Inverse solution

A discussion forum for old Hydrus-2D & Meshgen. Happy Posting!
Download PDF document with all topics (February 8, 2008, 1.9MB):
http://www.pc-progress.com//Downloads/F ... on_H2D.pdf
Post Reply
ebrahimian
Posts: 4
Joined: Thu Dec 02, 2010 7:37 am
Location: Iran

Inverse solution

Post by ebrahimian » Fri Dec 17, 2010 9:18 pm

Jirka,

I want to optimize 6 soil parameters (3 for soil hydraulic and 3 for soil solute transport).

How many parameters can HYDRUS_2D (old version) simultaneously optimize?

In your opinion, is it better to optimize parameters in two parts (first part for soil hydraulic parameters and second part for soil solute transport)?

Thanks again
Hamed

Jirka
Posts: 4716
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Post by Jirka » Fri Dec 17, 2010 9:49 pm

I do not remember what the limit was in the old HYDRUS-2D. Likely 10.

I would do that in three steps. First, I would optimize water flow parameters, then solute transport parameters, and finally both simultaneously.

You can look at the following paper (you can download from my page), in which we deal with the issue whether to optimize similtaneously or sequentially.

Jirka

Šim?nek, J., D. Jacques, J. W. Hopmans, M. Inoue, M. Flury, and M. Th. van Genuchten, Solute Transport During Variably-Saturated Flow - Inverse Methods, In: Methods of Soil Analysis, Part 1, Physical Methods, Chapter 6.6, Eds. J. H. Dane and G. C. Topp, Third edition, SSSA, Madison, WI, 1435-1449, 2002.

ebrahimian
Posts: 4
Joined: Thu Dec 02, 2010 7:37 am
Location: Iran

Post by ebrahimian » Thu Dec 30, 2010 6:43 pm

<b>Happy New Year!</b>

I want to simulate water and nitrate movements in furrow fertigation for maize field. When I am running the model for inverse solution (9 soil hydraulic parameters must be optimized for three soil layers) I got the following message after 3-4 hours.
"ORTHOMIN terminates - too many iterations, time step reduced.
Error when writing to an output file! "

The information about my project is shown below:

tini= 0
tfinal= 168 (hour)
dtInit= 1e-005
dtMin= 1e-006
dtMax = 12
MaxIter = 20
Water content tolerance = 0.001 (for faster run, because each run takes much time)
Pressure head tolerance = 1 cm

Soil depth= 60 cm
Furrow width= 75 cm
Mesh nodes number= 378
Mesh triangles number= 635
Soil materials= 3

Inverse solution:
Max number of iteration= 50
Number of data points in the OF= 18


Time step is really small. What is wrong? Should I use the finer mesh?

Thanks again
<b>Hamed</b>

Jirka
Posts: 4716
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Post by Jirka » Thu Dec 30, 2010 8:12 pm

Hamed,

This likely means that some of the soil hydraulic parameters are unrealistic. Note that during optimization, the parameters are changed by the program to minimize the deviations between measurements and model. You can check the last set of parameters in the I_Check.out file (or similar name). If you see some unrealistic numbers, you should constraint the optimized parameters.

However, note that you should not optimize 9 parameters, when you have only 18 data points. This will very likely lead to nonunique solution. See the following papers (you can download from the webpage with my CV) discussing these issues:

Šim?nek, J., and J. W. Hopmans, Parameter Optimization and Nonlinear Fitting, In: Methods of Soil Analysis, Part 1, Physical Methods, Chapter 1.7, Eds. J. H. Dane and G. C. Topp, Third edition, SSSA, Madison, WI, 139-157, 2002.

Hopmans, J. W., J. Šim?nek, N. Romano, and W. Durner, Inverse Modeling of Transient Water Flow, In: Methods of Soil Analysis, Part 1, Physical Methods, Chapter 3.6.2, Eds. J. H. Dane and G. C. Topp, Third edition, SSSA, Madison, WI, 963-1008, 2002.

Jirka

Post Reply