The Robust Decision Support process for participatory planning

Published: 12th July 2016 15:34Last Updated: 25th July 2016 15:52

Introduction

This description of RDS is taken from the final report for the Rios del Páramo al Valle project, which focused on building climate adaptation capacity in water resources planning. This report can be downloaded here.

The SEI practice of Robust Decision Support (RDS) is based on a theoretical decision making under uncertainty framework referred to as Robust Decision Making (RDM). RDM emerged from a program on strategic decision making under conditions of deep uncertainty within the RAND Corporation (Lempert et al., 2003). The starting point for the RDM framework is that traditional decision making approaches based on an assessment of the likely probabilities of future conditions do not respond well to a situation such as climate change, where there is no consensus about the likelihood of specific climate futures. SEI work with RDS has involved applying RDM theory to the challenge of water and watershed planning and decision making under climate change in a way that responds directly to the IWRM appeal for participatory water and watershed planning, based on a large body of literature (Folke et al., 2005; Pahl-Wostl, 2009; Pahl-Wostl et al., 2007). 

The central feature of the RDS practice is to acknowledge and intentionally incorporate the analysis of external factors such as climate change, but also potentially other factors such as population growth and economic development, into the evaluation of the potential benefits associated with specific water management adaptation actions. While grappling with the uncertainty associated with these external factors, decision makers engage in an iterative process of identifying the actions that can be taken at the watershed scale in order to reduce the climate vulnerability and increase the climate resilience of water systems.

The Method

The steps in the RDS process are shown in Figure 1, below.


Figure 1: RDS steps, timing, and participation levels.
Color coding indicates level of participation according to legend and approximate time for each step is included. Steps are shown in a linear way, but they overlap and can be iterative. [note: this is Figure 2, from page 29 of the Rios del Páramo al Valle report]

The steps of this process fall into two phases, preparation and investigation.

The preparation phase

The preparation phase, which generally takes around 12-24 months to complete, is designed to assure that all relevant stakeholders and decision makers are given the opportunity to participate in the critical problem formulation and analytical design process. The lower end of the timing of about 12 months corresponds to situations where a technical level on water modeling expertise exists among stakeholders. The higher end estimate of timing includes working with stakeholders to build capacity on water systems and management modeling. Specific steps in this phase are as follows.

  1. Identify decision space: Either by being introduced to it by key actors or by conducting a screening analysis of the challenges in a particular geographical or thematic context. Here the decision space means the forums within which watershed actors engage in discussions regarding potentially useful water management adaptations to climate change and other uncertainties, and take decisions to implement the most promising options (Pahl-Wostl, 2009). Level of Participation: Consultation
  2. Actor mapping: Within a decision space identify which actors to include in the negotiation and the deliberation process and the type of information they can provide for the analysis (Reed et al., 2009). Level of Participation: Information extraction
  3. Problem formulation: Whereby all of the key actors identified by the actor mapping participate in describing the decision space via the application of the XLRM problem formulation framework (Lempert et al., 2003). Level of Participation: Participatory research
  4. Model construction: To assemble the analytical tools and information to simulate the system. In the water resources related work described here, this model construction step uses SEI’s Water Evaluation and Planning (WEAP) system. The model constitutes a laboratory for testing possible watershed futures (Groves et al., 2008). Level of Participation: Co-Learning 
  5. Scenario definition quantifies plausible future ranges of the identified uncertainties. In the work reported on here, a key part of the activity is the generation of future climate projections scaled appropriately for evaluating climate change adaptations at the watershed scale (Peterson et al., 2003). Level of Participation: Cooperation

Two items referenced in this description of the RDS Preparation Phase warrant further presentation. The first is the XLRM problem formulation framework. XLRM is a tool developed by the RAND Corporation that divides a decision making process into four components:

  • X (eXogenous factors) represents the uncertain factors outside the direct control of the actors within a particular decision making process but which have the potential to influence outcomes.
  • L (Levers) represents the specific actions that are available to these actors as they seek to improve conditions or outcomes in the face of future uncertainty.
  • R (Relationships) is the suite of analytical tools deployed to capture the exogenous factors and represent the levers identified by the actors, which when deployed produce estimates of...
  • M (Metrics of Performance), which are the means by which individual actors will evaluate the outcomes associated with a specific action considered as part of the decision making process.

The R component of the XLRM framework pertains to the tools used to support the analysis carried out as part of the effort to evaluate the performance specific adaptation actions. These often include models of the watershed/water management system in question.

In the Rios del Páramo al Valle project, the primary model or analytical tool deployed was the Water Evaluation and Planning (WEAP) system which has been developed and supported within SEI for over 25 years. WEAP is an integrated hydrologic/water resources modeling platform that represents both the natural hydrologic or rainfall-runoff processes in a watershed as well as the physical and regulatory systems put in place to balance available supplies and existing demands as part of a multi-objective water allocation system. Over the years WEAP has been expanded to allow for the representation of groundwater hydrology, surface water quality, plant biomass production and many other processes at play within a watershed.

The investigation phase

In the RDS process, once the modeling platform has been constructed and calibrated based on historical climatic and hydrologic data sets, and potential future scenarios have been defined, the process switches to the Investigation Phase. During this phase, which takes approximately 12 months to complete, the models are run for each of several adaptation strategies articulated by the key actors (always including the ‘no action’ option in order to create a baseline for comparison), under each scenario related to future climate and non-climate (e.g. population growth rate, per capita consumption, regional economic development) uncertainties of concern. A set of scenarios produces a large data base of results covering many dimensions of performance (e.g. demand satisfaction, reservoir storage levels, hydropower generation, and ecosystem health), that are explored using innovative data visualization techniques which provide critical inputs to the decision making process. Specific steps in this phase include:

  1. Ensemble analysis: Within which the model constructed is run to combine all future uncertainties and actions, including the case when no action is taken. This case is critical in assessing the baseline vulnerability of the current system in the face of uncertainties. Level of Participation: Cooperation
  2. Output exploration: Uses innovative and interactive data discovery tools to explore the model results. This exploration is carried out in a participatory and dynamic fashion with key actors in the decision space. Level of Participation: Co-Learning
  3. Decision support: Based on the exploration of the outcomes, which are evaluated within the decision space, the performance of specific management actions can be evaluated relative to the no-action baseline and to each other. Upon viewing the results, actors can decide to either reformulate the problem or to accept a particular recommendation for a preferred course of action. Level of Participation: Co-Learning

Where the WEAP tool is employed, the exploration of WEAP outputs is simultaneously the most exciting and the most challenging step in the RDS process. It involves exploring, in close collaboration with watershed actors, the output of multiple model runs covering all combinations of future scenarios and possible adaptation responses, covering several dimensions of performance. The amount of information to digest is substantial and traditional techniques for sharing scientific and technical information with decision makers (maps, X-Y graphs, data tables) are not well suited.

In the Rios del Páramo al Valle project, SEI and its watershed partners worked with a leading edge data exploration and visualization software package called Tableau to produce sophisticated dynamic data visualizations. In testing the RDS method as part of this project, SEI and its partners tried to directly relate the steps in the process to both the connections between the various water and watershed planning instruments mandated by Colombian Law (POMCA, PORH, POT, PSMV) and to the guidance documents pertaining to the formulation of each individual instrument. As such, the project was able to produce results that are feeding directly into national level dialogues pertaining to climate change and water management in Colombia.