Geoscience Reference
In-Depth Information
7
Parameter Estimation and
Predictive Uncertainty
Far better an approximate answer to the right question, which is often vague, than an exact
answer to the wrong question, which can always be made precise.
John W. Tukey, 1962
The weight of evidence for an extraordinary claim must be proportional to its strangeness.
Pierre Simon Laplace (1749-1827)
On the basis of the quantity and quality of information that is available today, equifinality
(from practical or mathematical perspectives) cannot be totally eliminated from hydrologic-
response simulation for any model, simple or complex. One promising protocol for reducing
parameter uncertainty for physics-based simulation is an improved dialogue between ex-
perimentalists and modelers ...,sothat future field experiments and long-term observations
will better capture the non-intuitive nuances associated with the distributed response for
real systems.
Brian Ebel and Keith Loague, 2006
7.1 Model Calibration or Conditioning
It should be clear from the preceding chapters that limitations of both model structures and the data
available on parameter values, initial conditions and boundary conditions, will generally make it difficult
to apply a hydrological model (of whatever type) without some form of calibration. In a very few cases
reported in the literature, models have been applied and tested using only parameter values measured or
estimated
a priori
(e.g. Beven
et al.
, 1984; Parkin
et al.
, 1996; Bathurst
et al.
, 2004); in the vast majority
of cases, the parameter values are adjusted to get a better fit to some observed data. This is the model
calibration problem discussed in Section 1.8. The question of how to assess whether one model or set of
