Research focus: Quantitative modeling of human interactions with water resources systems

David Gold is an Assistant Professor in the Department of Physical Geography at Utrecht University. David's research advances emerging tools such as high-performance computing, multiobjective optimization, and artificial intelligence to understand and improve decision-making in water resources contexts.  David's current research projects include exploring drought vulnerability in the Upper Colorado River Basin, developing adaptive strategies to aid urban water supply planning in the Southeastern United States, and exploring pathways for sustainable and equitable water supply infrastructure investment in the Federal District of Brazil.

Ongoing Research: Visual analytics to support climate adaptation pathways

Funded by a Pathways to Sustainability Seed Grant, our team is developing a generalizable framework that uses Visual Analytics (VA) and eXplainable Artificial Intelligence (XAI) to facilitate the creation of multisectoral climate adaptation pathways. Water resources managers around the world are confronted by conditions of Deep Uncertainty – when decision-makers do not know or cannot agree upon probability distributions of key system inputs, outcomes of interest, and/or system boundaries. Under deep uncertainty, adaptive planning paradigms help planners develop robust adaptation pathways that avoid costly “lock-ins” by creating strategic visions for the future,  committing to minimum regret short-term actions, and discovering adaptation tipping points to guide future decisions. For example, recent work in the Rhine-Meuse Delta in the Netherlands used an adaptive planning framework to reveal how near-term flood protection decisions influence long-term regional risk, identify trade-offs between alternative strategies, and illustrate how careful planning can provide cost-effective means of maintaining safety standards for sea level rise scenarios up to 3 meters.

In recent years, tools from AI, including multi-objective evolutionary algorithms (MOEAs) and reinforcement learning (RL), have been shown to generate climate adaptation policy pathways that outperform policies developed solely by human decision-makers in stylized test problems. AI-supported methods can develop flexible and robust adaptation pathways by searching through millions of policy alternatives and incorporating large amounts of climate information into state-aware rule systems that guide adaption to changing climatic conditions. However, the complexity of policies generated by AI-assisted tools and the lack of transparency in how AI-assisted policies trigger actions have led to policymakers' distrust of these tools and created a barrier to implementing these strategies in real-world systems. This challenge mirrors difficulties in applying AI advances to other fields, such as healthcare and finance.

Our VA system employs a coupled MOEA/RL framework to discover adaptation pathways for a Delta System that balances multisectoral objectives. Using innovations from XAI, the system allows users to dynamically interact with the algorithm as it searches through pathway alternatives, and explore how the algorithm uses climate information to trigger adaptive action. We are building the system using an extensible design to allow for future integration with emerging visualization advancements such as mixed reality (e.g., Microsoft HoloLens) that provide users with immersive interactions with climate adaptation pathways policies. 

New Research project: Global potential for direct potable water reuse to alleviate water scarcity

Roughly two billion people worldwide lack access to safe drinking water. In water scarcity hotspots around the globe, non-renewable groundwater storage is being depleted, and surface water resources are overallocated (https://worldwatermap.nationalgeographic.org). This has led to an unsustainable situation, where there is a large water gap between the demand and availability of freshwater. Direct potable reuse (DPR) – introducing highly purified wastewater into urban water supply systems – is one promising solution strategy to reduce this water gap in urban areas. This project explores the potential of DPR systems to mitigate water scarcity at global, national, and regional scales.

As part of the National Geographic Society - World Water Map project (https://worldwatermap.nationalgeographic.org), our team is working with an international team of researchers from Utrecht university and the Joint Global Change Research Institute. Using the Global Change Analysis Model (GCAM) to measure the multisector implications of DPR under a range of socioeconomic and climate change scenarios, our work explores the impacts of DPR on the energy, urban, and agricultural sectors and map regional teleconnections resulting from DPR investments. 
 

Selected Publications

Gold, D. F., Reed, P. M., Gorelick, D. E., & Characklis, G. W. (2023). Advancing regional water supply management and infrastructure investment pathways that are equitable, robust, adaptive, and cooperatively stable. Water Resources Research, 59(9), e2022WR033671.

Gold, D. F., Reed, P. M., Gorelick, D. E., & Characklis, G. W. (2022). Power and pathways: Exploring robustness, cooperative stability, and power relationships in regional infrastructure investment and water supply management portfolio pathways. Earth's Future, 10(2), e2021EF002472.

Gold, D. F., Reed, P. M., Trindade, B. C., & Characklis, G. W. (2019). Identifying actionable compromises: Navigating multi-city robustness conflicts to discover cooperative safe operating spaces for regional water supply portfolios. Water Resources Research, 55(11), 9024-9050.

Trindade, B. C., Gold, D. F., Reed, P. M., Zeff, H. B., & Characklis, G. W. (2020). Water pathways: An open source stochastic simulation system for integrated water supply portfolio management and infrastructure investment planning. Environmental Modelling & Software, 132, 104772.

Hadjimichael, A., Gold, D., Hadka, D., & Reed, P. (2020). Rhodium: Python library for many-objective robust decision making and exploratory modeling. Journal of Open Research Software, 8.