Water resources management is a very complex task. Building a model of the system to support decision-making and planning requires representing both the supply and the demand side, each of which includes multiple elements.
On the supply side, there is the hydrology of the basin, as well as the potential impact of future climate change on it – such as the effects of warmer temperatures, changing rainfall patterns, and/or melting glaciers.
There can also be many competing demands for the water: for household use, industry, irrigation and energy production, for instance, as well as streamflow requirements to maintain healthy ecosystems and fish habitats. And infrastructure such as dams, reservoirs, wells and irrigation systems can reshape the system in key ways.
Scientists have developed ways to capture all these elements in integrated models, such as those produced by SEI’s Water Evaluation and Planning (WEAP) system. But a key challenge in working with such complex models is that it is very difficult to present the results to stakeholders. This not only hinders the models’ usefulness to decision-makers, but also reduces the quality of the models themselves, as stakeholders cannot easily provide feedback, based on their own knowledge and experience, to refine and improve the models.
At SEI, we are using data visualizations to help address this problem. The human mind is highly visual, so when people see information in graphic form, instead of just talking about it or reading numbers, they can recognize patterns.
Data visualizations have proven to be very useful for communicating large amounts of information. At SEI, we use Tableau software to distil the results of complex water system modeling exercises – in particular, the results of large-scale model runs that consider hundreds (or thousands) of scenarios at once.
We first used Tableau in Bolivia, in a project focused on water supply for the cities of La Paz and El Alto. We had developed a participatory approach to water resources planning, using the Rand’s Corporation XLRM framework: X for exogenous factors or uncertainties, L for policy levers or strategies, R for relationships, and M for performance metrics. XLRM is very useful for planning under deep uncertainty, but all the pieces make for a very complex model. The visualizations made it easier to discuss the different pieces with stakeholders and understand how they fit together.
Time series graphic was used to show the month in which the system fails to supply water for the city of La Paz under various climate projections. Climate models estimate a temperature rise of 2°C by 2050, leading to the dramatic shrinkage or disappearance of many small glaciers. The graph shows how by 2030 under the current state of the system, it cannot supply water to an increasing population.
The main goal of the Yuba River case study was integrated regional water resources management planning. Yuba River is a focal point of competing demands for water: hydropower, irrigation, municipal supply, flood control, recreation, fisheries, wetland habitat, and water transfers outside of the Yuba basin. The graphic shows time series of end of May storage for Bullard bar reservoir (1 of 14 performance metrics). The grey bars in the back indicate failure to perform to the desired levels.
Fused data in the Bolivia case study were represented as a scatter plot that shows the reliability values for the two cities, La Paz and El Alto, under the various climate and strategies considered. Other non-climate uncertainties can be evaluated by using the filters on the right.
A grid graphic was developed showing the percentage of times that a particular performance metric fails under the scenarios considered. The rows represents the combination of scenarios of the uncertainties. The columns corresponds to the performance metrics identified. The menu at right shows interactive input boxes for the incorporation of desired levels for each performance metric. Those values act as the thresholds to calculate system failure at the relevant time step. Cell colours in shades of red corresponds to vulnerability values above 50% and in shades of green, corresponds to values below 50%.
The final decision space visualization in the Bolivia case study was represented by grid graphic of reliability levels. The row blocks are the three performance metrics considered: demand satisfaction for La Paz, El Alto, and the evaluation of potential impacts to agricultural production. Nested within those rows are the climate projections. The columns corresponds to the strategies. Nested within the strategies are the non-climate uncertainties. Desired levels of reliability and demand satisfaction (coverage) are located on the right-hand menu. Orange represents failure to obtain desired reliability levels and blue represents above desired reliability levels.
The final decision space visualization for the Yuba case was grid graphic showing as column blocks the strategies and nested within the strategies are the performance metrics. The grid contains the difference in percentage points of vulnerability values when a strategy is evaluated. Green represents negative values, i.e. a reduction in vulnerability, and red shows positive values, i.e. an increase in vulnerability. Because of our ability to recognize patterns, we can quickly see which strategies have an integrated or comprehensive impact in the system.
A comet graph was use to evaluate the initial values of vulnerability and its corresponding changes. The graphic here shows the ecological flood plain flows starting point of vulnerability. We can see that while in the previous graphic it showed shades of pink, the initial values were low, and even though we see an increase in vulnerability it is still below the 50% value. Only large-scale projects like the ones shown on the last two columns of the previous graphic could produce multi-sector impacts.
In the Bolivia case study the main metric of performance was the water supply reliability of La Paz/El Alto. Agricultural water supply was also evaluated to make sure they can obtain reliable levels. The XLRM approach and the visualizations worked so well that adapted them for a very different setting, Yuba County, California. That was an even more complex situation, as there were many objectives with even more performance metrics, so we had to develop a wider range of visualizations.
The most important part of the visualizations is seeing how people engage with them, since they are interactive. Stakeholders at the table have similar but different conceptual model of the basin based on their expertise and type of work, so each brings something different to the discussion.
In these types of water systems, the optimal solution for everybody is no longer an option – there is simply not enough water. That reality becomes evident when looking at the future and the climate change projections. Compromises are needed, and discussions of “what I can live with” under the worst conditions are very important. They shed light on how integrated water resources planning needs to be collaborative and inclusive, taking a bottom-up approach.
Different types of visualizations meet different needs in a project, but collectively, they create a visual image of what we call the “decision space”. They allow the development of a shared mental model or conceptual representation of the basin. They also allow stakeholders to “play” with the interactive features, to evaluate the implications of different choices for vulnerability and overall performance.
Decision space visualization is a powerful way to foster effective and meaningful communication, knowledge exchange, and interactive learning. As the body of knowledge about participatory techniques continues to grow, we will continue to refine and improve our approach. Our experience to date suggests they can lead to better, more integrated, comprehensive and inclusive solutions to water scarcity issues.
