Commonly used horizontal diffusion and wavenumber truncation dealiasing met
hods in spectral models are verified using the National Center for Atmosphe
ric Research Community Climate Model version 3. For the same horizontal gri
d resolution, time step, physical processes, boundary conditions, and initi
al conditions, the simulated climate, using the horizontal diffusion alone
model, is better than that using the wavenumber truncation method. In compa
rison with the observed climate data, the global root-mean-square of simula
ted January monthly mean 500-hPa geopotential using the horizontal diffusio
n alone model is 25% less than that using the wavenumber truncation model.
However, for the same spectral resolution, the wavenumber truncation model
(high horizontal grid resolution) leads to more accurate solutions than the
horizontal diffusion model (low horizontal grid resolution).