| Title: | Simulated impacts of climate change and agricultural land use change on surface water quality with and without adaptation management strategies |
| Authors: | Mehdi, B., B. Lehner, C. Gombault, A. Michaud, I. Beaudin, M.-F. Sottile and A. Blondlot |
| Year: | 2015 |
| Journal: | Agriculture Ecosystems and Environment |
| Volume (Issue): | 213 |
| Pages: | 47-60 |
| Article ID: | |
| DOI: | 10.1016/j.agee.2015.07.019 |
| URL (non-DOI journals): | |
| Model: | SWAT |
| Broad Application Category: | hydrologic and pollutant |
| Primary Application Category: | climate change and land use change |
| Secondary Application Category: | BMP and/or cropping system assessment |
| Watershed Description: | 629 km^2 Pike River located at the northern tip of Lake Champlain, straddles the Province of Québec (530 km^2) and
the State of Vermont (99 km^2); see Fig. 1 in paper. Pike River is a subwatershed of the Missisquoi Bay, and was identified in 2002 by the Québec government as a priority watershed in need of integrated water resources management. The river is plagued with inputs of excessive nutrients from non-point source pollution stemming from agricultural and forest activities. In particular, the total = phosphorus (TP) in the Bay exceeds the Québec guidelines on surface water quality of 0.02 mg/L (MDDEFP, 2002), as well as the TP limit set for the Missisquoi Bay of 0.025 mg/L (LCBP, 2013).
Excess P contributes to regular cyanobacteria algae blooms (Blais, 2002; Simoneau, 2007) that have appeared almost annually in the Bay since 2000. |
| Calibration Summary: | Overall, the calibration and validation of SWAT were deemed
“satisfactory” according to the objective criteria listed in Moriasi
et al. (2007) and fell within the ranges of other SWAT studies
listed in Gassman et al. (2007) and Douglas-Mankin et al. (2010).
We confirm the common tendency of SWAT to perform less well
at the daily time step as compared to monthly or yearly steps
(Gassman et al., 2007). Therefore, we used monthly results of
SWAT in this study, so we focus on the performance evaluation
at this time step. The modelled transport of nutrient loads is
highly dependent on properly simulated streamflow and
snowmelt processes; our SWAT simulations for streamflow
had Nash-Sutcliffe Efficiency (NSE) values of 0.75–0.83
(Table S4a). The simulation of nutrients was satisfactory during
the calibration stage; except the sediment values were under-
estimated compared to observed values. The validation of
nutrients, however, had a rather poor model performance;
although this is not unusual (Gupta et al., 2009). The discrep-
ancies in our case were attributed to the short time periods of
available data as well as the weaker simulation of streamflow at
the outlets where the nutrients were calibrated (outlets 4 and 6).
The performance criteria for sediments and for TP were
somewhat better for the independent evaluation (1979–2007)
at outlet 18. |
| Validation Summary: | |
| General Comments: | |
| Language: | English |
| Keywords: | SWAT
Nitrate
Phosphorus
Streamflow
Scenarios
Management practices |