Simulating correlated noise

---

glossary of terms

What is correlated noise?

Correlated noise refers to any non-random interference in an astronomical observation. The causes can be immensely varied, ranging from cloud cover to instrumental temperature variation to stellar rotation. Not taking correlated noise into account can have severely detrimental effects on data analysis and results - skewing best fit curves and leading to spurious signal detections. That last effect is even more significant for data features with a low signal to noise (S/N) ratio, such as planet detections.

How do we simulate it?

There are several methods for simulating the effects of correlated noise, and statistical frameworks accounting for it in best-fit analysis, but short answer for how I chose to simulate it in my dissertation is: I didn't.

Simulating correlated noise from scratch would have required me to consider dozens of potential sources and hundreds of extra parameters. Even then, I would have no guarantee that my simulation would actually be representative of a real observation. Instead, I opted to take existing light curve data, isolate the non-microlensing segment, and inject my simulated light curve into it. This way I can be entirely confident that the noise present for the detection routine is realistic.