Calibration of MODFLOW, GSFLOW and PRMS

[Chang Liao]

A long question related to calibration.

I know USGS provide ModelMate, UCODE, PEST and PEST++ for model analysis and calibration for MODFLOW.

The GSFLOW also encourage individual calibration of PRMS and MODFLOW using PRMS-only simulation and MODFLOW-only simulation.

The GSFLOW (Instructions for Running GSFLOW Models) http://wwwbrr.cr.usgs.gov/projects/SW_MoWS/GSFLOW%20-%20Instructions%20for%20Running.html

also provide some simple steps to calibrate.

PRMS site MOWS also provide LUCA http://wwwbrr.cr.usgs.gov/projects/SW_MoWS/LUCA.html  for calibration using SCEUA global optimization method.

So, I am kind of confused which to follow in different cases in terms of calibration and analysis. 

Such as what to use if I run MODFLOW-2005, and what to use if I run GSFLOW with different simulations?

Also, I haven't finished reading the PEST++ UCODE manual, so what is the calibration method used in them? (I know PEST++ have the SCEUA utility)

[Francesca Moschini]

I'm calibrating PRMS using LUCA and preparing the files to calibrate GSFLOW using PEST (it's a pain though, the manual is huge!). I can't calibrate MODFLOW that much because I don't have observation wells, so I calibrated it manually using the river outflow. 
My LUCA calibration has been running for one week now and I have at least a week to go. A problem that I faced using LUCA is that the initial value for parameters has to be an integer, if I put comma values the program crashes. So in my case the calibration gets longer since my starting values are really far from coverage!

If you find easy way to calibrate GSFLOW, please share!

Best regards!

[Chang Liao]

I have similar situation here. Not enough observation data to verify. And I plan to use the discharge at the outlet to verify the MODFLOW.
So basically I can't get the well data, but I will use some soil moisture profile if possible.
Also welcome to send me email to discuss more details.

[Far_Naz]

Dear  Francesca Moschini

Would you please say  number of HRUs and property of computer you used for calibrating PRMS by LUCA . I tried LUCA example and it lasts so long.

Regards,

[Francesca Moschini]

Hello Far_Naz,

I have 223 hrus and I have used my pc to run the calibration. I don't have the specifications right now but it is not powerful at all!

It took me more than a week to calibrate PRMS and the results were not satisfying. Maybe it's better if you use PEST!

[ZAHEER AHMED]

please provide me some example of modlfow 2011.1

[Francesca Moschini]

Hello Chang,
Have you started calibrating gsflow?
In the end I calibrated PRMS with PEST and now I'm preparing the pest control file (.pst) for the GSFLOW model using TSPROC.exe.
However I have a problem matching the modeled streamflow from the  .sfrg1 file with the observed streamflow than I take from the statvar file.
Tsproc gives me the following error message :" Error: OBSERVATION_SERIES "str_obs" has been matched to MODEL_SERIES 

   "str_sim". However these series have different numbers of terms."

How did you prepare the file?

Thanks 

Francesca

[Francesca Moschini]

I solved just by removing the time :/

[Chang Liao]

Not yet. I have over 2 million HRUs and I have to do all the data preparation and simulation on HPC.  (I am pushing the limit of these tools as the USGS guy commented)

And it takes longer time to do every step. But I have almost finished both PRMS and MODFLOW simulation.

[Chang Liao]

Could anyone provide an example of calibration MODFLOW-2005 using PEST/PEST++?

[Aklilu Dinkneh]

Hello Francesca

I have a question regarding your PRMS calibration using Luca. did you do some sensetivity analysis to decide which parameters to calibrate? do you have any documentation which explain how to calibrate using Luca?

thank you for your help

[Francesca Moschini]

Luca gave me problems, so I decided to use PEST in the end. With PEST you can run a sensitivity analysis and decide which parameters to calibrate.
the documentation that I used when I tried to use LUCA was the one attached to the program when I downloaded it.

I strongly advice you to use PEST, check the swutils.pdf manual to understand how to write the files.

I can send you my files so that you can have a better idea, if you want.

[Sontaya Phakdiwatsakun]

Dear Francesca Moschini 

now I am deciding choosing model for calibration, now I use modflow 2005 for simulation.  

I am trying to use modelmate but I quite don't understand how to creat the parameter and observation file. 

modelmate is giving hugr problem, i didn't the solver ways at its manual.

I want to try using PEST

could you send me your files, it would be better for me to get better idea. 

Thanks

เมื่อ วันพฤหัสบดีที่ 7 พฤษภาคม ค.ศ. 2015 11 นาฬิกา 49 นาที 21 วินาที UTC+9, Francesca Moschini เขียนว่า:

[Francesca Moschini]

Hello Sontaya,

I didn't calibrate modflow, just PRMS. However if you read the gw utilities for Pest you'll find all the examples you need.

