Nehrkorn, Thomas, Atmospheric and Environmental
Research Inc., and Thomas AulignŽ, NCAR
Numerical Weather Prediction
models have dramatically improve in recent years, but they still often exhibit
significant phase errors. These errors are particularly difficult to correct in
data assimilation because they are usually not normally distributed. However,
they can have a major impact of the forecast of important meteorological
features such as cloud systems and hurricanes. We will present a package to
compute alignment fields based on observations and WRF model background.
Two-dimensional alignment is computed through a non-linear minimization
process, with a series of constraints applied on the alignment to insure a
physical result. A particular focus is devoted to the conservation of the
internal balance in the model fields. Results for a hurricane case and cloud-resolving
simulations will be shown, with a significant impact on the forecast skill.