Elevation problems on the boundary

Hallo there,

I am modeling a multilayer aquifer system. Attached are the picutres of the interpolated elevation 3D model. But it does not appear like what I want.  I tried different interpolation methods. but the elevations on one of the boundaries still dont reach convergence. It is a syncline structure. Only the hilly boundary could not be interpolated well. other parts of the model are OK. and generally are also OK.  I wonder why top slice 1 and 2 has no problem while other slices have shock there. And besides, this only shows up at this very starting boundary. Other boundaries are also OK.
Additionally, anyone who could help to tell the differences and applicable conditions of each interpolation methods? I think my original elevation.dat file is also not so well compiled. 
Your help is appreciated.

I'm no geologist, but that data doesn't look right to me. You should probably go back and check it... In your second image, it looks like you have all the stratigraphy equal to a thickness close to 0m at one point. You should probably get rid of outliers like that and smooth your surfaces below the topography first. Also, looks like the thicknesses of the stratigraphy vary wildly along the between the syncline and anticline, which isn't realistic...

FEFLOW only interpolates the data onto the mesh. If the data is bad, you're going to end up with a bad mesh...

Hi,adacovsk,
Thanks for your comments!
Yes, first of all, I think the original data file is not so good to be interpolated.
But I checked other cross sectional views of the model, they are however quite normally interpolated and there is no such up and down or "sharp" curves as I showed in those pictures. The thickness of each layer is definitely different in my elevation data base. And stratums are growing thicker in the direction to the center of the domain. Bad results only show up at this very boundary part. I tried to add and delete some data at the boundary but it does not work.  Anyway, thanks for your help~