inconsistency of sign or eigenvectors

I am trying to solve the transmission and reflection of a 2D photonic crystal.
Attached with the document where I only looks at single layer with only one fourier expansion. As can be seen on the last figure, there seems to be discontinuity and rapid changes on the graph. I have traced back to be the results of inconsistency of sign of the eigenvectors. Modelling for than single layer creates random spikes. Is there any way to fix such inconsistency?