A modeling study is undertaken to better understand the physics of katabatic flows. This study is divided into three topics; a comparison between a large eddy simulation (LES) and a mesoscale model of katabatic flows, a sensitivity study of katabatic flows to various physical parameters, and an investigation into the effect of subgrid scale terrain features on katabatic flow models. In the first topic, a comparison between LES, and a mesoscale model, ARPS, of katabatic flows is made to better quantify the accuracy of subgrid parametenzation in ARPS. It is shown that, although the modeled flows agree on a number of parameters, the LES model produces a lower and faster jet than that of ARPS, and also cools more near the surface. The momentum budgets of the two models agree well with each other. The ARPS model has a higher amount of TKE than the LES model, due to an overproduction by shear in the ARPS subgrid parameterizations. The second portion of this thesis represents a sensitivity study of katabatic flows to various physical parameters. The depth and strength of katabatic flows are shown to vary with surface heat fluxes, slope angle, and ambient stratification. Katabatic flows are shown to grow in depth and magnitude as slope angle increases, due to an increase in entrainment of overlying ambient air. The ratio of advection to mixing is shown to collapse to a near universal value regardless of surface heat fluxes. With increasing ambient stratification, entrainment in katabatic flows becomes small and the momentum equation is reduced to a two-way balance between buoyancy and drag. In this case, the heat flux of entrained air into the katabatic flow approaches that of the surface cooling, and the flow ceases to grow in the down-slope direction. Finally, predictions for bulk velocity and buoyancy strength scales are developed as a function of slope angle and surface heat fluxes. The last portion of this study focuses on the effect of subgrid scale terrain features on katabatic flows. It is shown that in areas of inadequate terrain resolution, the effect of the terrain smoothing routine in ARPS is to increase the slope height in areas of concave mountains. The concept of energy conversion in katabatic flows is introduced, and it is shown that the effect of raising terrain is to assign parcels more buoyant potential energy than they would otherwise have, and thus over-predict the magnitude of katabatic flows. Finally, an investigation into the effect of changing upper slope angle on katabatic flows over combined slopes is made. It is concluded that a combined slope cannot be predicted using a linear combination of simple slopes, since the transition portion of the slope results in a turbulent hydraulic jump with enhanced mixing. The magnitude of mixing in the turbulent hydraulic jump in combined slopes is shown to depend on the difference between upper and lower slope angle. / Graduation date: 2004