Understanding turbulence and correctly representing it in numerical weather, climate and air pollution models remains an outstanding problem in atmospheric sciences, adding to large uncertainties in weather prediction and climate projections. Using a large ensemble of turbulence measurement datasets obtained over a wide range of surface conditions (from flat to mountainous terrain), we will explore the nature of turbulence, how it is modified by terrain complexity, and search for frameworks that allow turbulence theories to be extended to such locations where the common assumptions are clearly violated.