Thread Subject:

generating covarying random climate

I am working with monthly climate climate file (4 variables - tmax, tmin, radiation, precip). I have built a climate generator that uses the mvnrnd and repmat functions to produces random climate series while maintaing covariance relationships among the variables.

mu=cat(2,climate_mu(:,1),climate_mu(:,2),climate_mu(:,3),climate_mu(:,4));
sigma=1.0*climate_sigma;

climateData=mvnrnd(repmat(mu,1200,1),repmat(sigma,[1 1 1200]));

This is intended to account for the covariance among tmax and tmin for example. Which is to say... that if it is a cold month, both tmax and tmain are likely to be cold.

However, I am wondering about covariance within a variable from month to month. By which I mean... does tmax of January, covary with the tmax of February? I suspect it does. In fact the entire series Jan - Dec) probably follows some kind of hidden sin wave.

Does anyone have a suggestion on how to investigate this? and/or how to solve the issue if these variables do co-vary temporally.