Is it possible to write final parameter from calibration back to an ArcSWAT project?

[Jane]

Dear SWAT group,

I have calibrated my SWAT Model with SUFI2 and I am now satisfied with my final parameters. How can I write these parameters back to my original SWAT Project to run my scenarios. Or is it possible to run the model in SWAT CUP to write the Output.mdb files. I tried several things now but nothing really worked. To do it manually seems not possible related to relative changes for some parameters. I could really need some advice?!

[Jaya Pudashine]

Have you got solution of this issue ? I m also having same problem

[Karim Abbaspour]

... The idea that calibration provides you with one set of parameters that you can be put back in the ArcSWAT project (for whatever reason) 

is really misguided and even wrong. Calibration can at best provide you with a possible range of parameters that you must propagate and obtain a possible range of solutions. You can do this with swatcup, therefore there is no need to put one set of parameters back in the ArcSWAT project.

Best, Karim

-------------------------------------------------

Dr. K.C. Abbaspour
Eawag, Swiss Federal Institute for Aquatic Science and Technology
Ueberlandstr. 133, P.O. Box 611, 8600 Duebendorf, Switzerland
email: abba...@eawag.ch
phone: +41 44 823 5359
fax: +41 44 823 5375
http://www.eawag.ch/index_EN

---

From: Jaya Pudashine <lette...@gmail.com>
To: swat...@googlegroups.com
Sent: Thursday, August 22, 2013 8:46 AM
Subject: Is it possible to write final parameter from calibration back to an ArcSWAT project?

Have you got solution of this issue ? I m also having same problem

--
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+unsub...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[Yuestas]

Hello Dr. Abbaspour, 

I guess what we are wondering is, once we spend time calibrating and validating our model and get a range of parameter ranges, is there anyway to re-use this calibrated model (i.e., for future predictions). Is it wrong to pick a certain parameter value from the range that we get from our calibration result to run future predictions? Looking forward to your clarifications.

Thank you for your time.

Yuestas.

[Karim]

Of course you can use your calibrated model for scenario analysis, and future prediction, and climate change analysis etc. 

Once you set up your project in ArcSWAT, your model is summarized in the files in the TxtInOut directory. In swatcup this initial model remains unchanged in the Backup directory. The calibrated parameter ranges are in the par_ inf. txt file. These two completely define your calibrated model. 

For any analysis you need to propagate the parameter ranges. So, if you want to do climate change analysis, for example, you need to update the pcp and tmp files to future values and run, say, 500 simulations to see the output. 

Another example, if you want to see the impact of decreasing fertilizer input on water quality, then you can use swatcup to change fertilizer input and do 500 runs to see the output. Your output in each case will have a distribution,  which you can compare with previous distribution, and so on. It is prudent and important to report this output distribution(uncertainty) because your model always remains uncertain even after calibration.

You can do all kinds of future analysis. You do not need to go to ArcSWAT for these, you can do all kinds of future prediction by making appropriate changes in the TxtInOut files using swatcup. 

If for some reasons you want to use the so called "best" parameter set to do some future analysis, the you can also easily do this in swatcup by placing the best parameter set in the txtinout files and making only on simulation. This is however not recommended.  

I hope this is clear.

Best, Karim

[Jane]

Thank you very much for your answere. This helped me very much to understand my problem.

Best Regards

[yudan1...@gmail.com]

Dear Karim

According to my understanding, the validation steps are:

1. Modify the par_inf.txt: change the range of the parameters.But I don't know where I can find the range after I calibrate the model.

2. Modify the File.cio: first make sure that your simulation period  covers your validation.

                                 change the NBYR to the number of  years validated.

                                  change the IYR to the beginning year of validation.

                                  chage the IDAL to ending julian day of simulation(365 or 366).

3. Modify the Observed_rch.txt and Oberved.txt: your observed data in validation.

4. Modify the SUFI2_extract_rch_def: beginning year of validation 

                                                       end year of validation

But how many simulations could I choose? Just one or more?

Dan

在 2013年8月23日星期五UTC+8上午7时27分04秒，k_abb...@yahoo.com写道：

[Karim Abbaspour]

it is in par_inf file...

