Located at the downstream reaches of the Nelson River Basin, the province of Manitoba is rich in water resources that support the development of hydroelectric projects. Accurate estimates of water yield are necessary to support hydropower planning studies, dam safety studies, and hydroelectric operations. The remoteness and expanse of the region hinders the implementation of additional streamflow gauges, hence deterministic hydrological modelling offers an alternative. The use of a hydrological model is beneficial in ungauged areas where calibration in one area permits translation of the model to other ungauged areas with similar land cover.

The lower Nelson River is located in the High Boreal Forest of northem Manitoba. The region contains complex hydrological processes that shape the streamflow hydro graph. Some of these processes include discontinuous permafrost, unique evapotranspiration (ET) processes, a sub-arctic climate, a landscape that is littered with numerous lakes, marshes, and swamps that may be hydraulically disconnected from the basin due to its hummocky terrain, and land cover consisting of various types of muskeg, forest, and treed rock. Snowmelt processes dominate the hydrograph resulting in peak annual streamflow during the spring freshet and near base flow conditions during the winter.

The main objective of this research is to apply a macro scale deterministic hydrologic model to the High Boreal Forest region of northern Manitoba that is capable of providing accurate water yield estimates. Model parameters for the dominant types of land cover are derived from the literature to initiate model calibration.

The SLURP hydrologic model is a semi-distributed conceptual macro scale model that is proven to work well under a wide variety of basin conditions. The SLURP model was applied to the Taylor River (899 km²) and the Upper Burntwood River (6859 km²). Geographic Information Systems were utilized to develop the physiographic parameters for each of the basin Aggregation Simulation Areas.

The SLURP model was calibrated and verified on the Taylor River for each year between 1985 and 2000. A relatively small number of parameters were adjusted in order to optimize the modelling performance. Overall, SLURP performed satisfactorily as iilustrated by an average annual Nash Sutcliff Efficiency (NSE) of 61% and Deviation from Runoff Volume (D_v) of ±12%. The same model parameters were applied to the Upper Burntwood River with the exception of elevated ET parameters. Using the same test years in the Taylor River application, the resulting NSE = 54% and D_v=±14%

A number of model deficiencies were determined during the application of the SLURP model to the test basins. A simplifiedfrozen soils model was tested in the SLURP model but it was abandoned since it resulted in a significant amount of overland runoff during the spring which did not sustain the streamflow hydrograph. The date dependent snowmelt rates were replaced with a single constant snowmelt rate. The premature generation of snowmelt in the simulated hydrographs was addressed by incorporafing a snowpack temperature deficit model into SLIIRP in order to simulate the effects of snow ripening. These two modifications provide the modelling flexibility needed to control the timing of initial snowmelt and the rate of snowmelt. The model performance improved during the spring freshet period for nearly all test years in both basins.

This research has demonstrated that the SL{JRP model is able to provide reasonable estimates of water yield for the Taylor River and Upper Burntwood River. Application of the model to other basins and additional model ref,rnements would continue to improve the model performance and would develop a better understanding of the hydrological processes of the High Boreal Forest.