Problems from inverse solution by using field soil moisture data

A discussion forum for old Hydrus-1D users.
Post Reply
Posts: 1
Joined: Mon Oct 17, 2022 1:35 am
Location: Australia

Problems from inverse solution by using field soil moisture data

Post by Chenting » Mon Oct 17, 2022 10:34 pm

Hello everyone,

I am using field soil moisture and actual weather data to estimate five van Genuchten parameters by inverse solution. I did three layers and two layers and got similar problems.

In the two layers of soil:
The soil profile is 23cm in depth, including two soil horizontals (0-7cm A11 and 7-23cm A12). The soil textures are sandy loam. The field data includes 100 days of soil moisture in 5 and 15-cm depths. Based on the information, I set up two materials and a 23cm soil profile with time-variable boundary conditions in the van Genuchten-Mualem model. The water flow boundary conditions are atmospheric BC with the surface layer as the upper boundary condition, free drainage as the lower boundary condition, and calculated pressure head as the initial condition.

However, if the initial parameters are set up as Sandy loam, the numerical solution cannot be completed as it does not converge. Therefore, I usually modify the lower and upper limits of the tension interval in iteration criteria to 0.001 and 1000, which leads to the calculation being completed successfully. Sometimes, even though these two values have been changed a lot, the program still crashes due to not converging.
Another interesting situation is if the initial parameters are set up as silt loam, the program can calculate successfully without any criteria change. This seems not reasonable.

In the three layers of soil, similar problems came again.

I would appreciate it if anyone could give me some suggestions to avoid the programming frequently crashing and stably obtain the results.

Post Reply