Evaluation of four hydrological models for operational flood forecasting in a Canadian Prairie watershed

Figures & data

Figure 1. The Shellmouth basin, Canada.

Figure 2. Combined land cover for the Shellmouth basin.

Figure 3. Digital elevation model (DEM) of the Shellmouth basin.

Figure 4. Sub-basins of Shellmouth basin, and hydrometric and meteorological stations.

Table 1. Calibrated parameters for the WATFLOOD model. Ass.: Assinboine; Kam.: Kamsak; Sturg.: Sturgis (see Fig. 4 for locations).

Routing parameters	River name
	Stony Creek	Shell River	Yorkton Creek	Ass. at Kam.	Lilian River	Ass. at Sturg.	White-sand
Lower zone coefficient	0.0001	0.0031	0.0050	0.0009	0.0023	0.0047	0.0049
Lower zone exponent	0.86	1.48	0.72	0.33	2.47	1.11	0.62
Overbank Manning’s n	0.4	0.4	0.7	0.4	0.4	0.4	0.4
Channel Manning’s n	0.074	0.056	0.075	0.070	0.039	0.014	0.067
Wetland or bank porosity	0.8	1	0.9	0.9	0.2	0.4	0.7
Hydrological and snow parameters	Land-cover type
	Crop	Grass	Deciduous	Coniferous	Mixed	Wetland	Water
Interflow coefficient	0.53	1.93	9.67	4.85	8.87	3.32	0.10
Infiltration coefficient, bare ground	35.7	35.4	40.6	31.3	41.6	303.0	0.0
Infiltration coefficient, snow covered	0.11	0.25	0.02	0.48	0.04	0.27	0.00
Upper zone retention (mm)	97.0	97.7	40.9	14.3	144	84.9	144
Recharge coefficient, bare ground	0.16	0.29	0.01	0.24	0.07	0.08	0.07
Melt factor (mm °C^−1 h^−1)	0.11	0.08	0.08	0.16	0.11	0.10	0.10
Base temperature (°C)	0	−1.2	−3.0	−3.2	−0.03	−3.1	−0.5
Sublimation rate (mm d^−1)	1.27	1.47	0.37	0.24	0.08	0.25	0.16

Table 2. Calibrated parameters for the HBV-EC model.

Parameter	Definition	Calibrated value
ForestTFRAIN	Fraction of rainfall reaching ground surface below forest	0.8
ForestTFSNOW	Fraction of snowfall reaching ground surface below forest	0.8
SnowAM	Factor controlling influence of aspect on melt factor	0
SnowTM (°C)	Threshold temperature for snowmelt	0
SnowCMIN (mm °C^−1 d^−1)	Value of melt factor on winter solstice for open areas	0
SnowDC (mm °C^−1 d^−1)	Increase in melt factor between winter and summer solstice	2
SnowMRF	Ratio between melt factor in forest and open areas	0.7
SnowCRFR (mm °C^−1 d^−1)	Factor that controls rate at which liquid water refreezes in snowpack	2
SnowWHC	Liquid-water-holding capacity of snowpack, expressed as fraction of snowpack water equivalent	0.05
SnowLWR (mm)	Maximum amount of liquid water that can be retained by snowpack	2500
SoilFC (mm)	Field capacity of soil	70
RunoffFRAC	Fraction of runoff directed to fast reservoir	0.1
RunoffKF	Reservoir coefficient	0.2
RunoffAlpha	Fast reservoir exponent	0.1
RunoffKS	Slow reservoir coefficient	0.05

Table 3. Calibrated parameters for the HEC-HMS model for different sub-basins. PX temperature is used to differentiate between precipitation falling as rain or snow.

	Sub-basin
	Shell River	Local 2	Local 3	Local 1	Assiniboine at Sturgis	Stony Creek	Whitesand River
Crop coefficient	1.5	1.3	1.3	1.2	1.02	1.1	1.4
Canopy max. storage (mm)	20	10	25	15	10	15	20
Max. infiltration (mm h^−1)	53	53	53	53	53	53	53
Base temperature (°C)	–1.5	–1.5	–1.5	–1.5	–1.5	–1.5	–1.5
PX temperature (°C)	1	1	1	1	1	1	1
Wet melt rate (mm °C^−1 d^−1)	4	4	4	4	4	4	4
Soil percolation (mm h^−1)	15	15	15	15	15	15	15
Lapse rate (°C 1000 m^−1)	–1.1	–1.1	–1.1	–1.1	–1.1	–1.1	–1.1
Groundwater 1 percolation (mm h^−1)	15	15	15	15	15	15	15
GW 2 percolation (mm h^−1)	15	15	15	15	15	15	15

Table 4. Typical ranges of calibrated parameter values of the HSPF model for different land types in the sub-basins.

