Numerical errors associated with the method of superposition for computing acoustic fields