How can I rescale normalized standard deviation in scikit-learn's Gaussian process regression?

Question

I am using scikit-learn for modeling with Gaussian Process Regression. My data are not normalized. The model always return standard deviation values between 0 and 1, which is no