Two questions on inverse modeling and one on IC

A discussion forum for Hydrus-1D users.
Post Reply
Posts: 8
Joined: Wed Oct 21, 2020 8:32 am
Location: Germany

Two questions on inverse modeling and one on IC

Post by mfd » Wed Sep 15, 2021 11:38 am

Dear Prof. Simunek,

Some questions I couln't find answers to in the manual, tutorials or literature:

I am modeling saturated-unsaturated conditions in a 2m water column in which I am measuring matric potential in three depths.
Thus, some matric potentials are negative, some are positive and for each time step, I have three simultaneous measurements.

Q1: In the data for the inverse model, I have to enter the depth of the measurement. It says "depth (negative)". If a sensor is in 40cm depth, should I put -40 or 40? I tried both but am not sure which is correct from my results.
With "position of observation node" - is that the node number or the depth?

Q2: For all time steps, I would like to enter the three simultaneously measured matric potentials in the three depths. Is this possible? I tried both Type 2 and Type 12, and the inverse model is converging but obviously something is wrong, as I got ok results when I entered matrix potential in only one depth, and with two or three depths the residuals are huge when I try estimate only n or alpha.
The reason I don't just enter GW table is that there are sometimes typical highly negative matric potentials near the surface due to ET, and I thought this would be great information for the modeling.

Q3 unrelated to inverse modeling:
My h(z) initial condition is not always linear, as mentioned, in certain conditions I measure strongly negative matric potentials towards the surface.
I have noticed that the model replicates my time series worse when I enter these measured non-linear initial condition h(z) versus a linear one assuming 0 at measured GWt able, as it always first wants to produce a static equilibrium with a linear h(z) that is not there in the measurements. Is this a typical issue with the model and would you advise to always begin time series with linear h(z), or does this simply point to wrong BCs or SHPs? (such as too large Ks)

Thank you for your time and
kind regards

Posts: 5343
Joined: Sat Mar 16, 2002 3:47 pm
Location: USA
Location: Riverside, CA

Re: Two questions on inverse modeling and one on IC

Post by Jirka » Wed Sep 15, 2021 9:20 pm

Q1: If you use type 12 (for pressure heads) or 13 (for water contents), then it should be -40, e.g.,
-40 0.35 13 iPrint Weight
If you use type 1 (for pressure heads) or 2 (for water contents), then you should give the observation node number, e.g.,
Time 0.35 2 iObs Weight

You should be able to see in the Fit.out file, if the values are correctly interpreted.

Q2: You should use Type 2 and the observation node number. You should look at the example FitTest2 (in the Inverse folder), where pressure heads at three depths are used.

Q3: Below the groundwater table (in the saturated zone), you should have hydrostatic equilibrium.


Post Reply