Resilient water infrastructure partnerships in institutionally complex systems face challenging supply and financial risk tradeoffs

doi:10.21203/rs.3.rs-3143783/v1

Download PDF

Article

Resilient water infrastructure partnerships in institutionally complex systems face challenging supply and financial risk tradeoffs

https://doi.org/10.21203/rs.3.rs-3143783/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 27 Aug, 2024

Read the published version in Nature Communications →

Version 1

posted

You are reading this latest preprint version

As regions around the world invest billions in new infrastructure to overcome increasing water scarcity, better guidance is needed to facilitate cooperative planning and investment in institutionally complex and interconnected water supply systems. This work combines detailed water resource system ensemble modeling with multiobjective intelligent search to explore highly uncertain infrastructure investment partnership design tradeoffs in the context of ongoing canal rehabilitation and groundwater banking in California. We highlight that severe tradeoffs can emerge between conflicting goals related to water supply deliveries, partnership size, and the underlying financial risks associated with cooperative infrastructure investments. We demonstrate how hydroclimatic variability and institutional complexity create significant uncertainty in realized water supply benefits and the potential for strong heterogeneity in partners’ financial risks that threaten infrastructure investment partnership viability. Importantly, the partnerships discovered by our multiobjective intelligent search framework are found to deliver substantially higher water supply benefits with a fraction of the financial risk compared to status quo planning processes. This work has important implications globally for efforts to use cooperative infrastructure investments to enhance the resilience and financial stability of water supply systems confronting increasing competition for scarce resources.

Scientific community and society/Water resources

Scientific community and society/Agriculture

Scientific community and society/Business and industry

Water scarcity impacts are increasingly frequent and severe in many parts of the world due to climate change, economic growth, and regulatory change^1–4. Many regions are investing heavily in new infrastructure such as reservoirs, canals, and groundwater recharge facilities to bolster their water supply reliability and avoid negative impacts on public health, food security, power supply, and ecosystems⁵. In the U.S., federal and state funding for water systems has declined in recent decades so that the large majority of water infrastructure spending comes from the local level^6,7. Much of this funding comes from municipal bonds issued by local governments and water utilities, which provide up-front capital for projects in return for the principle plus interest to be repaid over many years. When investing in new infrastructure, water providers must be careful to balance the benefits of supply investments with their associated long-term debt obligations that must be repaid using water sales revenue^8,9. This financial risk is especially important in light of concerns over the credit-worthiness of drought-exposed water providers, concerns that can lead to higher borrowing rates and contribute to growing water affordability challenges in many regions^10–14.

Collaborative partnerships, water supply regionalization, and consolidation have garnered attention for facilitating investment in large-scale infrastructure. Multiple water providers collaborating to finance, build, and operate shared facilities with joint benefits can be more cost effective than independent provisioning due to economies of scale, reduced redundancy, improved access to capital, and diversification of supply and demand patterns^15–20. Regions across the U.S., from California to Kentucky, have encouraged cooperative regional infrastructure investments with funding programs and technical assistance^21,22.

However, collaborative infrastructure investments also carry risks. Modern water supply systems are governed by a complex set of hydrologic, infrastructural, and institutional drivers and uncertainties^23–25. This presents a significant modeling challenge for water providers trying to estimate the expected benefits and costs of expensive infrastructure investments. For example, understanding the value of groundwater recharge facilities requires detailed modeling of short-timescale dynamics (e.g., atmospheric rivers or snowmelt-driven flood events), long-timescale trends (e.g., climate change and aquifer depletion), and time-evolving infrastructure constraints, water rights, and regulations^26–29. Moreover, water supply systems in many regions are increasingly interconnected, which can lead to complex emergent behaviors within the coupled network because each water providers’ future supply reliability depends not only on its own actions and capabilities, but also on the actions and capabilities of other water providers in the network^30–32. Collaboration amplifies this challenge because it requires understanding not only the total aggregate benefits and costs of an infrastructure investment, but also the uncertain distribution of these benefits and costs to individual partners over time^33–35.

