Magnetohydrodynamic coronal seismology is one of the main reasons for studying wa- ves in solar corona

7 MHD equations In our models we describe plasma dynamics in a coronal loop by the ideal magnetohydrodynamic equations The plasma energy density The flux vector

8 Numerical solution of MHD equations The MHD equations (1) – (4) are transformed into a conservation form For the solution of the equations in conservation form exist many numerical algorithms including professional software such as NIRVANA, ATHENA, FLASH,.... (www.astro-sim.org)

9 Numerical methods – I. There exist a lot of numerical methods used for the solution of equations in conservation form in numerical mathematics Generally we can use the two types of numerical methods  explicit methods – calculate the state of a system at a later time from the state of the system at the current time easy to programming  unstable in many cases  implicit methods – find the solution by solving an equation involving both the current state of the system and the later one unconditionally stable  difficult to programming (tridiagonal matrix; solution by Thomas algorithm)

10 Numerical methods – II. We use only explicit methods in our calculations for this reason we must use the artificial smoothing for the stabilisation of the numerical scheme Some mathematical definitions of numerical methods for PDEs  Consistency – the numerical scheme is called consistent if  Convergence – the numerical method is called convergent if

11 Numerical methods – III. For the solution of the MHD equations in a conservation form the methods of so-called flux limiters are used These numerical methods are able to jump down the oscillations near sharp discontinuities and jumps Generally, for the solution of PDE in conservation form in 1D we can write

12 Numerical methods – IV. Many authors often use the linear methods  upwind scheme  Lax-Wendroff scheme (downwind slope)  Beam-Warming scheme (upwind slope)  Fromm scheme (centered slope)

13 Numerical methods – V.

14 Numerical methods – VI. To avoid the “overshoots” we limit the slope by flux limiter methods  minmod  superbee  MC  van Leer And many others – van Albada, OSPRE, UMIST, MUSCL schemes

15 Numerical methods – VII.

16 1D model of acoustic standing waves There exists a lot of types of oscillations in solar coronal loop  acoustic oscillations  kink and sausage oscillations  fast and slow propagating waves,... Acoustic oscillations are easy to simulate, they can be modelled in 1D, without magnetic field, etc. Kink and sausage oscillations were directly observed (SOHO, TRACE) and there are many unanswered questions – excitation and damping mechanisms, etc. We focused on the impulsively generated acoustic standing waves in coronal loops

17 1D model – initial conditions Initial condtitions The length of the coronal loop was L = 50 Mm which corresponds to loop radius about 16 Mm. The loop footpoints were settled at positions x = 0 and x = L

18 1D model – perturbations Perturbations In the view of our interest to study impulsively generated waves in the solar coronal loops, we have launched a pulse in the pressure and mass density The pulse had the following form

19 1D model – numerical solution The numerical region was covered by a uniform grid with cells and open boundary conditions that allow a wave signal freely leave the region were applied The time step used in our calculations satisfied the Courant- Friedrichs-Levy stability condition in the form In order to stabilize of numerical methods we have used the artificial smoothing as the replacing all the variables at each grid point and after each full time step as

20 Results – 1D model Time evolution of velocity v(x = L/4,t), mass density  (x = L/4,t) (top panels) and spatial profiles of velocity v(x,  t), v(x,7.12T 1 ) (bottom panels); all for mass density contrast d = 10 8, pulse width w = L/40, and initial pulse position x 0 = L/4.

21 Results – 1D model Time evolution of velocity v(x = L/4,t), mass density  (x = L/4,t) (top panels) and spatial profiles of velocity v(x,  t), v(x,7.89T 2 ) (bottom panels); all for mass density contrast d = 10 8, pulse width w = L/40, and initial pulse position x 0 =L/2.

22 Time evolution of velocity v(x = L/4,t), mass density  (x = L/4,t) (top panels) and spatial profiles of velocity v(x,  t), v(x,11.00T 1 ) (bottom panels); all for mass density contrast d = 10 8, pulse width w = L/40, and initial pulse position x 0 = L/50. Results – 1D model

23 Time evolution of average pressure, increased by the factor 10 3, initial pulse position x 0 = L/4 (left top panel), x 0 = L/2 (right top panel) and x 0 = L/50 (bottom panel), mass density contrast d = 10 8 and pulse width w = L/40; note that x-axis is in the logarithmic scale. Results – 1D model

24 Fourier power spectra of velocities v for initial pulse position x 0 = L/2 (left) and x 0 = L/4 (right), mass density contrast d = 10 5 (top panels) and d = 10 8 (bottom panels) and pulse width w = L/40. The amplitude of the power spectrum A(P) is normalized to 1.

25 Results – 1D model Time evolution of total (red), pressure (blue) and kinetic (green) energies for various positions in numerical box. Left upper panel – whole simulation region, left upper panel – “transition region”, bottom panel – “coronal region”. The initial pulse position x 0 = L/4, d = 10 8 and pulse width w = L/40.

26 Results – 1D model Time evolution of total (red), pressure (blue) and kinetic (green) energies for various positions in numerical box. Left upper panel – whole simulation region, left upper panel – “transition region”, bottom panel – “coronal region”. The initial pulse position x 0 = L/2, d = 10 8 and pulse width w = L/40.

