I am analyzing time series concerning daily water temperatures in rivers and I stubmled upon a series with autocorrelation like this:

Autocorrelation function with maximum lag = 730

Moreover the series is stationary (I checked it with the Augmented Dickey-Fuller test). Partial autocorrelation plots looks like this (and I guess it indicates that it is AR(2) process):

enter image description here

So my problem is: what does this weird autocorrelation plot indicates? It seems to be telling me something, but I can not understand what it is exactly, so any help will be much appreciated.


What the autocorrelation plot indicates is that a good model for the stochastic process in question is $$X(t) = \cos(2\pi t + \Theta), -\infty < t < \infty$$ where $\Theta$ is a random variable that is uniformly distributed on $[0,2\pi)$. For a proof that the autocorrelation function of this process is indeed $\cos(2\pi \tau)$ where $\tau$ is the lag, see this answer on dsp.SE.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.