To date, little work has been done to help water providers understand the local-scale benefits and risks of investing in regional infrastructure partnerships. Water supply planning models are often highly aggregated, with many water providers grouped together, which limits their ability to discover local-scale tradeoffs for individual partners^23,33. Moreover, little guidance exists to facilitate the design of resilient and equitable water supply infrastructure partnerships in institutionally complex systems. Existing economic planning frameworks for partnership design and cost apportionment are generally based on expected value benefit-cost analysis and/or game theoretic methods^36–40. These methods are not well suited to realistic systems characterized by a large number of potential partners, interactive network effects, high uncertainty, and multiple conflicting objectives. In practice, infrastructure partnerships tend to emerge through pre-existing relationships or ad hoc arrangements. However, as we show in this study, this can lead to significant regret in complex water systems where partnership performance is challenging to predict.

In this work, we highlight the substantial water supply tradeoffs and financial risks that must be navigated by water providers when designing regional infrastructure partnerships. To do this, we develop a novel approach that uses detailed ensemble modeling of an interconnected water supply network under uncertainty, combined with multiobjective intelligent search to discover the optimal tradeoffs in infrastructure partnership design (Supplementary Fig. S1). We apply this approach in the context of ongoing canal expansion and groundwater bank development efforts in the San Joaquin Valley region of California, representative of the state’s efforts to encourage innovative partnerships through its Water Resilience Portfolio Initiative²¹. California is poised to spend billions of dollars on collaborative water infrastructure investments in the coming years^7,41–43. We find a wide diversity of alternative infrastructure partnerships that exhibit strong tradeoffs between the size of a partnership and partner-level financial risk, and between water supply benefits for the investing partners and the rest of the region. Our findings also emphasize the importance of accounting for the substantial uncertainty in infrastructure performance that emerges due to California’s extreme hydroclimatic variability. Uncertainty in overall infrastructure performance is accompanied by high degrees of heterogeneity in the water supply benefits and financial risks experienced by individual water providers under different partnerships. Lastly, we demonstrate the advantages of our multiobjective intelligent search framework through a baseline comparison focused on the ongoing efforts to rehabilitate the Friant-Kern Canal. As an example, one of the candidate partnership designs discovered using our intelligent search procedure achieves 67% higher water supply gains while also significantly lowering the risk of extreme cost burdens for investing water providers when compared to the existing status quo canal expansion partnership. These insights are only attainable by moving beyond traditional ad hoc planning practices, expected value benefit-cost frameworks that do not resolve key differences in partners’ individual water supply and risk tradeoffs, and highly aggregated regional water supply models that fail to resolve the institutional complexities and local-scale dynamics that drive consequential differences in partners’ investment benefits and risks. Our results have important policy implications for California and broader insights for other regions working to develop collaborative investment partnerships to increase the resilience of their water supply systems.

Water supply infrastructure partnership design

The San Joaquin Valley region in California is home to over 4 million people and over 5 million acres of irrigated farmland⁴⁴. The region’s long-standing water supply challenges have been amplified in recent years as climate change has made drought more frequent and severe, and as chronic groundwater overdraft has caused dry wells, widespread subsidence, water quality issues, and restrictive new regulations^21,44–46. Local and state agencies are working to address these challenges by investing in new infrastructure and cooperative water management strategies^20,26,42,47. In this study, we analyze two key ongoing infrastructure initiatives in the Tulare Lake Basin region of the San Joaquin Valley (Fig. 1). First, the Friant-Kern Canal is an important structure that primarily conveys San Joaquin River water south from Millerton Lake as part of the federal Central Valley Project. However, the capacity of the canal has been reduced by up to 60% due to subsidence-related damage^48,49. In 2021, a collection of water providers known as the Friant Contractors decided to invest roughly $50 million to rehabilitate the canal and expand its capacity, with various federal and state programs providing support for the rest of the estimated $500 million cost⁵⁰. We consider only the $50 million borne by local users in this study. The fact that we find significant financial risks even with this unusually favorable cost structure serves as a warning for other less-subsidized infrastructure investments in the region. We also consider a second infrastructure investment in a new groundwater bank, which would route surplus water available in wet periods into infiltration basins to replenish the aquifer and allow for additional groundwater pumping during dry periods^20,27,28,42. We estimate this project would also cost local water providers $50 million based on cost estimates for other groundwater storage projects being developed throughout the region⁴³.

