ABL simulation with experimental data: problems and doubts
 

Hi all,
i'm trying to do a "validation" test of an Atmospheric Boundary Layer experiment. The experimental data are found in this paper:

"Near wall characterization of the flow over a two-dimensional steep smooth hill", J. B. R. Loureiro Æ F. T. Pinho Æ A. P. Silva Freire

Actually i'm working on a flat surface, so i can verify if the velocity profile will be modified by the solver.
I have extracted the wind and turbulence profile points, and i have used this data as inlet condition in the setup.

I have had some problems with the roughness length: the paper report a value of 0.27mm, but if i use this value in the atmBoundaryLayerInletVelocity condition the velocity profile will be too modified.
I have used the velocity formula found in atmBoundaryLayerInletVelocity source code to find the value of z0 that best approximate the inlet velocity profile. This value is z0 = 5e-6. Yuo can see the approximation in the first image in the attachment.

With the turbulence profile of the paper, i've calculated the inlet profile for k, epsilon, using the formula i've found in cavity tutorial (se formula 2.8, 2.9).
In K,epsilon, i've inserted a list of values calculated at the center of a given inlet face of the mesh. Is that right?
In the internalField value for k, i've used the mean value of the face center k values.
In the internalField value for epsilon, i did not used the mean value of the face center epsilon values, because it was a really small one, and the solver crashes (floating point exception). So i've multiplied by 100 that value and solver was able to converge.

You can see the result in the second image in the attachment. What are the causes of the differences in the velocity profile?

After all this explanation, my doubts are:

1. Is there a better way to approximate the experimental inlet profile? In the first pic it is possible to see "big" variations from the experimental one and the approximation i've used.
2. Is it correct to use the formulas 2.8-2.9 in cavity tutorial to calculate k, epsilon inlet profiles?
3. How can i calculate the internalField value for k, epsilon?
4. In paraFoam, if i plot k and epsilon at inlet and outlet, the profiles are extremely different. Also, the inlet profile seems wrong in comparison to the data i've given in the input files.
5. What can i do to improve the accuracy of the result?

PS: in this link you can find the simulation.