Many landscape evolution models have considered the interaction of exogenic and endogenic processes. However, geomorphological processes have not been successfully incorporated in landscape evolution models. The thesis begins with a critical analysis of methodologies for the study of large-scale geomorphological processes. A framework based on a generalization of the relevant processes is recommended. Hillslope and channel submodels, which are based on typical processes operating in coastal regions of British Columbia, are introduced. The following hillslope processes are considered: (i) slow, quasi-continuous mass movements; (ii) fast, episodic mass movements; and (iii) weathering. The transport relation for fast, episodic mass movements was found to be nonlinear. Fluvial transport in both low and high-gradient channels and debris flow transport are considered in the channel submodel. A bed load transport equation, which is a revised version of the Bagnold stream power formula, is derived. Suspended load is calculated using a suspended load/contributing area correlation. Connections between hillslope and channel processes are considered to ensure adequate representation in the model. The hillslope and channel submodels are explored in one-dimensional and surface model runs for small drainage basins in the Queen Charlotte Islands, British Columbia. Tests of the fluvial submodel demonstrate the robustness of the bed load equation used in this study. A conceptualization of the landscape into unstable and stable regimes is introduced. Results of surface model runs emphasize the key role of low-order channels in transferring sediment from hillslopes to main channels. The exercise of constructing and running the model highlighted major gaps in our present understanding of geomorphological process operation and sediment routing. Suggestions for future research are extensive and are outlined in the concluding chapter of the thesis.

Many landscape evolution models have considered the interaction of exogenic and endogenic processes. However, geomorphological processes have not been successfully incorporated in landscape evolution models. The thesis begins with a critical analysis of methodologies for the study of large-scale geomorphological processes. A framework based on a generalization of the relevant processes is recommended. Hillslope and channel submodels, which are based on typical processes operating in coastal regions of British Columbia, are introduced. The following hillslope processes are considered: (i) slow, quasi-continuous mass movements; (ii) fast, episodic mass movements; and (iii) weathering. The transport relation for fast, episodic mass movements was found to be nonlinear. Fluvial transport in both low and high-gradient channels and debris flow transport are considered in the channel submodel. A bed load transport equation, which is a revised version of the Bagnold stream power formula, is derived. Suspended load is calculated using a suspended load/contributing area correlation. Connections between hillslope and channel processes are considered to ensure adequate representation in the model. The hillslope and channel submodels are explored in one-dimensional and surface model runs for small drainage basins in the Queen Charlotte Islands, British Columbia. Tests of the fluvial submodel demonstrate the robustness of the bed load equation used in this study. A conceptualization of the landscape into unstable and stable regimes is introduced. Results of surface model runs emphasize the key role of low-order channels in transferring sediment from hillslopes to main channels. The exercise of constructing and running the model highlighted major gaps in our present understanding of geomorphological process operation and sediment routing. Suggestions for future research are extensive and are outlined in the concluding chapter of the thesis. / Arts, Faculty of / Geography, Department of / Graduate

The main focus of the present study is to identify and integrate the factors affecting the degree of irregularity of five continental divide traces, as expressed by their fractal characteristics measured by the divider method. The factors studied are climate, relief and tectonic environment. The second objective of this study is to determine the relationship between uplift rates and divide trace fractal dimension. Analysis of the results suggests that the degree of irregularity of continental divide traces at fine scale (approximately 10-70 km of resolution) is strongly affected by both climate and tectonics. It is found that control of the factors is generally weaker at coarse scale (above approximately 70 km of resolution). Generic relief should be ranked below both climate and tectonic environment as a factor affecting the complexity of continental divide traces. In terms of the second objective, the fractal dimension at fine scales follows a weakly inverse relationship with uplift. At coarse scale, there is stronger inverse relationship between uplift rate and fractal dimension. / Introduction -- Methodology -- Geomorphic environment -- Evaluation of results -- Significance of control factors -- Conclusion. / Department of Geological Sciences

Previous works have applied the Divider Method to the shapes of drainage divides as measured from maps. This study focuses on the shapes of several drainage divides measured in the field at very fine scale. These divides, chosen for their sharp crests, include portions of the Continental Divide in Colorado and badlands-type divides in Arizona, Wyoming, South Dakota, and Texas. The badlands type divides were delineated using a laser theodolite to collect data at decimeter point spacing, and the Continental Divide segments were delineated using pace and bearing at a constant point spacing of 30 meters. A GIS was used to store and visualize the divide data, and an automated divider analysis was performed for each of the 16 drainage divides.The Richardson plots produced for each of the drainage divide datasets were visually inspected for portions of linearity. Fractal dimensions (D) were calculated using linear regression techniques for each of the linear segments identified in the Richardson plots. Six of the plots exhibited two distinct segments of linearity, nine plots exhibited one segment, and one plot exhibited no segments of linearity. Residual analyses of the trend lines show that about half of the Richardson plot segments used to calculate D exhibit slight curvature. While these segments are not strictly linear, linear models and associated D values may still serve well as approximations to describe degree of divide wandering.Most (20 out of 21) of the dimensions derived from the Richardson plots for the drainage divides fall within the range from 1.01-1.07. The D values calculated for the Continental Divide range from 1.02-1.07. The dimensions calculated for the badlandtype divides were distributed evenly across the range of 1.01-1.06, with a single exceptional D value at 1.12. Only four of the divide D values fall within a range of 1.06–1.12, the range for D established for drainage divides in published map-based studies, despite the apparent dominance of erosion processes on the measured divides. / Department of Geology

Empirical observations have established connections between river network geometry and various hydrophysical quantities of interest. Since rivers can be decomposed into basic components known as links, one would like to understand the physical processes at work in link formation and maintenance. The author develops a natural stochastic geometric model for this problem, for the particular type of link known as exterior links. In the model, the distribution of distance from a uniformly distributed point to a fixed graph is computed. This model yields an approximate expression for the distribution of length of exterior links that incorporates junction angles and drainage density, and compares favorably with observed length distributions. The author goes on to investigate related mathematical questions of independent interest, such as the case where the previously mentioned graph is itself a realization of a random process, and in so doing derives a formula for the first contact distribution of a general random Voronoi tesselation (also associated with the names of Dirichlet and Thiessen). Since this random tesselation is a natural starting point for modelling spatial processes in a wide variety of fields, these results should find immediate applications. It is also shown how these results can be interpreted as a generalization of a classical problem considered by Buffon. / Graduation date: 1991