The model is helpful in elucidating how different processes and properties in the system interact. We are always constrained to investigate a limited part of the whole system with respect to both time and space. The model can be used as a tool to extend our knowledge.
The fundamental physical equations are well known and accepted but we still have to test their validity at different field scales. A general problem is that our knowledge of soil properties normally originates from small soil samples. The role of small soil units compared to larger units is not well understood and we have to find out how we can combine information, which represents different scales. Areal mean values of soil properties such as the hydraulic conductivity are hard to determine even from intensive measurement programmes and it is not certain that the use of an areal mean will be the best choice for the model simulations. The dynamical interaction between the plant and its environment is a newly developed part of the model and is thus continously updated as new experiences are gathered.
One important aspect when testing the model is that parameter values should ideally have been estimated independently of the field measurements, which are used to test the model predictions. In such a case we will learn about how the system behaves even when model predictions fail. On the other hand we will seldom learn about how nature behaves by using calibration procedures even if good agreements between simulated and observed variables are obtained. The estimated parameter values that result in a good agreement must always be compared with other independent estimates if a model application is to have scientific interest.
1) Do not be happy just because the model output is in agreement with observations; try instead to find out why there are no discrepancies.
2) Be happy when the model and the reality are different; then you have a key to new knowledge.
3) The model can provide you with a much better answer to an applied question than is possible with many field investigations. In many cases we cannot wait for the results from long-term field investigations.
4) An adviser using a good mathematical model will certainly be efficient if he/she is successful in combining the results from the model with critical thinking. The model will stimulate an examination of problems if the adviser as well as the scientist gets an opportunity to play with the model.
5) An adviser who believes too much in the figures from a mathematical model will be equally poor as the one who fully trusts results from field investigations.