PBIAS in SWATCUP different than PBIAS calculated in spreadsheet

[Mike Bullard]

Here is summary_stat output for a daily time step SWAT project.  PBIAS is shown as 10.8%.  But when I calculate in a spreadsheet using (sum of obs - sum of sim)/(sum of obs)  I am getting   -1.29%

Variable   p-factor r-factor R2     NS     bR2   MSE       SSQR    PBIAS KGE  RSR   MNS VOL_FR --- Mean_sim(Mean_obs) StdDev_sim(StdDev_obs)

FLOW_OUT_97 0.21    0.00    0.66    0.64   0.3829 6.0e+000 1.9e+000 10.8 0.64 0.60 0.53 1.12        2.48(2.78)          2.94(4.11)

Is SWATCUP using the formula (sum of obs - sum of sim)/(sum of obs) to calculate PBIAS?  If so, then I will keep trying to find my error.

Thanks for any assistance.

[charlesi @iastate.edu]

Mike,

I have observed the same thing (and posted about it a few weeks ago).  In fact, my PBIAS  calculations in a spreadsheet outside of SWAT-CUP seem to match the VOL_FR stat as reported in SWAT-CUP (Sufi2).

--Charles

--
You received this message because you are subscribed to the Google Groups "SWAT-CUP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to swat-cup+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Abbaspour, Karim]

Did you use any weights? What do you mean match VOL_FR?

Karim

--

To unsubscribe from this group and stop receiving emails from it, send an email to swat-cup+u...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

 

--

You received this message because you are subscribed to the Google Groups "SWAT-CUP" group.

To unsubscribe from this group and stop receiving emails from it, send an email to swat-cup+u...@googlegroups.com.

[Mike Bullard]

Hi Karim,

No weighting (I think) , here is the objective function observed.txt:

1 : number of observed variables

5 : Objective function type, 1=mult,2=sum,3=r2,4=chi2,5=NS,6=br2,7=ssqr,8=PBIAS,9=KGE,10=RSR,11=MNS

0.1 : min value of objective function threshold for the behavioral solutions

1 : if objective function is 11=MNS (modified NS),indicate the power, p.

　

　

　

FLOW_OUT_97 : this is the name of the variable and the subbasin number to be included in the objective function

1 : weight of the variable in the objective function

-1 : Dynamic flow separation. Not considered if -1. If 1, then values should be added in the forth column below after observations

-1 : constant flow separation, threshold value. (not considered if -1)

1 : if separation of signal is considered, this is weight of the smaller values in the objective function

1 : if separation of signal is considered, this is weight of the larger values in the objective function

10 : percentage of measurement error

1826 : number of data points for this variable as it follows below. First column is a sequential number from beginning

: of the simulation, second column is variable name and date (format arbitrary), third column is variable value.

[Abbaspour, Karim]

Send me best_sim file to check.

Karim

[Mike Bullard]

Hi Karim,

I checked the best_sim file and the PBIAS calculated from that matches what is shown in the summary_stat. 

However, the two PBIAS I am comparing are the PBIAS in SWATCUP, and the PBIAS from the calibrated SWAT model.

To calibrate the SWAT model, I am taking the best params from SWATCUP and using the manual edit in SWAT to update those params (just once--and I am keeping track of that very carefully) in SWAT, re-run SWAT.

I then get the sim from SWAToutput.mdb in the tables out folder. 

Maybe I am not doing the manual calibration correctly. 

In SWATCUP,  I do:

1.     CN2                     Relative         -0.3

2.     ALPHA_BF          Replace          0.026

3.     SOL_AWC           Relative          0.7

In ArcSWAT, I do:

1.     CN2                     Multiply By      0.7

2.     ALPHA_BF          Replace By     0.026

3.     SOL_AWC           Multiply By      1.7 

If all this is correct, I will keep checking my work.  Maybe just run both SWATCUP and ArcSWAT without calibration and compare the PBIAS. 

Thanks,

Mike

[charlesi @iastate.edu]

In my spreadsheet I've calculated PBIAS as:

100* [ Sum(Obs-Sim) / Sum(Obs) ]

The resulting PBIAS is different from the PBIAS reported in SWAT-CUP, but the equation in the SWAT-CUP user manual is the same:

--Charles

On Wed, Sep 21, 2016 at 7:49 PM, Erick Sebadelhe <erickse...@hotmail.com> wrote:

eu
Dear Mike,
according to Moriasi et al (2007) PBIAS = sum (obs-sim) * 100 / sum (obs). Thus, if you are using PBIAS = (sum of obs-sum of sim) / (sum of obs) as you said, the results should be different.

--
You received this message because you are subscribed to the Google Groups "SWAT-CUP" group.

To unsubscribe from this group and stop receiving emails from it, send an email to swat-cup+unsubscribe@googlegroups.com.