SWAT+ Toolbox Update

1,693 views
Skip to first unread message

Celray James CHAWANDA

unread,
Dec 8, 2020, 3:24:12 PM12/8/20
to SWAT+

SWAT+ Toolbox v0.5

Dear all, I am excited to let you know that version 0.5.0 of SWAT+ Toolbox is now available with built in updates.

Automatic calibration has been updated with multi-site calibration and parameter density plots.

There are also other user interface improvements and bug fixes. There will be an update soon to enable adding different forms of phosphorus in observations for calibration and some improvements, so keep checking software updates regularly by clicking “check for updates" in the ‘About’ section.

You can freely download through

the Open Water Network Website: http://openwater.network/assets/downloads/SWATPlusToolboxv0.5.0Installer.exe

Or through OneDrive: https://onedrive.live.com/?authkey=%21AKaNcf%2D4p%2DkVBC4&id=BB883823C529A9AF%211027878&cid=BB883823C529A9AF

SWAT+ Toolbox will also be on the SWAT Website later.

Your feedback is highly appreciated as it is helping shape SWAT+ Toolbox and also find and fix bugs which I would not find on my own.

Regards to all,

James

live_par_dens_and multi_cal_plots.jpg
about_page with update button.png

ana maria paez trujillo

unread,
Dec 9, 2020, 5:48:03 AM12/9/20
to SWAT+
Hello James,

Thanks for the tutorials and the contributions that you are making, they are very helpful.
 
I'm working on my model calibration and I would like to use SWAT+ Toolbox. I installed the version 0.5.0 of SWAT+ Toolbox, the option appeared in the SWAT+ editor, but I couldn't open it, then I tried to open it directly from the SWAT+ Toolbox icon but it didn't work. These are the problem details that I got. I installed the previous version and I got the same problem. Do you how could I solve them?



Kind regards
Ana Maria

ana maria paez trujillo

unread,
Dec 9, 2020, 5:54:18 AM12/9/20
to SWAT+
  Hello James,

Thanks for the tutorials and the contributions that you are making, they are very helpful.
 
I'm working on my model calibration and I would like to use SWAT+ Toolbox. I installed the version 0.5.0 of SWAT+ Toolbox, the option appeared in the SWAT+ editor, but I couldn't open it, then I tried to open it directly from the SWAT+ Toolbox icon but it didn't work. These are the problem details that I got. I installed the previous version and I got the same problem. Do you how could I solve them?

Kind regards
Ana Maria
Problem details.pdf

Celray James CHAWANDA

unread,
Dec 9, 2020, 1:14:24 PM12/9/20
to SWAT+
Dear Ana,

I guess you are referring to QSWAT+ when you say 'it appeared in SWAT+ Editor'. It seems you are running Windows 7. SWAT+ Toolbox depends on .Net Framework 4.7.2+ which is usually available on Windows 10 but you have to install on Windows 7. Make sure this is installed on your computer, you can download it here: http://openwater.network/assets/downloads/NDP472-KB4054530-x86-x64-AllOS-ENU.exe

If you still face problems, I would be interested to talk to you and find out what is causing any further issue. In that case, let me know when you would be free using swatplus_celrayjames[at]outlook.com.

Kind Regards

ana maria paez trujillo

unread,
Dec 11, 2020, 4:58:05 AM12/11/20
to Celray James CHAWANDA, SWAT+
Hello James, Thanks for your answer. I installed  Net Framework 4.7.2+, and  SWAT+ Toolbox is working. I'm exploring the toolbox and so far it is working pretty well.

I have some questions. I can not find the PBias report in the calibration section, in the video tutorials it appears next to the NSE and RMSE reports. Did the location of the PBias report change in version 0.5.0? I Where could I find that report?
The second question is about the calibration process. I'm working on my model calibration and I completed the sensitivity analysis. There is a parameter with high sensitivity and the others report very low sensitivity. What does it mean for the calibration? If the sensitive parameter gets a stable value during the calibration process what is going to happen with the calibration of the other parameters? Should I calibrate the sensitive parameter first, and then calibrate the less sensitive parameters?

Kind regards
Ana Maria


--
You received this message because you are subscribed to a topic in the Google Groups "SWAT+" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/swatplus/mdvMPioKTIY/unsubscribe.
To unsubscribe from this group and all its topics, send an email to swatplus+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/swatplus/0b0ace6e-393c-48e8-af62-fb50be7c6109n%40googlegroups.com.

Jacob Wood

unread,
Dec 11, 2020, 2:23:59 PM12/11/20
to SWAT+
Hello Jame, and thank you for this useful tool.  In attempting to calibrate flow at the monthly scale, I am receiving an error indicating that there are duplicate timesteps (see attached).  I am also attaching my observation file for review.   I am certain there are no duplicate dates in the data, and the date format is dd/mm/yyyy.  

Have you encountered this error before, or do you have any suggestions for a solution?

