Mountain weather has a remarkable range of phenomena, which are signified by two flow types: modifications to the background (∼100 km, synoptic-scale) flow driven by regional pressure gradients and thermal circulations (valley/slope flows, mesoscale) generated by local heating/cooling of the ground surface. Current state-of-the-art mesoscale models typically employ a grid resolution of ∼1 km, with the assumption that over such scales there is sufficient spatial homogeneity to represent heat, mass and momentum transport within a grid cell. Nevertheless, because of topographic heterogeneities, differential heating/cooling and flow interactions, the flow within a mesoscale model grid element can be highly inhomogeneous, and this exemplifies the difficulty of mathematically representing (parameterizing) sub-grid processes of a slope-scale grid.  The aim of the MOUNTAIN TERRAIN ATMOSPHERIC MODELING AND OBSERVATIONS (MATERHORN) PROGRAM was to conduct fundamental research to improve weather predictions for mountainous terrain, and improvements in understanding of sub-grid scale processes and their parameterizations were a significant part of it. To this end, field and laboratory studies were conducted and theoretical formulations were developed, which are described in this thesis. A suite of flow diagnostic techniques were used. At the core of the work are the subtopics of upslope flow separation in mountainous terrain, multi-scale interactions of slope and valley flows and measurement of turbulence in katabatic flows. Owing to the vast range of scales involved, new cutting edge techniques were developed and deployed for process identification. These include tower mounted three-dimensional hot-film combo probes, consisting of sonic anemometers co-located with hot-film anemometers. The combo probes follow mean winds using a feedback control loop and use a Neural Network to calibrate the hot-films in-situ. Once calibrated, these probes can measure from mesoscale flow down to the Kolmogorov scale. Also deployed were three scanning Doppler LiDARs in coordination to visualize the velocity structure and to obtain three-dimensional velocity virtual towers up to 300 m AGL. Turbulence in slope and valley flows, upslope flow separation, flow collisions and interactions between different types of flow are discussed in this thesis, with particular emphasis on quantitative results of consequence for numerical modeling.