A Physical Scaling Method to Incorporate Regional Physical Characteristics in Future Climatic and Hydrologic Predictions

In this thesis an original Physical Scaling (SP) method for downscaling Global Circulation Model (GCM) based climatic projections has been developed, tested and applied over a study region. The model formulation can take into account regional physical characteristics like land-cover and elevation into the model formulation. A thorough verification of the method and its extension: SP with Surrounding pixel information (SPS) method has been performed and their performance towards downscaling GCM based precipitation, surface temperature and air temperature has been compared with many state-of-the-art downscaling models like Bias Correction Spatial Downscaling (BCSD) method, Statistical DownScaling Method (SDSM) and Generalized Linear Modeling (GLM). The SPS method extends SP method by also taking into account neighborhood physical characteristics into the downscaling process. A major benefit of the presented downscaling approaches is that they can account for non-stationarity in physical characteristics of the region of interest like changes in land-cover as well as their neighborhoods. This represents a major contribution in the field of statistical downscaling literature since it brings the benefits of physically based dynamic downscaling into a statistical downscaling framework. Proposed models are used to isolate physically sourced climatic and hydrologic contributions in four catchments located within the southern Saskatchewan region of Canada. Contributions towards flood magnitudes are also studied for low to high return period flooding events. Results indicate that the contributions of catchment physical characteristics towards shaping climatic and hydrologic regimes in the analyzed catchments are statistically significant. Further significant variability in the detected changes exists over catchment space and analyzed time-period. Finally the results from this thesis highlight the importance of further exploration of physically driven climatic changes, and the need to find out how to incorporate them while making future streamflow predictions. The developed SP and SPS methods are highly relevant and useful in a non-stationary world which is set to experience rapid climatic and geophysical changes in the future.