Using the cooperative infrastructure investment context of the Friant-Kern canal rehabilitation and its potential augmentation with a new groundwater bank, we simulate the performance of alternative infrastructure partnerships using the California Food-Energy-Water System (CALFEWS), a highly detailed water resource systems model that simulates daily reservoir operations, environmental flow rules, water rights, interbasin transfer projects, and conjunctive surface and groundwater supply management at the level of individual water providers^23,33. The high-fidelity system representation of the model allows us to generate new insights into infrastructure partnership design compared to existing lower-fidelity water supply planning models that cannot resolve the multi-timescale dynamics of managed aquifer recharge or the multi-spatial scale distribution of water supply benefits and financial risks for individual water providers^23,33. Each partnership is evaluated across an ensemble of daily-time step 30-year hydrologic sequences from a synthetic streamflow generator that captures the spatial correlation across the state as well as interannual persistence of wet and dry periods (see Methods and Supplemental Note S1). This allows us to explore the impact of California’s severe internal hydroclimatic variability on the water supply benefits and financial risks resulting from these infrastructure investments^51,52. The synthetic streamflows used in this study serve two major technical benefits: (1) they permit a well-founded statistical representation of the plausible flood-drought regimes that regional infrastructure investments will face in the near-term and (2) they are intentionally strongly optimistic in neglecting longer-term climate changes to show that even without climate change effects, there are immense uncertainties that challenge our understanding of investment partnerships’ water supply and financial risk tradeoffs.

Partnership performance is assessed over the 30-year simulation (a common municipal bond payback period⁵³) according to five decision-relevant metrics related to partnership size, water supply benefits, and financial risk. We combine this ensemble modeling framework with a multiobjective intelligent search framework that leverages the Borg Multi-Objective Evolutionary Algorithm (MOEA)^54,55 to test over 600,000 candidate infrastructure investment partnerships to discover the best-performing partnerships that represent the optimal tradeoffs across the five metrics (Supplementary Fig. S1). The elements of partnership design that are optimized can be described with three questions: (1) In which project does the partnership invest (canal expansion, groundwater bank, or both)? (2) Which subset of 42 water providers in the region should participate? (3) What “ownership share” should be assigned to each partner, governing its capacity in the project and its share of annual debt payment obligations? Additional details on our partnership design and evaluation framework can be found in Methods and Supplementary Notes S2-3.

Understanding diverse partnership design options

Our multiobjective intelligent search process yields a set of 374 approximately Pareto-optimal partnerships (hereafter referred to as “optimal tradeoff partnerships”), each of which represents a different balance of compromises across the five conflicting objectives. The vast majority of optimal tradeoff partnerships (99%) were found to invest in both canal expansion and groundwater banking (Fig. 2a). This suggests a synergistic relationship. The surface water gains from the Friant-Kern canal expansion generally occur during relatively high-flow periods, when available water supply exceeds immediate demand, so it is complementary to pair the expansion with additional storage^29,56. The remaining 1% of optimal tradeoff partnerships invest only in a groundwater bank, with no partnerships investing in canal expansion alone. This highlights the importance of considering the interactive effects across alternative infrastructure projects and the potential for synergistic gains when bundling regional investments together.

There is a high degree of diversity in the way that partnerships can be designed. The optimal tradeoff partnerships range in size from 2 to 24 water providers, with 60% of candidate partnership designs having 10 to 17 partners (Fig. 2a). There is also diversity in the concentration of ownership across the partnerships. The smallest partnerships tend to distribute ownership roughly equally (e.g., three partners with shares of ~ 33% each). Larger partnerships tend to have a few partners with disproportionately large ownership shares and more partners with rather small shares (Fig. 2b, Supplementary Fig. S2). Considering the geographic context of these partnerships (Figs. 2c-e), we find that certain providers almost always participate and generally carry large ownership shares (black), while others rarely participate and tend to take rather small shares (yellow). There are also providers that regularly participate with small shares (light brown) or that irregularly participate with large shares (teal). These results demonstrate the complexity of partnership design and the range of roles that different water providers can play in partnership creation given their unique contexts: the local infrastructure networks, water rights, hydrologic context, and other factors that impact their ability to procure and store additional surface water as a result of the collaborative investment. Moreover, many providers can span a range of ownership shares across the optimal tradeoff set depending on how the rest of the partnership is constructed (Fig. 2d), which highlights the importance of capturing local-scale dynamics and network interactions between water providers when designing collaborative infrastructure investments.

