Modelling streamflow from forested watersheds on the Canadian boreal shield using the Soil and Water Assessment Tool (SWAT)

Abstract

The Forest Watershed and Riparian Disturbance (FORWARD) Project was initiated in 2001 in Canada. The main objective of the project is to investigate the effects of tree harvesting upon streamflow and nutrient export from forested watersheds, and to develop hydrologic and water quality modelling tools to predict these effects. For this purpose, the FORWARD Project has been adapting the Soil and Water Assessment Tool (SWAT) model for boreal forest watersheds. The SWAT model was originally developed for the management of agricultural watersheds in the USA. Therefore, the FORWARD project researchers modified the SWAT model to make it more suitable for simulating hydrological processes occurring in boreal forest watersheds. The modified model is called SWATBF.

SWATBF was successfully tested on the western Boreal Plain where the soil mantel is thick. However, the model must be tested before applying it on eastern Boreal Shield watersheds where the soil layer is thin. Therefore, the focus of this study was to investigate the applicability of SWATBF for Boreal Shield watersheds and to investigate differences in calibration parameters and their values, and model set-up for hydrological simulation between Boreal Shield and Boreal Plain watersheds.

Initial set-up and testing of the model showed that a simplified version of SWATBF provided acceptable performance in Boreal Shield watersheds. Hence the simplified version of the SWATBF model was used in this modelling investigation. The simplified SWATBF omitted the Boreal Plain litter layer and wetland representations.

Two types of tests were conducted to verify the applicability of the SWATBF model: (1) split-sample test; and (2) proxy-basin test. In total, six case studies were performed from the two different tests. In general the simplified SWATBF model was able to predict the pattern of monthly and daily streamflow in all six case studies. The performance of the model was much better for simulation of monthly runoff compared to daily runoff. However, it was found that the model underestimated the many of the daily peak flows in all case studies. Potential sources of model error are discussed.