Channel LES and explicit filtering
 

I am trying some calculation with LES dynamic model in a 3D channel. Many authors use top hat filtering function to perform the explicit filtering operations. The convolution integral is evaluated by means of linear interpolation of the function to be filtered and dicrete integration rule (Zhang). In a laminar flow the right hand term in Germano's identity (i.e. the difference between the corresponding components of the coarse grid stress tensor and the filtered fine grid ones) should be zero as states Ghosal. But in a laminar channel flow the wall to wall velocity profile is parabolic and in my calculation the trace of this term shows a maximum on the first point near the wall. Is the problem due to a too coarse grid resolution (32x16x32 and same geometry of Kim and Moin DNS) or to the filtering operation? Thank in advance for any help.