PESTPP-SEN Control variables

[Felipeisu]

Dear developers.
I am currently working on my research project at the Leibniz University of Hannover in Germany. I would like to use PESTPP-SEN for my sensitivity analysis. I am not quite sure how to set up the PESTPP-SEN control variables in the enhanced control file format. Would it be possible to receive an example of an enhanced control file containing the Morris Method control variables?

I thank you in advance.

Kind regards,
Felipe Silva

[jtwhite79]

You can see a pestpp-sen enhanced control file here:

https://github.com/usgs/pestpp/blob/master/benchmarks/mf6_freyberg/template/freyberg6_run_sen.pst

As you can see, its using all default arguments for the Morris Method (so there aren't any pestpp-sen specific arguments in the "* control data keyword" section). If you want to change one or more of the default argument values, you need to add new lines in the "* control data keyword" section, one per argument, and the format is key-whitespace-value. So to change the number of samples per parameters (the morris "R" value) from 4 (the default) to 6, you would add

gsa_morris_r 6

somewhere between lines 3 and 11 in that control file (the order doesnt matter). The Morris control variables are described here:

https://github.com/usgs/pestpp/blob/master/documentation/pestpp_users_manual.md#723-control-variables

[Felipeisu]

thank you very much for your reply.

[Felipeisu]

hello,

I would like to ask a question regarding PESTPP-SEN.

I have 33 parameters, 11 parameters for each soil profile (soil column divided in 3 parts).
I want to do the sensitivity analysis separately for each of the strata.

in control file *parameter groups I have put each of the 3 zones.
in *parameter data I have the question with the PARGP variables. Here I put the name of the group associated to each PARNME. Is this procedure correct?
in the PARTIED variable I have to rewrite the same variables as in PARGP?

in the control variable of PESTPP-SEN
tie_by_group(true) and enforce_tied_bounds(true) is it correct?

I would be very grateful for your help.

Thank you very much
Kind regards,
Felipe Silva

[jtwhite79]

Doing a sens analysis with only 3 adjustable parameters seems like it might be too under parameterized so that the results might not be very meaningful. For Morris, with 33 parameters and 4 steps per parameter (the default), it will take about 135ish runs. Can you not just do the full compliment of parameters? How long does the model take to run?

[Felipeisu]

Hello
Thank you for your prompt reply.
The thing is that I have already done the analysis of the 33 parameters with Morris and it turned out well.

So now I want to do it again but using the Morris method that works with groups of parameters.
I explain it like this:

I have 33 parameters in total corresponding to 11 parameters for a Peat soil type, 11 parameters for a Mineral soil type and 11 parameters for a Bed Rock soil. The outputs of my model are temperature and humidity.
I would like to know how to set up the control file to get sensitivity results for my parameters but with respect to the soil type, not as I have already done in general.
As far as I understand the Morris method with group of parameters, it is a matter of varying by modeling all the parameters at the same time but of the same group, which would be in my case : 11 parameters per soil type.

How could I configure the control file to achieve this?
thank you very much
Felipe Silva