Thank you,

Jacob



IMG_1952.jpg
SCC_FLOW_MONTHLY.csv

Celray James

unread,
Dec 11, 2020, 4:07:31 PM12/11/20
to Jacob Wood, SWAT+

Indeed, If your system date format is mm/dd/yyyy instead of dd/mm/yyyy, SWAT+ Toolbox thinks the day is changing and the month is constant (01/01/yyyy to 01/02/yyyy can be a month difference or a day difference depending on the date format of your system). Take a look at this to see how you can change date format for your system: https://youtu.be/1uN3NtH1jLk?t=642

 

Let me know how that goes and if you need further assistance

 

           With Kind Regards,

           Celray James CHAWANDA

--

You received this message because you are subscribed to a topic in the Google Groups "SWAT+" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/swatplus/mdvMPioKTIY/unsubscribe.
To unsubscribe from this group and all its topics, send an email to swatplus+u...@googlegroups.com.

Celray James

unread,
Dec 11, 2020, 4:36:29 PM12/11/20
to ana maria paez trujillo, SWAT+

Hi Ana,

 

Great to hear that it works now!

 

All objective functions are available in the manual calibration section. In the automatic calibration, you only see the objective function that you have selected for use when you run the calibration algorithm. However, you can still watch how the objective functions such as PBias and RMSE change during automatic calibration if you monitor the objective functions in the manual calibration section. If you want to check the evolution of the selected objective function, Click one of the newly introduced navigation arrows within the automatic calibration section.

 

Sensitivity analysis helps us narrow down our efforts during calibration so that we can focus on the most influential parameters. The idea is so that you focus on few parameters instead of trying parameters that do not make a difference at all in your simulations.

 

If you are doing a manual calibration, then indeed you can start with the most sensitive parameter and go down the parameter list according to sensitivity. However, it is slightly different when you are dealing with automatic calibration. In that case, you chose a number of parameters to be included in calibration depending on how much time you are willing to dedicate. You can remove the non-sensitive parameters. This is because if you reduce the number of parameters for automatic calibration to only the more sensitive parameters, you arrive at your solution faster.

 

Note that sometimes even with little sensitivity, some parameters might still add that small model performance boost (even if it is little). So there is a trade-off between potential better performance if you include the parameters with low sensitivity and additional time it would take during calibration to realise those small gains.

 

I hope that was helpful.

 

           With Kind Regards,

           Celray James CHAWANDA

 

Kind regards

Ana Maria

El martes, 8 de diciembre de 2020 a las 21:24:12 UTC+1, Celray James CHAWANDA escribió:

SWAT+ Toolbox v0.5

Dear all, I am excited to let you know that version 0.5.0 of SWAT+ Toolbox is now available with built in updates.

Automatic calibration has been updated with multi-site calibration and parameter density plots.

There are also other user interface improvements and bug fixes. There will be an update soon to enable adding different forms of phosphorus in observations for calibration and some improvements, so keep checking software updates regularly by clicking “check for updates" in the ‘About’ section.

You can freely download through

the Open Water Network Website: http://openwater.network/assets/downloads/SWATPlusToolboxv0.5.0Installer.exe

Or through OneDrive: https://onedrive.live.com/?authkey=%21AKaNcf%2D4p%2DkVBC4&id=BB883823C529A9AF%211027878&cid=BB883823C529A9AF

SWAT+ Toolbox will also be on the SWAT Website later.

Your feedback is highly appreciated as it is helping shape SWAT+ Toolbox and also find and fix bugs which I would not find on my own.

Regards to all,

James

Lucas Graunke

unread,
Dec 14, 2020, 1:28:19 AM12/14/20
to SWAT+
Hi James, 

First off, thank you for creating such a useful tool for swat+ plus users! I'm really enjoying getting to use this new software. 

I am using version 0.5.1 of swat+ toolbox and have a couple of questions. 

For my calibration, I am using three different sites with three different observation files. How do I know if the toolbox is using all three observation files when doing the calibration, or do I need to make one big observation file for the multi-site calibration? 

With the old version of the toolbox, the Current Best values in the automatic calibration area would fill and update with each run of the calibration, but I am only seeing 0.00 for all of my Current Best values. 

In the table that lists all of my observation channels, their NSE number and the weight, the NSE number updates for every run of the calibration but the graph below called OBJ. FX. Value Evolution stays at -1000 NSE for the entire calibration. The NSE values in the table above are not -1000. 

Also the graph that shows the evolution of all the parameters values does not have any data in it, even after running the calibration for 700 iterations. 

If you have any advice on how to solve these issue I would really appreciate the input! Thanks again for all of your help!

Lucas 

Celray James CHAWANDA

unread,
Dec 14, 2020, 9:24:23 AM12/14/20
to SWAT+
Hi Lucas!

Great to hear that you are running version 0.5.1! The first person I know to have used that new 'check for updates' button!

For the answers that I provide to your great questions, refer to the attached picture.

