Uncertainty

The ideal EM for use in a given situation should provide enough information to identify sources of uncertainty, ideally quantitatively.  The degree of confidence that can be placed in results from a documented application of an EM will be higher if modeling included and documented any of the following procedures, which are all reflected in ESML descriptors:

• Model goodness of fit (tendency of model estimates to fit a given data set)
• Model validation (evaluation of model performance against other data or models). ?
  ESML uses the categories in “Characterising performance of environmental models” by Bennett et al. (ND Bennett, BFW Croke, G Guariso et al. 2013, Environmental Modelling & Software 40:1-20) to classify validation approaches.
• Characterization of predictor or response variable variability
• Sensitivity analysis (assessment of response variable sensitivity to predictor variable variation)
• Uncertainty analysis (evaluation of the sources and magnitude of uncertainty in EM responses)

View the ESML Data Map to see a full listing of ESML descriptors.