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

[Karim Abbaspour]

do as many simulations as you did to calibrate your model without any further changes to your calibrated parameter ranges, which reside in the par_inf file of your last calibration iteration. You need to propagate the uncertainty (parameter ranges) to validation period also.

Best 

Karim

---

From: "yudan1...@gmail.com" <yudan1...@gmail.com>
To: swat...@googlegroups.com
Sent: Monday, September 23, 2013 10:50 AM

Subject: Re: Is it possible to write final parameter from calibration back to an ArcSWAT project?

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

[yudan1...@gmail.com]

 Do you mean the par_inf file in the "..\Iterations\Iter1\Sufi2.In". I have checked my par_inf file but I don't find the so called 'best' parameters set after my calibration. Because I find the parameter ranges are the same as I set the SWATCUP project at the begining. Do you think the parameter ranges are changing during the calibration process? 

在 2013年9月23日星期一UTC+8下午4时54分37秒，k_abb...@yahoo.com写道：

[yudan1...@gmail.com]

Dear Karim:

I succeed in calibrating one iteration(about 500 simulations). In the calibration output I get the New_par.txt. Then I want to do validation, can I just copy the new parameter ranges in the  New_par.txt and paste to the next iteration's input file- par_inf.txt?

 At the same time I modify the File.cio, Observed, rch.txt and Oberved.txt, SUFI2_,extract_rch_def  as I mentioned before.
Do I follow the right steps for validation?
在 2013年9月23日星期一UTC+8下午4时54分37秒，k_abb...@yahoo.com写道：

[Liem, Nguyen Duy]

Dear all,

Sorry for mistake!

It is right here:

http://www.youtube.com/watch?v=7R4xtFHBInY

Liem

2014/1/6 Liem, Nguyen Duy <nguyen...@gmail.com>

Dear Jane

Here is video for you:

http://www.youtube.com/watch?v=7R4xtFHBInY

I hope you succeed!

Liem

2013/8/7 Jane <frau.jan...@gmail.com>

Dear SWAT group,

I have calibrated my SWAT Model with SUFI2 and I am now satisfied with my final parameters. How can I write these parameters back to my original SWAT Project to run my scenarios. Or is it possible to run the model in SWAT CUP to write the Output.mdb files. I tried several things now but nothing really worked. To do it manually seems not possible related to relative changes for some parameters. I could really need some advice?!

--

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

--

Nguyễn Duy Liêm

Bộ môn Tài nguyên & GIS,

Khoa Môi trường & Tài nguyên,

Trường Đại học Nông Lâm TP. HCM.
Địa chỉ: phường Linh Trung, quận Thủ Đức, Tp. HCM, Việt Nam

Website:  http://gis.hcmuaf.edu.vn/ 
Email: nguyen...@gmail.com

          nguyen...@yahoo.com

Mobile: (+84) 0983-613-551

Nguyen Duy Liem

Department of Natural Resources and GIS, 

Faculty of Environment and Natural Resources

Nong Lam University- Ho Chi Minh City.
Address: Linh Trung Ward, Thu Duc District, Ho Chi Minh City, Vietnam.

Website:  http://gis.hcmuaf.edu.vn/  
Email: nguyen...@gmail.com

          nguyen...@yahoo.com

Mobile: (+84) 0983-613-551

--

Nguyễn Duy Liêm

Bộ môn Tài nguyên & GIS,

Khoa Môi trường & Tài nguyên,

Trường Đại học Nông Lâm TP. HCM.
Địa chỉ: phường Linh Trung, quận Thủ Đức, Tp. HCM, Việt Nam

Website:  http://gis.hcmuaf.edu.vn/ 
Email: nguyen...@gmail.com

          nguyen...@yahoo.com

Mobile: (+84) 0983-613-551

Nguyen Duy Liem

Department of Natural Resources and GIS, 

Faculty of Environment and Natural Resources

Nong Lam University- Ho Chi Minh City.
Address: Linh Trung Ward, Thu Duc District, Ho Chi Minh City, Vietnam.

Website:  http://gis.hcmuaf.edu.vn/  
Email: nguyen...@gmail.com

          nguyen...@yahoo.com

Mobile: (+84) 0983-613-551

[m. domingo]

If I understand correctly, you should use the same parameter ranges in validation that you used in your last calibration run. By doing so you are testing the strength of your model configuration. If you were to add the new ranges suggested after your last calibration run, you would no longer be comparing apples to apples.

