What you're seeing is a model of a spinning star that is covered in starspots (grey circles), and transient flares (red dots). The starspot sizes and impact on the resulting light curve (bottom panel) are generated using a real and super neat starspot modeling code one of my collaborators has built. The flares occur at random (based on some "rate" parameter I've set), but their sizes and shapes are based on results from a pair of papers my PhD thesis advisor and I wrote in 2014.
Remember, when studying stars with Kepler all we get is the light curve (bottom panel), and it is very difficult (sometimes impossible) to infer the true spatial distribution of spots and flares on the surface. A model like this can help us visualize what really may be happening on the star, and how our intuition can betray us.
I've tried to make everything going in to the light curve as physically realistic as possible, making this a "phenomenological model" of sorts. As a result, we get a model that looks very similar to real data. For example, when I run the model forward 10 more rotations and stack all the data, you see there is no strong correlation between # of flares and rotation phase. This despite the presence of a large active polar starspot that dominates the spot light curve.
This looks very similar to a result from Hawley & Davenport et al. (2014). A modified version of the published figure is below that also shows no correlation with flare energy (labeled Ekp) and rotation, indicating big flares and small flares come from all over the star.
I originally made this toy model and animation to help explain results from Kepler in a talk. The cool thing is that we might actually be able to do some real science using this model, both in generating mock data to use for training, and in estimating flare rates and flare properties for real stars!
Of course, the code to make this is available on GitHub as well.
Written Friday, February 12, 2016