Optical Illusion in Visualization

I was recently working on a few basic tests of MHD codes and one of the tests I did was the 2D rotor problem. In this test a circular region of high density initially starts out rotating. There is a magnetic field pointed in the +x direction. As the simulation progresses the magnetic field distorts the spinning high density region and causes magnetic waves to propagate into the surrounding low density region. Thus this test is a good one to show the basic structure of magnetic waves.

This test was done using Athena and the visualization was done in Paraview.  The 2D surface is colored according to the log10 of the density and the vector arrows show velocity and are colored according to the magnitude of the magnetic field.



One thing that came up with this visualization is that at the very beginning there is an optical illusion. Even though the gas is initially spinning in a counter-clock-wise direction because it is slowing down and begins to change direction, it appears as if it is spinning in the opposite direction. Because of the way the velocity vectors change length and direction at the beginning it gives the impression that the gas is spinning in the clock-wise direction. This is an excellent instance of an optical illusion in a visualization. Sometimes when scientists are trying to visualize their data they run the risk of having it create an optical illusion that will mess up their interpretation of the data. The way to fix this is to use a different visualizing method, in this case I could use particle tracers or stream line tracers to fix the problem.

But this is just one of the hazards of trying to use visualization to tell us something about our data. Still, visualization is better than just about any other method in giving a complete picture of what it going on. We just have to be mindful that sometimes our minds will play tricks on us.

The Sun and I

As most of you know, I have ventured into the realm of solar modeling in my research here at Colorado. Specifically, I work of global-scale models of the sun's convection zone, which extends from about 75% of the sun's radius to about 98% of the sun's radius. In this area, energy is being transported mainly by convection - hot fluid rises and cold fluid sinks. When you add a magnetic field, this creates a self-sustaining dynamo that turns convective energy into magnetic energy, which creates a sustained magnetic field in the sun's convection zone.

Magnetic fields play an important role in solar activity. When magnetic field lines poke up out of the solar surface, you get sun spots, flares, and coronal mass ejections - the fireworks. In order to better understand what is driving these explosive events, we need to better understand what is driving the sun's magnetic field and how it changes in time.

Below is a movie created from data from our numerical models of the sun's convection zone. This particular model has been "spun-up" to three times the solar rotation rate in order to exaggerate the effects and speed up the solar cycles. A normal solar magnetic cycle takes 22 years - in this simulation, we see strong magnetic variability in cycles that take about 2 years of simulation time. The movie specifically shows some 3-D visualizations of the toroidal magnetic field (i.e. the magnetic field with the dipole part removed). The positive (in the direction of the sun's rotation) field is displayed in red (strongest) to yellow (weakest). The negative field is displayed in blue (strongest) to purple (weakest). For ease of viewing, only the strong fields are displayed.

As you can see, the magnetic field has organized itself into two bands of magnetic field in the tropics. In this cycle, both bands have been greatly weakened in the second image - particularly the negative (blue/purple) band. Since our model uses constant inner and outer boundary conditions, this clearly shows that the sun's magnetic variability can be caused purely by oscillations in the convection zone.

Aside from the science we are able to do with these types of visualizations, it's also just fun to be able to actually see the 3-D data we work with. In the future, I think that 3-D visualizations are going to become more and more common as more numerical models progress into three dimensions.

These videos were created using Vista, developed at the San Diego Supercomputing Center. Special thanks to Steve Cutchin at SDSC for working with me to get Vista running with our data.