Mike

[Abbaspour, Karim]

Correct!

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

[LEGEND LEGEND]

how can i use swat cup for bmps mnalysis? i have to change single value of any parameter, i did so but the result in not changing? For filter strip, i changed filterw.mgt value from zero to 2,4,10 and 15 but sediemnt yield remained constant in swat cup. what may be the reasons?

[Eva Mia Siska]

Dear Dr. Abbaspour,

I understand your point below, but my study requires me to look into basin average outputs such as actual evapotranspiration, lateral flow, etc which are defined by output.std. If I wish to use the best simulation, how can I obtain these values from SWAT-CUP files? Or do you have suggestion on what would be the most proper way to display these basin average values which reflecting the calibration results?

Another question: if I understand correctly, your suggestion is that we need to stick to the results in form of sets of parameters ranges. Are you suggesting that we also should have range of model efficiency? say NSE is 0.76-0.85?

Thank you in advance for your clarification?

[Karim Abbaspour]

To use the best simulation, I have written many times how you can put the final best parameters in the swat files and make a run to get the so-called best values. Look at page 66 or 65 in the manual. Basically, follow the step below:

If you want the SWAT files to reflect the best parameters you obtained in an iteration do the following:

1- Note the number of the best simulation in the Summary_Stat.txt file

2- In the SUFI2_swEdit.txt set the starting and ending simulation values both to the number of the best simulation in step 1.

3- Under Calibration, run SUFI2_run.bat. Do not run SUFI2-Pre.bat.

This command will replace the parameter values and set them to the best values of the  iteration.

 

Then use swatcup to visualize the results by using an irrelevant parameter and the  following setup below:

 

In par_inf.txt

1   : Number of Parameters

1   : number of simulations

r__WDPQ.bsn             0       0

 

In SUFI2_swEdit.def

1      : starting simulation number

1    : ending simulation number

 

Then make a complete run: SUFI2_pre, -run, -post  to visualize the results and also output.std to get average basin values.

More correct would be to make an iteration run with as many runs as you made in the final calibration run, extract what you need from various swat output files and then use the outputs to calculate the uncertainty in your numbers.

Yes, you can also report a range in model efficiency from all the runs. These are outputted in goal.txt in the last column. There you see a number of bad simulations as well, which you can avoid by setting a behavioral threshold.

Please see the paper below on how to use the uncertainty in scenario analysis.

http://www.mdpi.com/2073-4441/10/1/6

Karim    

To unsubscribe from this group and stop receiving emails from it, send an email to swat-cup+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/swat-cup/c44985bf-42d0-4c34-9552-005a15c4b240%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Eva Mia Siska]

Dear Dr. Abbaspour,

Thank you for your kind and clear explanation. I will try as suggested.

Best regards,

Eva 

[Mishkat Ul Saba]

Sir Kareem as you have mentioned that we can do any kind of Analysis using swat-cup, for example for future prediction we can change pcp and Tmp files in textinout folder of swatcup to get future river flows as particular outlet or sub basin wise flow ?

[Yuhang.Charles Wei]

Thanks Dr Abbaspour, the methods to write best parameters to the SWAT files and get corresponding output.std is super helpful!

Yuhang

[Dpaceism]

Greetings Dr. Abbaspour,

I calibrated my model for monthly discharge and now I realize that my sediment data has daily observation. In this case I need to run SWAT daily. This will change my TxTinOut file that is to be imported in SWAT CUP. How do I handle this situation now? I was thinking of writing the best parameter set back to Arc SWAT(to have the model is calibrated for the flow), run the model daily and calibrate for sediment again, but it will not be a correct approach. Hoping to getting your response.

Thank you for your time and consideration.

Sincerely,

Dipesh

[2w2e]

SWAT always runs daily. You just have to tell SWAT to write daily data, and provide daily data in the swatcup observation files. no need to change txtinout. 

Also, there there is no need to put parameters back in ArcSWAT. It actually makes no sense. This has bee discussed several times. Perhaps see:

https://www.youtube.com/watch?v=ZxHpPYe9iB8&t=2131s