SKYNET ALGORITHM FOR SINGLE-DISH RADIO MAPPING
Undergraduates: Dylan Dutton, John Martin, Logan Barnes
Faculty Advisor: Dan Reichart
Department: Physics & Astronomy
Collecting data and taking measurements has always been the most fundamental concept in science. Unfortunately, this data can often become contaminated and radio astronomy is of no exception to these contaminations. A combination of exceedingly sensitive detectors on modern radio telescopes and crowding of the radio spectrum due to the use of everyday devices (cell phones, Wi-Fi, etc.) has caused contamination of astronomical data to become a serious concern. The scope of this project aims to reduce these contaminants to sub-noise levels by using a combination of local modeling and a rigorous outlier-rejection algorithm. These common contaminants can be separated into three categories: En-route signal drift, radio-frequency interference (RFI), and elevation-dependent signal. Preliminary simulation-based testing shows that the algorithm is effective at removing large-scale structures. For example, en-route drift and long-duration RFI are reduced by factors of ~61 and ~590, to ~5% and ~36% of the noise level when using a 6-beamwidth scale, respectively. Large-scale astronomical and elevation-dependent signals are reduced by factors of ~510 and ~710 to ~9% and ~5% of the noise level, respectively, when using a 6-beamwidth scale.