Kaylah Breanne McGowanEnhancing LIGO's Sensitivity via Detector Characterization & Noise Mitigation
Discipline: Physics
Subcategory: Astronomy and Astrophysics
Session: 3
Room: Woodley Park
Kaylah Breanne McGowan - Fisk University
Co-Author(s): Dr. Gabriela Gonzalez, Louisiana State University, Baton Rouge, LA; Dr. Kelly Holley-Bockelmann, Vanderbilt University, Nashville, TN; Dr. Karan Jani, Vanderbilt University, Nashville, TN; Dr. Brian O’reilly, LIGO Livingston, Livingston, LA
Non-stationary scattered light creates noise in the Advanced LIGO (Laser Interferometer Gravitational-Wave Observatory) gravitational wave detectors signals, impeding on the frequency of gravitational wave detections and quality of the data collected. Scattered light produces low frequency noise (20-40 Hz) that couples into the gravitational wave sensitive frequency band (10–100 Hz) through light reflected in mirrors and other surfaces. The scatter noise caused by low frequency motion can be identified as arches in the time-frequency plane of the gravitational wave channel. Gravitational waves are the key prediction of Albert Einstein’s General Theory of Relativity. The direct detection of these waves by LIGO since 2015 has opened a new observational window into previously hidden aspects of the cosmos. These waves are produced from extreme cosmic phenomena such as mergers of black holes, neutron stars and supernovae. Accurately measuring gravitational waves has important applications to fundamental physics, providing insights into the nature of space, time and matter at cosmological scales. In the Advanced LIGO era, following a major upgrade of the interferometers, sensitivity has increased to allow for frequent gravitational-wave detections and an increased range of 150 Mpc for binary neutron star (BNS) detections. As the detector power and sensitivity increases, the need for noise characterization tools follows, especially in the low frequency range where gravitational wave signals occur. To further improve the sensitivity of LIGO we investigate scattering surfaces with the intent to reduce the scattered light effect. There are at least two different populations of scattering noise known. We investigate the multiple origins of one of them by implementing an algorithm that uses a witness channel and scattering event time identified by machine learning tool GravitySpy. The algorithm filters out noise and calculates properties of the arches, including the velocity of the arches, which are used to trace the surface creating the scattering noise. Using this algorithm, we investigate the changes in scattered light properties such as the frequency, velocity, duration and rate at which they occur during O3 when compared to O4 after an extensive upgrade period. We find the number of scattered light occurrences has decreased in O4 at the LIGO Livingston Observatory (LLO), but the shape and duration have changed, indicating a change in scattered light origins from the previous run. The algorithm was tested on O4 data as well as a simulated noise model and we find that the algorithm accurately calculates the expected velocity and frequency of the arches in over 90% of cases. The algorithm will further be implemented to investigate several witness channels associated with surfaces inside of the detector. Future research involving LIGO Hanford (LHO), a site identical to the LLO, will be conducted using this tool. Future plans for the algorithm include comparing the Fast-Fourier Transform (FFT) method to a Hilbert-Huang Transform (HHT) method.
Funder Acknowledgement(s): This study was supported, in part, by a grant from Department Of Education awarded to Arnold Burger PhD, Director for the Center for Biological Signatures and Sensing (BioSS), Fisk University, Nashville, TN.
Faculty Advisor: Dr. Karan Jani, karan.jani@vanderbilt.edu
Role: I created the scattered light arch analysis algorithm using python based data analysis techniques and the GWPY package. I perfected the filtering techniques and created a simulated arch signal with noise identical to those found in the detector's real data.