Good luck!

Francesca

[Chang Liao]

MODFLOW packages only support limited parameters options, and they can be specified using pval file.

However, for PEST calibration, you can add as many parameters as possible if you want.

But you should be careful with the parameter within PEST control file, since PEST only works when MODFLOW could converge and produce results.

The more parameters you added, the longer it takes to run. So I recommend run Sensitivity first, and then choose those interested for calibration.

[Sontaya Phakdiwatsakun]

so how can I generate PVAL file ? 

I try to use modelmate but it requires PVAL file, I don't have any idea how to get that file. 

--
You received this message because you are subscribed to a topic in the Google Groups "MODFLOW Users Group" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/modflow/VuvxWA2awoA/unsubscribe.
To unsubscribe from this group and all its topics, send an email to modflow+u...@googlegroups.com.
To post to this group, send email to mod...@googlegroups.com.
Visit this group at http://groups.google.com/group/modflow.
For more options, visit https://groups.google.com/d/optout.

[Chang Liao]

PVAL - Parameter Value File

Take a look at:

http://water.usgs.gov/ogw/modflow/MODFLOW-2005-Guide/parameter_value_file.htm

[Sontaya Phakdiwatsakun]

Thank you Chang Liao

I am trying :) 

[Aklilu Dinkneh]

Hello Again,

Yes please that will save me a lot of time(if you can send me your file)

thanks a lot 

aklilu

[Francesca Moschini]

Hi, I've contacted you on hangouts 

[Chang Liao]

So after certain time of research, I was able to calibrate the PRMS and MODFLOW-NWT using PEST. 

As it takes days to finish the calibration if you have more than 10 parameters during MODFLOW calibration, I have applied the BeoPEST to achieve my goal on HPC. (16+ simulation runs simultaneously.)

I have not yet fully examined the results, but looks like both PEST and BeoPEST work fine even technical supports are quite limited.

As I first raised this topic, I still have not got the answer about how to use SCE-UA, etc. global optimization algorithm in PEST instead of the default GML one (it is powerful though)?

http://www.pesthomepage.org/Global_Optimizers.php

[Simon Woodward]

Well done Chang.

Yeah BEOPEST is pretty easy to use once you know how it works. I’ve been using that too. The PEST manual is pretty comprehensive, although not always easy to understand (very dense).

 

 

--
You received this message because you are subscribed to the Google Groups "MODFLOW Users Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to modflow+u...@googlegroups.com.

[Prashanth Khambhammettu]

If you download the latest version of the PEST addendum from pesthomepage.org, you will find a section describing various global optimizers including the SCE algorithm. 

-Prashanth Khambhammettu

--
You received this message because you are subscribed to the Google Groups "MODFLOW Users Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to modflow+u...@googlegroups.com.
To post to this group, send email to mod...@googlegroups.com.
Visit this group at http://groups.google.com/group/modflow.
For more options, visit https://groups.google.com/d/optout.

--

------------------------------------------------------------------------------
It's not who you are underneath, but what you do that defines you.
------------------------------------------------------------------------------

[Far_Naz]

I have followed this topic and I also want to calibrate PRMS, MODFLOW, and finally GSFLOW by PEST.

 I get confused with PEST utilities, I have some question that I hope you could answer them.

1-      As most of the document about pest are so comprehensive. I need a brief document for practical use of PEST with examples. (Do you have a suggestion?)

2-      I realized that For calibration PRMS, I can use  GET_MUL_SERIES_STATVAR black in TSPROC. But I want to calibrate my model in monthly (not daily) time scale. How can I make an instruction file in this case?

3-      For MODFLOW 2005 calibration which utilities are useful? (An example would be so practical)

Thanks ahead for your help.    

[Chang Liao]

Some personal thoughts for your questions:

1: Using PEST based on the manual is not as difficult as you thought once you understand the data flow and logic. But it takes time to use it wisely and efficiently.

2: Some utilities including TSPROC and PAR2PAR are basically helping you handling data (input or output) . These types of work can also be done using other simpler scripts (such as transfer from daily to monthly, and preparing inputs for TSPROC takes time again!). 

3: By far, I could see many improvements in modflow calibration using PEST. Many factors are affecting the result. I believe the model itself is the most important part. And choice of observations parameters or data are also important.

[Chang Liao]

Mr Simon:

May I know what version of beopest were you using? 

Beopest is sweet only when there is no error pop out. But it randomly gave me some error and I have no idea how to fix it even look at the source code.

The compiling was quite successfully and I have already finished some runs. Then I had a new error saying:

"Fortran runtime error, end of record" 

[Simon Woodward]

I would guess it is a MODFLOW error; with a certain parameter set, MODFLOW is not completing its run and so the output file is not created properly.

[Académie Scientifique]

i want to calibrate PRMS model with luca but l'application not work i dont know why . stopet at step 1 pleause help me if you can