27 1D – gravitational stratification To create more realistic model the gravitational stratification was added We consider a semi-circular loop with the curvature radius R L, in this model we incorporate the effect of loop plane inclination the shift of circular loop centre from the baseline was omitted

28 1D – gravitational stratification – I. The gravitational acceleration at a distance s measured from the footpoint along the loop, is The MHD equation of motion has the following form For the plasma pressure in the loop we can write

29 The temperature profile was calculated by means of this formula 1D – gravitational stratification – II. The length of the coronal loop was L = 100 Mm in this case which corresponds to loop radius about 32 Mm. The mass density was calculated from

30 Gravitational stratification – first results in 1D Time evolution of velocity v(x = L/4,t), mass density contrast d = 10 2, pulse width w = L/80, and initial pulse position x 0 = L/4 and x 0 = L/2, inclination angle  = 0° (blue line) and  = 45° (red line).

31 2D modelling of magnetoacoustic standing waves We consider a coronal slab with a width w = 1Mm and mass density  i, embedded in a environment of mass density  e The pressure, mass density, temperature and initial pulses in pressure and mass density are calculated similarly as in 1D model

32 Numerical solution in 2D For the solution of 2D MHD equations the Lax-Wendroff numerical scheme was used, this method is often used for the solutions of MHD by many authors Step 1 Step 2 The stability condition

33 Results – 2D model Time evolution of velocity v(x = L/4, y = 0, t) (left top panel). Spatial profile of x-component of velocity – v x at time t = 8.17 T 1 (right top panel) and the corresponding slices of v x along y = H/2 (x = L/2) – bottom left (right) panel; all for mass density contrast d = 10 8, pulse width w = L/40, and initial pulse position x 0 = L/2.

34 Results – 2D model Time evolution of velocity v(x = L/4, y = 0, t) (left top panel). Spatial profile of x-component of velocity – v x at time t = 6.15 T 2 (right top panel) and the corresponding slices of v x along y = H/2 (x = L/4) – bottom left (right) panel; all for mass density contrast d = 10 8, pulse width w = L/40, and initial pulse position x 0 = L/4.

35 Modelling of wave trains in 2D The wave trains were directly observed and discovered by SECIS (Solar Eclipse Coronal Imaging System) Observed in Ondrejov in radio waves The theoretical description is needed – the comparison of observed and modelled tadpoles → what type of waves are present

36 We study impulsively generated magnetoacoustic wave trains propagating along a coronal loop The problem is modelled by means of 2D model presented before, but magnetic field is parallel to the y axis The equilibrium is perturbed by a pulse in velocity, situated at L/4 of the numerical domain Modelling of wave trains in 2D

37 Wave trains – first results – I. The spatial profile of the velocity v x at time t = 30 s from initial pulse (left upper panel), and corresponding slices of v x along x axis (y = H/2) (right upper panel) and along y axis (x = L/4). Initial pulse position x 0 = L/4, mass density contrast d = 10 8, pulse width w = L/40

38 Wave trains – first results – II. Time evolution of mass density  (x = L/2,t), (top panel) and corresponding wavelet analysis (bottom panel); all for mass density contrast d = 10 8, pulse width w = L/40, and initial pulse position x 0 = L/4.

39 Conclusions – I. Computer modelling seems to be very useful tool for the understan- ding of processes in solar coronal loops The next step in our research will be the extension of current model to three dimensions (by means of mentioned software – Athena, Nirvana, FLASH...), including the source terms such as cooling term, heating term, gravitational stratification, etc. By means of this model we could investigate effects like attenuation of waves in coronal loops, plasma energy leakage by the dissipation into solar atmosphere and more very interesting problems in solar coronal physics...

40 More informations about 1D or 2D models can be found in  Jelínek P., Karlický, M.: Numerical Modelling of Slow Standing Waves in a Solar Coronal Loop, Proc. 12th ESPM, Freiburg, Germany, 2008  Jelínek, P., Karlický, M.: Computational Study of Implusively Generated Standing Slow Acoustic Waves in a Solar Coronal Loop, Eur. Phys. J. D, after revisions. Conclusions – II.

41 References [1] M. Aschwanden, Physics of the Solar Corona (Springer, Praxis Publ., Chichester UK 2004). [2] T. J. Chung, Computational Fluid Dynamics (Cambridge University Press, New York USA 2002). [3] E. R. Priest, Solar Magnetohydrodynamics (D. Reidel Publishing Company, London England 1982). [4] M. Selwa, K. Murawski, S. K. Solanki, A&A 436, 701 (2005). [5] Tsiklauri, D., Nakariakov, V. M., A&A, 379, 1106 (2001). [6] Nakariakov, V. M. et al.: Mon. Not. R. Astron. Soc., 349, 705 (2004).

42


Download ppt "Numerical MHD modelling of waves in solar coronal loops Petr Jelínek 1,2 and Marian Karlický 2 1 University of South Bohemia, Department of Physics České."

Similar presentations


Ads by Google