From: Kirk on
I have a series of multivariate monthly climate data that I am modeling with mvnrnd. However, the data generated from mvnrnd are temporally independent. This is not the case for the measured observations from which the means and covariances come from.

I would like to add a autoregressive model to the mvnrnd generated data. The idea is to be able to stochastically generate new data, but to model both within month covariance of the four variables, and the temporal covariance components.

Any suggestions would be appreciated.