simple-example.1

Test case 3
Directory: simple-example.1
Batch file: dupuit.bat
Process-model data: dupuit.in.tpl
Comment: All files are included in one directory. Only regression automated
analyses are performed. New users requested a very simple exmaple.
All input is in ucode_dupuit.in. The model is dupuit.exe and solves
for head in a system with a horizontal no-flow lower boundary, fixed
head on the left and right, and a recahrge flux on the top boundary.
This is presented in most beginning ground water books and is
illustrated with dupuit.htm in: simple-example.1/html/dupuit.htm.

ucode_dupuit.bat runs the regression based on input in ucode_dupuit.in

ucode_dupuit.in directs ucode to substitute parameter values in
dupuit.in.tpl, execute dupuit.bat (which runs the dupuit code in the
code_dupuit folder and generates output dupuit.out in the folder
simple-example.1), then based on the instructions in ucode_dupuit.ins
ucode extracts the results of the simulation from dupuit.out and uses
those values to calculate residuals and sensitivities, ucode
iteratively uses residuals and sensitivities to update the parameters
and repeats until the parameters change less than tolpar. If ucode is
successful, default residual analyses are executed from
ucode_dupuit.bat

The observation data provided include a small amount of noise.
Values below are in meters and days
The true heads and flow at the left and right boundary:
heads from left to right at
5m, 15m, 25m, 35m, 45m are:
5.50475 5.89724 5.81146 5.22393 3.91489
with noise as entered for observations in the regression they are:
5.505 5.897 5.812 5.223 3.915
flow at the left boundary is:
-0.00498
with noise as entered in the regression:
-0.0050
flow at the right boundary is:
0.00872
with noise as entered in the regression:
0.0088

New users may wish to experiment with alternative observations which
can be generated by running the forward model with altered input in
dupuit.in and getting the results from dupuit.out.

New users may also experiment with estimating other parameters, such as
the heads at the boundaries in various combinations with K and w.

One might also experiment with various settings in the ucode input.

Notice that if the flow is not included as an onservation in the
regression then only the ratio of K and W can be estimated, not
their absolute values because the parameters are completely correlated
when only heads are used as observations.