Model performance

To use environmental models effectively for management and decision making, establishment of an appropriate level of confidence in their performance is vital. The work reviewed techniques available in different fields. 

A five-step procedure for performance evaluation of models was suggested, with the key elements including:

1.  (re)assessment of the model's aim, scale and scope;
2.  characterization of the data for calibration and testing;
3.  visual and other analysis to detect under- or non-modeled behavior and to gain an overview of overall performance;
4. selection of basic performance criteria; and
5. consideration of more advanced methods to handle problems such as systematic divergence between modeled and observed values.

Last edited: 21 May 2014