Statistical test to asses models' spread

Hi Folks, I have 3D matrix (tas) containing mean global temperatures from 20 different models, so the shape of the matrix is tas(20,180,360), where 180 and 360 are the latitude and longitude points. I would compute the model agreement for each point of the domain, namely I would check if there is a large spread in models output. To do so, I was thinking to make a t-test or anova, to check if each model significantly differs from the ensemble mean. Do you know how can I do that with matlab?

Thanks, Markus

[Markus]

Just wondering if this question is not clear enough? Seems strange that I am getting no replies even though it's been posted for a while now. Best, Markus