Key performance tradeoffs for water supply investment partnerships

We find that water providers face severe tradeoffs across five decision-relevant metrics when designing collaborative infrastructure investments (Fig. 3). The optimal tradeoff partnerships can increase average surface water deliveries to partners by anywhere from 37 to 116 GL/year. For context, 100 GL (81 kAF) is equivalent to 15% of Lake Millerton’s capacity or 4% of Kern County’s average irrigated acreage⁵⁷. With increased surface water deliveries, most partnerships are able to reduce their average groundwater pumping, which could help them meet their obligations under the Sustainable Groundwater Management Act⁴². However, the effect on groundwater pumping varies widely across the optimal tradeoff partnerships, from a reduction of 74 GL/year to an increase of 3 GL/year on average.

In addition to water supply volumes, we also calculate the worst-off partner cost of gains, defined as the annual debt service payment divided by the annual captured water gain for the worst-off partner in each partnership. The 90th percentile cost of gains across alternative hydrologic scenarios, which we use as a proxy for partner-level financial risk, ranges from $78/ML to over $1000/ML in different partnerships (Fig. 3; see Methods for the detailed definition). For context, water providers in this region generally charge $32–154/ML ($40–190/AF) for irrigation water⁵⁷, which is used to pay for debt service on infrastructure investments, as well as the cost of procuring water and other operating expenses. Financial distress can occur when a water provider agrees to future debt obligations as part of an infrastructure partnership but does not end up receiving substantial water supply benefits from the project. Optimal tradeoff partnerships with very high costs of gains may be unable to raise customer rates high enough to pay off the debt associated with the new infrastructure. This could lead to default, credit downgrade, deferred maintenance, or other challenges^9–11. Significant increases in customer water rates are also problematic due to the water affordability and water quality challenges in many low income communities, as well as the thin profit margins in many agricultural regions where higher rates for irrigation water can make farming infeasible^{12,14,44,58,59}.

In general, the partnerships that achieve the largest surface water gains and groundwater pumping reductions tend to have many partners (Fig. 3, Supplementary Figs. S3-S4). Larger partnerships are expected to have larger and more diverse sources of water supply, water demand, and local infrastructure, and thus are more capable of fully utilizing the new shared infrastructure throughout the year and across a range of conditions. However, the largest partnerships also tend to produce the largest financial risks for their worst-off partners. Almost all partnerships with 20 or more partners have at least one partner paying over $1000/ML, while each partnership in which all partners pay under $150/ML have 11 or fewer partners (Supplementary Fig. S3 & S5). It is intuitive that a larger number of partners makes it harder to satisfy all partners’ needs. Yet this finding also points to an important tension in California’s recent efforts to incentivize large-scale collaborative water supply investments²¹. Large-scale collaboration can indeed be beneficial, as evidenced by the positive relationship between partnership size and captured water gains and pumping reductions (Fig. 3, Supplementary Fig. S4). However, it is also critical to understand the significant financial risks that can arise from these arrangements and to ensure that all partners are receiving adequate benefits to justify their debt.

