Dear all,

I have some troubles with inverse simulations of particle transport model (Hydrus 1D). I noticed that the weights coded in the "FIT.in" file were different from the codes in the "Fit.out" file. For instance, data with a type 4 and a given weight of 1 had a weight of 5 in the fit.out file. Can hydrus deal with different weights? If so, how does it deal with it?

In one of the studied colums, all particles were filtered by the column. The aim is to simulate the evolution of the "sorbed front" into the column over time. Therefore, I would like to fit a model by taking into account the "sorbed" amount in the solid phase and a concentration of "0" in the eluent. However, according to the fitting results, I have the impression that the inverse simulation cannot take into account very low concentration as calibration data. Is that possible? Is there a minimum concentration value I should use? Or can I only use the concentration sorbed in the soil in that case?

Thank you!

## Inverse data - transport model

### Re: Inverse data - transport model

Of course, Hydrus can deal with different weights. There are three types of the weighting of inverse data: a) No internal weighting, b) Weighting by mean ratio, and c) Weighting by the standard deviation. This is all described in the manual and the help.

The objective function is defined as the sum of squared deviations. Obviously, small numbers will likely lead to small deviations and thus their weight in the objective function is small. One can either use higher weight for small numbers, or one can use the log transformation of concentrations in the objective function. That would give higher weight to small numbers and deviations. Again, it is described in the help and manual.

J.

The objective function is defined as the sum of squared deviations. Obviously, small numbers will likely lead to small deviations and thus their weight in the objective function is small. One can either use higher weight for small numbers, or one can use the log transformation of concentrations in the objective function. That would give higher weight to small numbers and deviations. Again, it is described in the help and manual.

J.