Spatial scale and hydrological model response
Lacroix, Martin P.
This thesis focuses on the effects of varying spatial scale on the numerical simulation of hydrological processes. Specifically, it examines the impact of changing the number and sizes of sub-basins on the performance of the SLURP hydrological model. Automated watershed segmentation and parameterization was essential for this scaling analysis. The TOPAZ digital landscape analysis model was used to process a raster digital elevation model (DEM) to derive a wide range of topographic and topologic variables that are physically meaningful to watershed runoff processes. TOPAZ automatically subdivides a main watershed into a variable number of sub-basin units at specified levels of detail or scale. The SLURPAZ interface was used to manipulate relevant output data from TOPAZ along with land cover, routing and climate station data to provide all the necessary inputs for the SLURP hydrological model. Results of an application using these techniques is presented for a northern alpine basin (i.e. Wolf Creek, Yukon, Canada). A comparative analysis is presented from an earlier hydrological simulation, for which the physiographic parameters were derived manually using a geographical information system (GIS), with a hydrological simulation of the same scale, where the physiographic parameters were derived automatically using the digital landscape analysis model, TOPAZ, and the SLURPAZ interface. Generally, simulations from the hydrological model, SLURP, are very similar when comparing the automated and manual runs. The major benefits of using the automated segmentation parameterization techniques versus the manual ones are the cost and time factors, and that the automated methods offer rapid parameterization at several scales or levels of detail. Several model runs, varying in scale from 1 to approximately 1600 sub-basins, are discussed. Varying sub-basin scale showed interesting trends at two levels of detail. First, as the number of Aggregated Simulation Areas (ASAs) increases, it is necessary to have a sufficient number of ASAs to allow variations in water balance components to be adequately represented. Secondly, results suggest that with an increasing number of sub-basins, accuracy of the hydrological model outputs tend to increase up to a certain threshold. Thereafter, any further subdivision does not enhance model performance significantly. Another comparison ensued where a short channel versus a long channel, both for the whole basin as one ASA, in order to evaluate SLURP's 'within-ASA' routing sensitivity. Additionally, the 'between-ASA' storage routing method was evaluated by assessing the sensitivity of the Alpha and Beta routing parameters. Optimal scales found in previous research from varying the size and number of subbasins are also compared to those found in this research. Automated parameterization allows for scaling studies to take place in a more rapid and cost effective manner. SLURPAZ acts as a useful link between the SLURP hydrological model and the TOPAZ digital landscape analysis model. Such tools offer the benefits of speed, precision and reproducibility in drainage basin analyses.