We also find a strong tradeoff between water supply benefits for the partners vs. other non-partner water providers in the region (Fig. 3). New infrastructure projects alter water providers’ path-dependent supply and demand behaviors, with ripple effects throughout the broader interconnected water supply network. This can lead to indirect effects on water availability for other water providers in the region that are not party to the investment. We find that the non-partner impacts vary greatly across the optimal tradeoff set, from gains of 28 GL/year to losses of 27 GL/year on average. In general, the partnerships with the largest captured water gains and pumping reductions for the partners tend to significantly reduce deliveries to non-partners (Supplementary Fig. S4). This suggests that these infrastructure partnerships benefit not only from newly captured “surplus” water that would have otherwise flowed out of the region, but also from reallocated water that would have otherwise been delivered to their neighbors. This raises important questions of whether non-participating water providers have the ability to block infrastructure investments that could negatively impact them. Water providers investing in new infrastructure in California must navigate the state’s complex web of different water rights, environmental laws, and administrative procedures^25,26. In particular, as in other western states, many changes related to water rights or diversion patterns require that the change will cause “no injury” to other water right holders; this can add significant expense and delays to the approval process^60,61. Moreover, the diversity of legal regimes and water laws in the state (e.g., prior appropriative rights vs. riparian rights vs. federal/state water supply contracts) complicates the evaluation of alternative water supply options, and more guidance is needed from the state to clarify procedures and facilitate more collaborative partnerships. Excluding partnership designs with significant negative impacts on non-partners from the optimal tradeoff set to account for potential political and legal constraints causes the best achievable captured water gains and pumping reductions to be reduced by 16% and 6%, respectively (Supplementary Fig. S6). This negative interaction represents a major challenge for supply reliability and groundwater sustainability efforts and points to the need for more coordinated regional infrastructure planning.

It is perhaps surprising that such a wide diversity of performance tradeoffs is possible, given that 99% of these partnerships invest in the exact same infrastructure: canal expansion and a groundwater bank (Fig. 2a). This demonstrates the underappreciated and critical role of partnership design itself in governing the performance of collaborative infrastructure investments. The variety of partnership designs associated with different performance tradeoffs means that it is not possible to establish a single “best” partnership a priori without first establishing decision-making preferences (i.e., how much to weigh partnership size vs. captured water gains vs. worst-off partner cost). Moreover, the striking performance tradeoffs that emerge from nuanced changes in partner selection and ownership distribution highlight the significant advantages of pairing detailed water supply models like CALFEWS²³ (capable of resolving daily-timescale coupled hydrologic, infrastructural, and institutional dynamics at the scale of individual water providers) with multiobjective intelligent search algorithms like the Borg MOEA (capable of exploring a much wider range of candidate partnership designs compared to traditional ad hoc planning processes).

Navigating uncertainty and heterogeneity in tradeoffs

We also find that hydroclimatic variability leads to significant uncertainty in the performance of alternative infrastructure partnerships. When an investment partnership borrows money in municipal bond markets to pay for a new infrastructure project, it is subject to significant uncertainty related to the future realized water supply benefits and sales revenues. This uncertainty can be highly decision-relevant when there is a wide range of performance values across plausible alternative future scenarios. The metrics in Fig. 3 represent aggregate performance computed across an ensemble of 64 sampled 30-year sequences of multi-site correlated daily streamflows. This ensemble captures a wide but plausible range of hydrologic conditions that investments could confront over a 30-year bond payback period (see Methods and Supplementary Notes S1-S2). The captured water gain and pumping reduction metrics in Fig. 3 are calculated as the mean values across the 64 sampled 30-year daily streamflow sequences, while the worst-partner cost of gains metric is calculated using the 90th percentile across scenarios. However, a deeper examination of performance in individual sampled 30-year daily hydrologic sequences reveals significant levels of uncertainty in the benefits that water providers can expect due to internal hydroclimatic variability. These uncertainties can be highly meaningful in the context of risky large-scale infrastructure investments.

We now highlight a “Compromise Partnership” selected as an example for navigating the tradeoffs in the optimal partnership set. The Compromise Partnership has 17 partners investing in both canal expansion and groundwater banking (Fig. 4a) and is selected for its relatively high performance across all five performance objectives (see Methods). Like all partnerships in the optimal tradeoff set, the Compromise Partnership is subject to significant performance uncertainty (Fig. 4b). Although the expected value of captured water gain is 84 GL/year, the captured water gain in individual 30-year daily hydrologic sequences can range from 50 to 124 GL/year, or 60–147% of the expected value. The relative variability of pumping reductions spans a similarly wide range. The captured water gain for non-partners is even more variable, ranging from a gain of 28 GL/year (3.9x larger than the expected value of 7 GL/year) to a loss of 51 GL/year (a loss 7.2x larger in magnitude than the expected gain). The worst-partner cost of gains also spans a wide range, from $137/ML in the best scenario (47% of the 90th percentile metric of $291/ML) to $742/ML in the worst scenario (255% of the 90th percentile metric). This cost differential could easily be the difference between a project that affordably improves surface water access and reduces groundwater overdraft and one that provides little water supply benefit and overburdens water providers with debt and their customers with rate increases.

