On central-difference and upwind schemes