2022RAA....22j5012Y

An ensemble pulsar timescale derived from the traditional Wiener filtration still contains some high level noise. To improve this situation we developed a Wiener filtration algorithm of the ensemble pulsar timescale based on a power-law model of power spectrum for pulsars. Our algorithm has three strengths: (1) mitigating spectral leakage between frequency bins for Fourier techniques; (2) using a power-law model for power spectrum; (3) signal realization in original residuals of data set by the power-law model. According to improved algorithm we constructed an ensemble pulsar timescale EPT-TAI using timing data with respect to International Atomic Time (TAI) about 16 yr time span of ten pulsars from International Pulsar Timing Array second data release (version A). The results show that EPT-TAI detected correctly the differences TT(BIPM2015)-TAI between terrestrial time TT(BIPM2015) and TAI. Fractional frequency stability σ_z analysis shows that EPT-TAI does not indicate red noise for 16 yr time interval, and fractional frequency stability for 8 yr and longer time intervals is slightly better than that of TT(BIPM2015)-TAI. Stability for short time intervals of TT(BIPM2015)-TAI is better than that of EPT-TAI, but TT(BIPM2015)-TAI shows red noise for longer time intervals. Using the same algorithm we also derived an ensemble pulsar timescale EPT-TT(BIPM2015) with respect to TT(BIPM2015). The fractional frequency stability curve of EPT-TT(BIPM2015) shows similar characteristics as that of EPT-TAI but with slightly lower values.