The benefits and risks of different partnerships can be very unevenly distributed across the project partners due to the complexity of the interconnected water supply system dynamics and the heterogeneity of local contexts for water providers. The partners within the Compromise Partnership experience widely different expected costs of gains for their water supply benefits (Fig. 4c), which is common across the optimal tradeoff partnerships. The heterogeneity of expected costs stems from the similarly heterogeneous expected captured water gains at the partner level (Supplementary Fig. S8a). The latter is not inherently problematic so long as each partners’ captured water gain is appropriately matched to its ownership share in the project and thus its share of the annual debt payments. For example, if Provider A receives twice as much captured water gain as Provider B, but also makes annual debt payments that are twice as large, then the two providers are effectively paying the same unit cost for their captured water gains. However, we find the ownership shares to be imperfectly matched in this and other partnerships, so that some providers pay more than their “fair share” on average and others pay less (Fig. 4c).

Moreover, partners can experience widely varying degrees of uncertainty related to their individual performance tradeoffs (Supplementary Fig. S8). The aggregate cost of gains for the Compromise Partnership (Fig. 4c, black distribution) has a relatively low chance of exceeding $100/ML in any of the sampled hydrologic scenarios. However, the cost of gains for the worst-off partner (red distribution) has a much wider range, reaching almost $800/ML in the most challenging scenario. This risk is not evenly distributed, with some partners experiencing a disproportionate share of extreme costs (e.g., Providers 3 and 4) and others experiencing uniformly low costs across the sampled hydrologic scenarios (e.g., Providers 7 and 12). This is critical because the heterogeneity of water supply benefits and financial risks could threaten the cooperative stability of the partnership itself if partners do not perceive the investment to be sufficiently fair and locally beneficial^30,33.

As noted before, these results (Fig. 4) are a strongly optimistic portrayal of partnership performance under conditions of uncertainty and heterogeneity, which reinforces the importance of our results given the impacts of anthropogenic climate change. Climate change is expected to make dry conditions increasingly frequent and severe in California and many other regions^2,62,63. Our expressly optimistic framing of future hydroclimatic conditions in California neglects any nonstationary effects from anthropogenic warming or low-frequency decadal climate oscillations. However, even under such optimistic conditions, the extreme levels of interannual hydroclimatic variability experienced in California^51,52 lead to significant decision-relevant uncertainty in cooperative infrastructure investment outcomes over the next several decades. Water providers therefore must find a way to improve surface water reliability and reduce groundwater overdraft without exacerbating the growing water affordability concerns for agricultural and urban water users^12–14. We caution that decision-makers should carefully consider the impacts of uncertainty and financial risk before committing to significant debt associated with new infrastructure partnerships. However, this will require updated planning frameworks that move beyond expected value-based benefit-cost analyses (e.g., exploratory modeling and robust decision-making approaches^64–66), as well as detailed water supply models that can account for complex local-to-regional scale dynamics in coupled hydrologic, infrastructural, and institutional systems under diverse conditions²³.

Regret of the current baseline partnership design

Lastly, we use our results to highlight the significant regret associated with the current baseline planned Friant-Kern Canal rehabilitation partnership. In practice, infrastructure partnerships are generally established via pre-existing relationships or ad hoc processes. Most infrastructure investments are then evaluated using low-resolution models that fail to capture key system features (i.e., interdependent flood-drought dynamics, institutional constraints, infrastructure operations, etc.) and the preferred alternative is selected based on highly aggregated traditional expected benefit-cost analyses. These processes fail to grapple with the complex tradeoffs, uncertainties, and heterogeneities in modern interconnected water supply systems, and therefore are ill-equipped to design robust and equitable infrastructure partnerships. To elucidate the regret associated with current baseline planning processes, we analyze the ongoing rehabilitation of the Friant-Kern Canal by the Friant Contractors, a collection of water providers with contracts to receive Central Valley Project water from Millerton Lake via the Friant-Kern Canal (Fig. 1)^48–50. We model a “Status Quo Partnership” (Fig. 5a) after this real-world example, as opposed to the partnerships in the optimal tradeoff set which are designed via our multiobjective intelligent search process (see Methods).

