Least-squares estimation and ANOVA for periodic autoregressive time series

The periodic correlation exists throughout the whole process in a analysis of variance (ANOVA) type model where the error terms consist of a periodic autoregressive time series. This paper studies the asymptotic property of least-squares estimators and linear testable hypotheses with a modified F-test in the analysis of variance akin to periodic autoregressive series. The techniques are applied in making inference on the quarterly streamflow in Asotin, WA.