babyֱ��app

Smith, Rebecca ¹ ; Kasprzyk, Joseph ²

¹ University of babyֱ��app at Boulder
² University of babyֱ��app at Boulder

In light of the unpredictable effects of climate change and population shifts, responsible resource management will require new types of information and strategies going forward. For water utilities, this means that water supply infrastructure systems must be expanded and/or managed for potential decreases in overall supply, increases in extremes, and a vulnerable environment. Multiobjective Evolutionary Algorithms (MOEAs) are tools that efficiently generate and evaluate planning alternatives to find a large number of portfolios that balance between water utilities’ conflicting performance objectives. Research applications have shown MOEAs to be useful for finding innovative approaches to challenging management problems. However, they have rarely been used in practice and their role in the context of real-world planning has not been fully established.

Our study bridges the gap between research applications of MOEA-assisted optimization and the needs of utilities by collaborating with six Front Range water providers to develop and evaluate results from an MOEA case study. At the second of two practitioner workshops, we engaged with water managers over structured activities and facilitated discussions to create a new understanding of how the tool can both support their current needs and expand their ability to holistically manage the built and natural systems we depend on. This presentation will focus on one result of this collaborative effort- the use of Multivariate Regression Trees (MRTs) to extract valuable information from the many optimized portfolios produced by MOEAs. MRTs can be used to, among other things, reveal subsets of management decisions (out of the many possible options) that lead to good system performance across multiple future scenarios. This type of information would provide a quantitative basis for utilities to establish “low regrets” management strategies.