The NANOGrav Nine-Year Data Set: Excess Noise in Millisecond Pulsar Arrival Times

TL;DR

This paper analyzes excess noise in millisecond pulsar timing residuals from the NANOGrav nine-year data set, identifying red and chromatic noise components and their implications for gravitational wave detection.

Contribution

It characterizes and decomposes excess noise in pulsar timing data, revealing red and chromatic components, and proposes a scaling law relating noise to pulsar spin parameters and observation span.

Findings

26 out of 37 pulsars show inconsistencies with white-noise-only models.

15 pulsars exhibit red power spectrum in excess noise.

The excess noise scales with pulsar spin parameters and data span.

Abstract

Gravitational wave astronomy using a pulsar timing array requires high-quality millisecond pulsars, correctable interstellar propagation delays, and high-precision measurements of pulse times of arrival. Here we identify noise in timing residuals that exceeds that predicted for arrival time estimation for millisecond pulsars observed by the North American Nanohertz Observatory for Gravitational Waves. We characterize the excess noise using variance and structure function analyses. We find that 26 out of 37 pulsars show inconsistencies with a white-noise-only model based on the short timescale analysis of each pulsar and we demonstrate that the excess noise has a red power spectrum for 15 pulsars. We also decompose the excess noise into chromatic (radio-frequency-dependent) and achromatic components. Associating the achromatic red-noise component with spin noise and including additional power-spectrum-based estimates from the literature, we estimate a scaling law in terms of spin parameters (frequency and frequency derivative) and data-span length and compare it to the scaling law of Shannon \& Cordes (2010). We briefly discuss our results in terms of detection of gravitational waves at nanohertz frequencies.