For equally spaced discretization, let h=step-size. For not equally spaced let h=max{h1,h2,h3}. hi is step-size in ith direction. Then find an operator which produces the error between numerical and exact as O(h^n), then the operator is nth order accurate. For example, central difference scheme is second-order accurate
Thank you prof. for your very good suggestion. Please suggest some journals or books for how to find the operator in case of unequal space data. Thank you very much.