For a few decades now, decision-makers in water resources management have been able to use Decision Support Systems (DSS) to evaluate solutions to complex problems related to water supplies and the environment (Labadie et al. 1989; Loucks and da Costa 1991).
For this purpose, several computer models have been developed, such as AQUATOOL, AQUATOR, MIKE-BASIN, RIBASIM, WARGI and WEAP. In these kinds of program, systems are represented by a set of nodes (water supplies, demands, junctions, aquifers…) connected by links (rivers, channels…), and a time-extended simulation can be executed to obtain the results based on system performance criteria (Loucks and van Beek 2017).
In complex water resources systems, these tools are essential in order to make decisions based on knowledge, for example, to choose between building new infrastructure or increasing the efficiency of existing systems to control demand, or simply to determine whether certain demands can be satisfied in the long term. Planning requires information, and these DSS are an important source of information (Loucks 1992).
In general, results are obtained by applying mathematical optimisation techniques (nonlinear optimisation models, dynamic programming, linear programming, artificial neural networks, genetic algorithms…), trying to satisfy the demand with objectives based on economic or priority criteria (Loucks and van Beek 2017). In each time interval calculated, the simulation results are typically flow series in links and water inputs/outputs in nodes (or storage levels in reservoirs and aquifers). From these results, it is usual to calculate performance parameters, to evaluate the degree to which the water demand is satisfied (Andreu et al. 1996).
These calculations must be performed on fixed data, because the algorithms are deterministic: a particular set of data produces a particular set of results, and any change in the data would lead to different results. But we usually try to study the future behaviour of a water resources system, and future values of some data are difficult to determine. In particular, natural water supplies must be estimated taking into account the climate change and the future basin evapotranspiration (linked to land uses and to climate). And future water uses are uncertain too: for example, future irrigation water demand will depend on crop patterns, irrigated surfaces, climate (rain and temperatures), irrigation systems efficiencies and socioeconomic factors.
In short, the results are sensitive to uncertain data, so that a "static analysis” (a single-scenario simulation) may be insufficient, also taking into account the transcendence of the decisions made on the basis of these models.
This work aims to show the importance of “dynamic analysis” (multi-scenario simulations) to better understand the water resources systems and to determine their future resilience under uncertain scenarios. To do that, a study case is shown where different scenarios are calculated, with different natural water supplies and different demands, to graphically visualise the variation of the results under different scenarios, with the aim of improving the decision support offered by these models.