Evaluate the skill scores uncertainty

[ocefpaf]

For Taylor diagrams we could use a subset of independent observation to evaluate how the model performs to the same metric. For the Mean Bias and RMSE a bootstrapping could do the trick.