I found it really interesting so I set out on a mission to find a way to describe the function apparent in the scatter plot. I searched for the best way to find the correlation in a dataset with an obvious non-linear function. Most of my clues to answers were found in comments at stackexchange. I ended up settling on polynomial regression. I found an implementation in javascript that I could understand and modify for my needs and I built a little application to extract the scatter plot points from the original image and use the points to feed the regression analysis and then plot the results. It worked first try. I was shocked. LOL The data I sampled was the dots that were all the darkest blue on the original image. I wanted the whole set but I couldn't get it to work without making multiple passes with the different pixel colors so I just used the sample since it was pretty close to representing the whole. It was just a test anyway. I'll refine it later on and get all the points. Anyway, it's fun playing around with this stuff.

polynomialregressionceressst.jpg