1: \begin{abstract}

2: 		A flexible model for non-stationary Gaussian random fields on hypersurfaces is introduced.

3: 		The class of random fields on curves and surfaces is characterized by a	power spectral density of a second order elliptic differential operator.

4: 		Sampling is done by a Galerkin--Chebyshev approximation based on the surface finite element method and Chebyshev polynomials.

5: 		Strong error bounds are shown with convergence rates depending on the smoothness of the approximated random field.

6: 	\end{abstract}

7: