Empirical verification of the truncation error (2)

• Assume $R^n = C\Delta t^r$
• $C$ and $r$ will vary with $n$ - must estimate $r$ for each mesh point
• Use a sequence of meshes with $N_t = 2^{-k}N_0$ intervals, $k=1,2,\ldots$
• Transform $R^n$ data to the coarsest mesh and estimate $r$ for each coarse mesh point