Improving pulsar timing by detecting and correcting for pulse shape changes

Pulsar timing derives its strength as an observational technique from the close connection between the rotation of a pulsar and the measurable times of arrival (TOAs) of the radio pulses it emits. However, this connection is not perfect. TOAs are typically measured by comparing an intensity profile created by averaging many observed pulses to a long-term average profile (template). This means that changes to the shape of the observed profile can affect the TOA estimate. Such pulse shape changes occur for a number of reasons, including the intrinsic variability of the pulsar emission, nulling and mode changing, and changes in the amount of dispersion or scattering by the interstellar medium.

I will present techniques I have developed for quantifying the errors in time-of-arrival estimates associated with pulse shape changes, and extensions to the standard estimation algorithms which may be used to correct these errors. In addition, I will describe observations of a recent and dramatic example of a pulse shape change, seen in one of the millisecond pulsars most frequently observed by pulsar timing arrays (PTAs), PSR J1713+0747, in April 2021. This new event follows two much smaller events, which may be of a similar nature, seen in the past 15 years. This event and others like it, if they cannot be successfully mitigated, may make it significantly more challenging for PTAs to detect and characterize the low-frequency stochastic gravitational wave background.