Geology Reference
In-Depth Information
Fig. 2.7 Hypothetical
relation of data interval on
modeling error
Data interval for modeling: The case studies in this topic also address a mostly
ignored area in data-based hydrological modeling
data time interval for models.
Modern data collection and telecommunication technologies can provide us with
very high resolution data with extremely
—
fine sampling intervals. We hypothesized
that both too large and too small time intervals were detrimental to a model
s per-
formance, which has been illustrated in this topic with particular reference to rainfall-
runoff modeling. The data time interval is a major factor affecting forecast perfor-
mances of node-based models, particularly neural network models. The performance
of neural network models is highly time-dependent [
10
]. Very large and small data
time intervals could have negative effects on modeling results. The hypothetical
condition for the effect of data time interval on modeling is shown in Fig.
2.7
.
'
2.5.4 Flexibility for a Model
The
flexibility of a model increases as the number of parameters goes up. However,
the modeler should be careful in increasing the
flexibility of a model by addition of
extra parameters, and in most cases it may cause irrationality. Flexibility of a model
is dependent on the assumptions and rules employed during its development. The
hypothesis is that less complex models are less
flexible because of adoption of
tough restricting assumptions to reduce the parameters. In general, complex models
have minimal assumptions, and thus they are more
flexible and applicable to a wide
range of scenarios. Because of this
flexibility factor, applications of less complex
models are limited only to situations where the assumptions are valid. The
flexi-
bility of a model is invariably accompanied by extensibility of the mathematical
code which is determined by the level of complexity of the model. In data-based
modeling, the modeler should always be aware of
five basic aspects before mod-
eling: (1) has the selected model suf
cient rigor to represent the process; (2) ade-
quacy of the selected model for simulation of hydrological processes; (3)
flexibility
