Information Technology Reference
In-Depth Information
5. An evolving system: Uncertainty, DSS and adaptive management
Living in a changing world, it is evident that even if planning and management are
implemented as particular actions, they are an ongoing process over the long-term.
Consequently, Integrated Water Resources Management is portrayed as a spiral where the
implementation of past plans is monitored and the process is re-evaluated and re-directed
based upon our most current, new information. In other words, we have to plan for an
uncertain future, then deal with it when it becomes the present, and learn from it when it
becomes the past. Such an acknowledgement is the basis of adaptive management.
Everyone knows the future is uncertain, but how do Decision Support System Models deal
with uncertainty? To what extent and how is uncertainty incorporated into DSS and how is
it communicated? The truth is that uncertainty is a difficult concept to work with and is
often not well represented in models and decision support tools. Many systems dynamics
models state as a disclaimer that the specific values provided by the model are to be
interpreted as a relative measure in comparison to other alternatives, but never as absolute
numbers. This is well accepted because it still allows the comparison of different
management alternatives and an overall view of their impacts in the entire system. While
uncertainty can be accounted for in specific model components (physical land surface and
hydrologic models) once the intention to do so is there, it may be harder to represent it
accurately in systems dynamics models, perhaps due to the inability to accurately represent
and blend uncertainties from many different model components of the system (i.e.
behavioral and socio-economic components).
There are many sources of uncertainty in simulations: uncertainty contained in the input
data (climate change projections), in the model structure formulation (recharge, runoff and
evaporation transformations), and arising from issues related to boundaries and scales (e.g.,
regionalizing soil parameters).
Uncertainty inherent to structural representations of the physical world reflects the lack of
proper understanding of physical processes or our inability to represent them properly,
much less crossing boundaries of scale. As an example, in basins in Arizona that constitute
some of the most instrumented and studied watersheds in the world, the quantification and
the spatio-temporal characterization of natural recharge into the regional aquifer remains a
formidable challenge. The estimates currently used in hydrologic models are based on
empirical relationships aggregated at the basin scale that were developed 20 years ago
(Anderson, 1992).
When developing a DSS model, different sources of uncertainty can be represented in
different ways. During a collaborative process, stakeholders and decision-makers can decide
on what sources and measures of uncertainty need to be explicitly represented in the model
and which ones may better be addressed through other means. For example, climate change
projections are very uncertain but a multi-model envelope of uncertainty can easily be
represented using the wettest and driest models (or hottest and coldest) as the extreme
cases, and assuming that future rainfall (or temperature) will fall somewhere in between
these extreme cases. All the projections of climate models falling within the wettest and
driest models can be averaged, providing what can be used as the highest-likelihood
possibility (Hagedorn et al. 2005). Such envelopes of uncertainty in inputs that drive land-
surface and hydrologic models can easily be propagated or transmitted from the input
variables to the output variables (Serrat-Capdevila et al. 2007). On the other hand, there are
uncertainties regarding issues that are difficult to quantify but still have important impacts
Search WWH ::
Custom Search