comparison of multiple point and single point calibration … · 2015. 10. 16. · comparison of...
Post on 08-Mar-2021
5 Views
Preview:
TRANSCRIPT
Comparison of multiple point and single point calibration
performance for the Saginaw River Watershed
Fariborz Daneshvar, A1. Pouyan Nejadhashemi1, Matthew R. Herman1
1Department of Biosystems and Agricultural Engineering, Michigan State University
2015 SWAT Conference at Purdue
October 14-16th
SWAT Model Calibration-Validation
22015 SWAT CONFERENCE
SWAT uses a large number of default input parameters for model development
Subbasin level parameters (e.g. ground water, soil, routing …)
Basin level parameters (e.g. surface runoff lag coefficient)
Calibration-Validation process is required to verify the accuracy of the model for the simulated
condition
Single point calibration is the common approach for water quantity and water quality calibration
Challenges
32015 SWAT CONFERENCE
Single point calibration will not account for spatial variation of hydrological processes.
Multiple point calibration was introduced to tackle this issue
Multiple point calibration software (e.g. SWAT-CUP) still uses unique basin level parameters for the
whole watershed.
Time consuming process!
Solution
42015 SWAT CONFERENCE
Introduce a new multiple point calibration process which is:
Faster
Can use multiple values for basin level parameters.
Study Area
52015 SWAT CONFERENCE
Figure 1- Study area
Saginaw River Watershed
Largest HUC-6 (040802) in Michigan
Total drainage area 16,120 km2
Drains to the Lake Huron
Defined as an area of concern by EPA
Mostly agricultural and forest lands (74%)
SWAT ProjectSWAT 2012 (rev-622)
Watershed Delineation
Predefined reaches and Subbasins by Michigan Institute of Fisheries
Research (IFR), based on National Hydrology Dataset Plus (NHDPlus)
Land-use
Cropland Data Layer 2013
Soil
SSURGO soil
Slope
5 slopes (1-2,2-4,4-7,7-10,>10)
Climate Data
NCDC data for 1990-2010 period
62015 SWAT CONFERENCE
Figure 2- SWAT overlapping layers
Calibration Criteria
Criteria Very Good Good Satisfactory Unsatisfactory
NSE 0.75 – 1.0 0.65 – 0.75 0.5 – 0.65 < 0.5
PBIAS (%) < ±10 ±10 - ±15 ±15 - ±25 > ±25
RSR 0.0 - 0.5 0.5 – 0.6 0.6 – 0.7 > 0.7
72015 SWAT CONFERENCE
Table 1- Flow calibration criteria1
1 Moriasi D.N., Arnold J.G., Van Liew M.W., Bingner R.L., Harmel R.D., Veith T. L., 2007, Model evaluation guidelines for systematic quantification of accuracy
in watershed simulations. Trans. ASABE. vol. 50(3), pp. 885-900.
Nash-Sutcliffe Efficiency (NSE)
Percent Bias (PBIAS)
RMSE-observations Standard-Deviation Ratio (RSR)
Single Point Calibration
82015 SWAT CONFERENCE
Saginaw River (USGS 04157000) station is used for flow calibration
13242 out of 13831 reaches (~96% of watershed) drain to this river
All reaches/Subbasins parameters (e.g. ground water, soil,
management, etc.) are assumed to be the same.
Monthly flow calibration:
Calibration: 2002-2003
Validation: 2004-2005
Calibration Validation
NSE PBIAS (%) RSR NSE PBIAS(%) RSR
0.72 6.02 0.53 0.85 -1.74 0.39
Figure 3- Single point Calibration
Single Point Calibration
92015 SWAT CONFERENCE
No spatial variation is considered
Each year of SWAT run ~ 5 min.
Cons.:
Figure 3- Single point Calibration
Multiple Points Calibration
102015 SWAT CONFERENCE
Using 9 USGS stations for calibration
USGS 04144500
(1053 reaches)
USGS 04155500
(553 reaches)
USGS 04154000
(627 reaches)
USGS 04151500
(2140 reaches)
USGS 04148140
(773 reaches)
USGS 04147500
(1311 reaches)
USGS 04156000
(2682 reaches)
USGS 04148500
(1048 reaches)
USGS 04157000
(3055 reaches)
Figure 4- Multiple point Calibration
Multiple Points Calibration
112015 SWAT CONFERENCE
Daily Flow Calibration
Calibration: 2001-2005
Validation: 2006-2010
Calibration Validation
USGS Station NSE PBIAS (%) RSR NSE PBIAS(%) RSR
04144500 0.64 20.88 0.6 0.62 9.12 0.62
04147500 0.61 9.43 0.62 0.65 -10.04 0.59
04148140 0.54 9.69 0.68 0.53 -24.98 0.68
04151500 0.59 15.14 0.64 0.68 12.62 0.57
04154000 0.54 5.51 0.68 0.54 13.24 0.68
04155500 0.63 20.1 0.61 0.59 1.29 0.64
04148500
04156000
04157000
Calibration Validation
USGS Station NSE PBIAS (%) RSR NSE PBIAS(%) RSR
04144500 0.64 20.88 0.6 0.62 9.12 0.62
04147500 0.61 9.43 0.62 0.65 -10.04 0.59
04148140 0.54 9.69 0.68 0.53 -24.98 0.68
04151500 0.59 15.14 0.64 0.68 12.62 0.57
04154000 0.54 5.51 0.68 0.54 13.24 0.68
04155500 0.63 20.1 0.61 0.59 1.29 0.64
04148500 0.65 26.3 0.59 0.76 8.08 0.49
04156000 0.73 3.27 0.52 0.74 8.98 0.51
04157000
Calibration Validation
USGS Station NSE PBIAS (%) RSR NSE PBIAS(%) RSR
04144500 0.64 20.88 0.6 0.62 9.12 0.62
04147500 0.61 9.43 0.62 0.65 -10.04 0.59
04148140 0.54 9.69 0.68 0.53 -24.98 0.68
04151500 0.59 15.14 0.64 0.68 12.62 0.57
04154000 0.54 5.51 0.68 0.54 13.24 0.68
04155500 0.63 20.1 0.61 0.59 1.29 0.64
04148500 0.65 26.3 0.59 0.76 8.08 0.49
04156000 0.73 3.27 0.52 0.74 8.98 0.51
04157000 0.79 11.66 0.46 0.8 10.3 0.44
Figure 4- Multiple point Calibration
Multiple Points Calibration
122015 SWAT CONFERENCE
0
200
400
600
800
1000
1200
1/1/2001 1/1/2002 1/1/2003 1/1/2004 1/1/2005 1/1/2006 1/1/2007 1/1/2008 1/1/2009 1/1/2010
Series1 Series2Obs. Sim.
Figure 5- observed flow vs. simulated flow for USGS 04157000 station
Conclusion
132015 SWAT CONFERENCE
Longer calibration period can be achieved.
Multiple point calibration will speed up the process.
SWAT parameters can be changed locally, which is more realistic compared to
previous approaches.
Acknowledgement
We would like to thank the Environmental Science and Policy Program (ESPP) at Michigan State University for funding this project
142015 SWAT CONFERENCE
Thank You
152015 SWAT CONFERENCE
top related