Parameters and lookup tables can be linked to simulation results. This way a large model can be split up into several smaller models.

To link a parameter to a simulation result, first edit the parameter and go to the Values tab. In the table, make sure that all parameter properties are shown by clicking the button in the lower left corner of the table.

Locate the Link field in the end of the table, and click the field. A button appears named Select. Click the button, and choose a simulation result from the window that appears.

To link a lookup table, simply click the link button located over the data table in the Values tab. The process is the same as for a parameter.

The big advantage with using linked results, compared with using normal import/export from/to Excel, is that linking can be done to probabilistic results.

Note that parameters that are linked to probabilistic results, must be selected for probabilistic analysis (see probabilistic settings). Likewise, parameters linked to deterministic results cannot be selected for probabilistic analysis.

Consider a model which first calculates the release of a chemical into the atmosphere. Next, a large, complex and time consuming model calculates the transport of said chemical from the release point to another area. Finally, a third model computes the uptake into local flora. These models require different expertise - it is likely that different people will develop different parts of the model.

Instead of creating one large model, three separate models are developed. Each model is given lookup tables and parameters for the input into the model.

When the time has come to run simulations, the first model is run probabilistically. The outflow results and all relevant parameter samples are saved as an archive and exported to a binary file.

This file is imported into the project for the transport model, and are linked to its input parameters and lookup tables. The transport model is now run probabilistically - a simulation that takes several days - and the output data and parameter samples are saved.

Finally, the developer of the final model can import and link with these results. When the final simulation is finished, a sensitivity analysis can be performed which includes the parameters related to both the release and to the atmospheric transport.

The advantages are many:

• Model developers can work (more or less) independently on their own models.
• When a model late in the chain is run, and errors are discovered, the whole chain does not need to be re-run.
• Running smaller models is generally faster than running one large model.

• Parameter
• Lookup table
• Results window