Creating a combined telescope dataset - adeich/APF_telescope_behavior_study GitHub Wiki

The Lick Observatory Automated Planet Finder (APF) telescope measures the radial velocity of stars--periodicity in this measurement over days, months, and years often reveals a planet orbiting that star. The APF infers the radial velocity by recording the star's spectrum and measuring its red shift. The APF has a measurement precision of 1 m/s -- this is, for perspective, out of a typical ~30,000 km/s radial velocity of a given star.

The APF measures several dozen stars every night, pausing for a few minutes on each. Most of these stars are observed for the purpose of hunting for planets. But there are three, specific stars that APF keeps returning to, usually observed once per night, which are already known to have zero planets. These stars serve as a barometer of velocity measurement error.

The spread of velocity measurements of these three standard stars has been much larger than expected for the APF; often by around a factor of 2 or 3. Our intention here is to study and characterize this spread in velocity and to hunt for its correlation with other, simultaneously recorded telescope sensor data. Ultimately we would like to (1) pinpoint the biggest instrumental correlations of the additional error and (2) make a strong enough prediction of error so as to anticipate it.

The sensor data ("telemetry") is recorded around once per second during telescope exposures. Some of the more important-seeming data channels include:

• Wind speed and direction.
• Temperatures--- averaged around the dome; of each mirror; of the air above a mirror.
• Telescope direction: azimuth and elevation.
• Stellar activity: a characterization of how noisy the stellar light is at the moment.