Low-level Data Analysis, Calibration, and Ground Modeling with COMAP

Abstract

The CO Mapping Array Project (COMAP) is a line intensity mapping experiment currently in its Pathfinder phase, where it is targeting CO emissions at redshifts z=2.4-3.4. A main goal of the Pathfinder phase is to build a solid analysis and modeling framework and achieve detection of CO at the current redshifts. Central to this effort is the proper understanding and handling of different signal systematics. In this thesis, we perform a comprehensive analysis of COMAP signal systematics and suggest improved methods for two of the current filters in the low-level data analysis pipeline, as well as the calibration. We claim that the currently employed frequency filter, primarily meant to target gain fluctuations and temperature continuum sources, is inadequate at handling the latter. We propose a new frequency filter, which performs a joint maximum likelihood fit of both quantities. One of the most critical systematics in the COMAP data is ground pickup by the far sidelobes of the telescope, currently handled by the pointing template filter. We propose a new way of constructing pointing templates, using ground pickup maps created from the COMAP data itself. Using a destriper mapmaking model, we create examples of such ground maps, and perform a preliminary analysis of the viability of introducing such a data-based pointing template model. We complement this analysis with ground pickup maps from a simulated beam profile. We find that our new frequency filter outperforms the current filter at removing temperature systematics in simulated data. The filter also significantly improves the removal of continuum foregrounds. With Jupiter as a case-study, it reduces the mean squared signal residual from 51 sigma to 1.5 sigma. Our destriper demonstrates the feasibility of producing data-driven ground maps, which can be employed in a more refined pointing template than the one currently employed, but also highlights several challenges which must be overcome. Our simulated ground pickup qualitatively corresponds well to maps produced by our destriper, but there are discrepancies, especially in the observational range of CO2, which warrants further analysis. We find the simulated beam profile to have a complicated frequency structure, which results in a ground profile that depends non-trivially on both frequency and pointing. Finally, our calibration analysis concludes that a calibration vane angle of 69 degrees or below produces acceptable results, and we use this analysis to implement a new and more robust scheme for hot load measurements.