HSPF parameter	Units	Definition	Range of calibrated values
CEPSC	mm	Interception storage capacity	2.5–17.8
UZSN	mm	Upper zone nominal soil moisture storage	33–38
NSUR	-	Manning’s n for overland flow	0.35–0.45
INFILT	mm h^−1	Index to infiltration capacity	0.8–13
INTFW	-	Interflow inflow parameter	15–25
IRC	-	Interflow recession parameter	0.88–0.90
LZSN	mm	Lower zone nominal soil moisture storage	127–305
LZETP	-	Lower zone evapotranspiration parameter	0.06–0.65
KVARY	mm^−1	Variable groundwater recession	5.1–7.6
AGWRC	-	Base groundwater recession	0.84–0.92
DEEPFR	-	Fraction of groundwater inflow to deep recharge	0.0
BASETP	-	Fraction of remaining evapotranspiration from baseflow	0.01–0.30
KMELT	mm °C^−1 h^−1	Degree-day factor	0.25–ß10

Figure 5. Simulated and observed flows using WATFLOOD, HSPF, HBV-EC and HEC-HMS models for the calibration period (2005–2015).

Figure 6. Simulated and observed flows using WATFLOOD, HSPF, HBV-EC and HEC-HMS models for the validation period (1994–2004).

Table 5. Comparison of the performance of various hydrological models in simulating runoff in the Shellmouth basin. NSE: Nash-Sutcliffe efficiency criterion; r: correlation coefficient; RMSE: root mean square error; MARE: mean absolute relative error; D_v: deviation of runoff volume (underlined bold indicates the best model performance and bold indicates the second best model performance).

Model	Calibration data					Validation data
	NSE	r	RMSE	MARE	Dv	NSE	r	RMSE	MARE	Dv
			(m^3s^−1)	(%)	(%)			(m^3s^−1)	(%)	(%)
WATFLOOD	0.49	0.77	44.1	184.8	31.4	0.48	0.74	30.1	198.3	53.7
HSPF	0.44	0.85	46.3	128.0	33.8	−0.63	0.81	53.4	211.1	156.3
HBV-EC	0.37	0.82	49.2	185.3	48.6	−1.96	0.62	72.0	500.1	250.5
HEC-HMS	0.39	0.69	48.4	231.5	29.5	−0.40	0.60	49.5	587.9	164.0

Figure 7. Scatter plots of measured vs computed flows using WATFLOOD, HSPF, HBV-EC and HEC-HMS models for the calibration and validation datasets.

Table 6. Comparison of the performance of the hydrological models in simulating runoff in the Shellmouth basin for the spring runoff period (underlined bold indicates the best model performance and bold indicates the second best model performance).

Model	Calibration data					Validation data
	NSE	r	RMSE	MARE	Dv	NSE	r	RMSE	MARE	Dv
			(m^3 s^−1)	(%)	(%)			(m^3 s^−1)	(%)	(%)
WATFLOOD	0.30	0.68	68.6	89.0	18.2	0.49	0.72	47.4	121.4	20.9
HSPF	0.22	0.85	72.5	76.3	43.3	−0.50	0.80	80.8	195.7	129.5
HBV-EC	0.14	0.82	75.9	111.8	60.0	−1.14	0.64	96.6	369.7	174.2
HEC-HMS	0.45	0.72	60.8	88.0	6.4	0.20	0.66	59.2	224.9	65.7

Figure 8. Scatter plots of measured vs computed flows using WATFLOOD, HSPF, HBV-EC and HEC-HMS models for the calibration and validation datasets during the spring runoff period.

Figure 9. Simulated and observed flows using WATFLOOD, HSPF, HBV-EC and HEC-HMS models for operational forecasting in 2016 and 2017.

Table 7. Comparison of the performance of the different models for operational flood forecasting in 2016 and 2017 (underlined bold indicates the best model performance and bold indicates the second best model performance).

Model	NSE	r	RMSE	MARE	Dv
			(m^3s^−1)	(%)	(%)
WATFLOOD	0.30	0.65	44.1	135.4	40.6
HSPF	0.83	0.91	21.9	45.0	–7.5
HBV-EC	0.80	0.90	23.4	42.3	–14.7
HEC-HMS	0.01	0.42	52.1	155.9	–0.7

Figure 10. Scatter plots of measured vs computed flows using WATFLOOD, HSPF, HBV-EC and HEC-HMS models for operational forecasting in 2016 and 2017.