The first question:
To use multi-site calibration, all you have to do is check the 'multi-site calibration' checkbox and assign weights as shown in the screenshot attached (highlighted in green). Once you check the 'multi-site calibration' checkbox, the observation selection will become unavailable because it will use all observation files based on the weights. You can see that as you do the automatic calibration, the BEST OBJ FX is a 'weighted average' of the best objective function of each observation. I am very interested to see how this goes since you are one of the first to test out this new feature.

The other questions:
These questions are related because they point to either of the following situations. But before I mention the situations, let me say that at the beginning of the automatic calibration using NSE as objective function, the default best NSE value = -1000
when you get a better NSE, then it increases to the new best accordingly.

Situation 1: You selected 'Minimise' as the direction of the algorithm search (highlighted in green). in this case, any better NSE you get is greater than -1000 and the algorithm is going to keep the smaller (-1000; because of minimise) and you will see the behaviour you have described.
Situation 2: You correctly selected Maximise for NSE but your model actually has NSEs less than -1000. In this case, it will keep the default -1000 since it is better than any NSE value you get after each iteration. The easiest way to check if your model has a reasonable starting NSE is to check model performance on the manual calibration page.

I would love to hear how you progress with your calibration. Let me know if you need further assistance and keep me posted

Regards,
James
direction_and_weights.png

Lucas Graunke

unread,
Dec 15, 2020, 11:48:47 AM12/15/20
to SWAT+
Thank you James! 

Once I had set weights for all of my sites then everything worked out perfectly. 

Thanks again for the help! 

Lucas 

David Haro Monteagudo

unread,
Oct 8, 2021, 12:31:41 PM10/8/21
to SWAT+
Hi James,

I just started using SWAT+ Toolbox v0.7.6 and I noticed that it takes way too long to run the model (more than 5 minutes for a 3 subbasins model and 35 years of daily data). Is there something I am doing wrong? I just moved to SWAT+ from traditional arcSWAT and SWAT-CUP a few days ago and I just feel everything is slower.

Cheers,
David

Celray James

unread,
Oct 8, 2021, 4:01:22 PM10/8/21
to David Haro Monteagudo, SWAT+

Dear David,

 

SWAT+ Toolbox does not affect how fast your model runs. You can check this by running the model in SWAT+ Editor or using the Executable from the website in your TxtInOut.

 

Here are a few things you may want to check to improve how fast it runs.

  1. Level of detail. The more HRUs you have the slower it runs. If you do not need a spatially very detailed model, you may want to simplify your model by reducing the number of HRUs.
  2. What you are printing. Only print the output files you require. The more files you print (especially if you print all daily output files) the slower it will run as the files take long to write to disk (which is also dependent on your computer’s hard disk performance)
  3. Do you need to run all 35 years for calibration/validation? If you are calibrating, setting a shorter period (about 7 years should be enough in your case) could get things done quicker.

 

Other than that, if you notice a difference in run times between SWAT+ Toolbox and SWAT+ Editor, please alert me so I can follow up on your case.

 

With Kind Regards,

Celray James CHAWANDA

 

--

You received this message because you are subscribed to a topic in the Google Groups "SWAT+" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/swatplus/mdvMPioKTIY/unsubscribe.
To unsubscribe from this group and all its topics, send an email to swatplus+u...@googlegroups.com.

David Haro Monteagudo

unread,
Oct 11, 2021, 12:59:30 PM10/11/21
to SWAT+
Hi James,

You are right, it does not. It is actually SWAT+ what takes longer to run in comparison to traditional SWAT. The problem is that this will affect the possibilities of running sensitivity analysis and/or calibration of large models in the toolbox. Any ideas why SWAT+ takes that much to run? The same model that would run in less than 30 seconds in SWAT, now takes more than 10 minutes in SWAT+.

Harun Alp

unread,
Feb 1, 2022, 1:53:22 AM2/1/22
to SWAT+
Hi James,
 I am trying to calibrate my QSAT+ model but I have some questions in my mind about multi-site calibration. So I need to read the documentation but I can not open the calibration section in the link you have given. (https://celray.github.io/docs/swatplus_toolbox/calibration.html) May I ask your help on this issue?

Best Regards
Harun

8 Aralık 2020 Salı tarihinde saat 23:24:12 UTC+3 itibarıyla Celray James CHAWANDA şunları yazdı:

Fereshteh Kordrostami

unread,
Jul 20, 2023, 1:58:03 PM7/20/23
to SWAT+
Hello James, 

I would like to know if it is possible to run the SWAT+ Toolbox using parallel processing.

Thanks, 
Fereshteh



Celray CHAWANDA

unread,
Jul 21, 2023, 1:43:57 PM7/21/23
to swat...@googlegroups.com

Hi Fereshteh,

Please use the latest version of SWAT+ Toolbox available from the website.
Currently, Parallel processing is not supported.

C. James.

--
You received this message because you are subscribed to the Google Groups "SWAT+" group.
To unsubscribe from this group and stop receiving emails from it, send an email to swatplus+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/swatplus/7bf313d9-14dd-4b57-a8d2-bfeebe34d76an%40googlegroups.com.
Reply all
Reply to author
Forward
0 new messages