Frequency-dependent stochastic radio time-scales in OT 081 from UMRAO monitoring

Abstract Using multi-decade monitoring by the University of Michigan Radio Astronomy Observatory (UMRAO), we test whether the dominant stochastic radio memory time-scale of OT 081 (PKS 1749+096; z = 0.322) changes between 4.8, 8.0, and 14.5 GHz. Each log-flux-density light curve is modelled with a continuous-time autoregressive moving-average (CARMA) process. Model adequacy is assessed with residual-whitening diagnostics and calibrated posterior predictive checks (PPCs), which compare observed variability summaries with the same summaries measured from light curves simulated from the fitted model. The quantity pcal is the calibrated predictive probability for the adopted discrepancy statistic; values near 0.5 indicate that the observed statistic is typical of the simulated reference distribution. The adopted models show no clear adequacy problem at 8.0 and 14.5 GHz, whereas 4.8 GHz is borderline in the global time-domain PPC (pcal = 0.0430), with the mismatch localised to 20–50 d lags. We use the longest CARMA damping time-scale, τlong, as the primary memory scale. The fractions of approximate-posterior draws in which τlong is shorter at 14.5 GHz than at 4.8 and 8.0 GHz are 0.995 and 0.943, respectively. Although the first fraction is numerically larger, the 14.5-versus-8.0 GHz comparison is diagnostically cleanest because both bands satisfy the adopted adequacy checks; the 4.8 GHz result points in the same direction but is interpreted cautiously because of its localised PPC warning. Centroid lags from the z-transformed discrete correlation function (ZDCF), defined as positive when lower-frequency variations lag higher-frequency variations, are mostly positive but broad. These results provide an observational reference for comparing radio stochastic time-scale hierarchies with VLBI core-shift constraints in blazar jets.