Image Details
Caption: Figure 8.
40 ft gain calibration data, with the noise diode first on and then off, and the best-fit model. Circled points have been robust-Chauvenet rejected, including data taken during the transitions from off to on and on to off, and RFI-contaminated data. The background level increased during the calibration, but our model accounts for this: simply averaging each level, instead of modeling each level with a line, would have underestimated the result.
© 2019. The American Astronomical Society. All rights reserved.