Although it does provide significant benefits to its 16 partners, the Status Quo Partnership is outperformed on all five aggregate performance metrics by the Compromise Partnership (Fig. 5b). The Status Quo Partnership produces 40% lower captured water gains and 34% lower groundwater pumping reductions on average. The Status Quo Partnership also results in slightly negative impacts on non-partners in the region, while the Compromise Partnership provides positive gains. Accounting for uncertainty, we see that the Status Quo Partnership significantly underperforms relative to the Compromise Partnership in the majority of the sampled hydrologic scenarios for all three water supply metrics.

The Status Quo Partnership also carries significantly greater financial risk. The worst-partner cost of gains exceeds $1000/ML in all 64 sampled hydrologic scenarios. Two water providers in particular (Partners 8 and 14) receive marginal or even negative captured water gains, and thus experience costs over $1000/ML, in a large majority of scenarios (Fig. 5c, Supplementary Fig. S9). Three other providers also exceed $1000/ML in a smaller set of scenarios. In contrast, there are no sampled hydrologic scenarios where any of the 17 water providers in the Compromise Partnership experience costs of gains above $742/ML, and in 90% of scenarios the worst-off partner pays less than $291/ML.

Three major differences in partnership design contribute to the Compromise Partnership’s dominance over the Status Quo Partnership. First, it couples the canal expansion project with new groundwater banking facilities, which help to capture more surplus water during high-flow periods when all partners’ immediate demands are already satiated. Partnerships in the optimal tradeoff set with more than four partners exclusively invest in both partnerships (Supplementary Fig. S3), suggesting that the performance gains from the synergistic pairing are worth the higher price tag. Second, although the Compromise Partnership and the Status Quo Partnership have significant overlap in participating providers (Figs. 4a & 5a), the former removes several Friant Contractors with marginal benefits and significant financial risks. It also adds several other non-Friant providers that stand to benefit from the infrastructure investment. Widening the net beyond the Friant Contractors allows for a more diversified portfolio of water supplies and demands and increases the utilization of the infrastructure across a range of seasons and conditions. Lastly, the ownership shares and annual debt payments are better matched to partner-level captured water gains in the Compromise Partnership than the Status Quo Partnership, which helps to equalize the cost of gains across partners. These results highlight the significant regret associated with traditional ad hoc partnership design processes and the substantial improvements that can be achieved by combining detailed ensemble water supply modeling with multiobjective intelligent search. In future work, these tools can be used to complement more traditional stakeholder-based collaborative planning efforts through iterated processes of search, evaluation, and human feedback^67–70.

According to the American Society of Civil Engineers, the U.S. will need to invest $109 billion per year in water infrastructure over the next 20 years to overcome the current investment gap in light of the deteriorating condition of the existing infrastructure stock as well as factors related to climate change, economic change, and regulatory change⁵. Collaborative partnerships and regionalization can be effective tools for facilitating investment by lowering costs and increasing flexibility. However, collaboration also brings new challenges related to the design and evaluation of robust infrastructure investments in institutionally complex water supply networks. Addressing these challenges will be critical to achieving California’s water supply resilience and groundwater sustainability goals^21,42. More broadly, these insights should be highly relevant to many other regions around the world that are working to adapt to increasing water scarcity, especially in regions with complex water institutions and interconnected infrastructure systems.

Our results have several key implications for water providers seeking to meet their supply reliability and groundwater sustainability goals through collaborative investment. First, water providers and planning agencies should assess financial risk as a key metric for infrastructure partnership design, along with more traditional planning metrics such as water supply reliability. Second, the severe tradeoff between partnership size and worst-partner cost of gains highlights an important tension between collaborative inclusivity and financial risk which must be navigated by water providers when developing regional partnerships. Third, given the strong tradeoff between water supply benefits for partners vs. non-partners, there is a need for state and national planning agencies to provide more guidance on the role of external stakeholders in the planning process when infrastructure investments have the potential for negative external impacts. Fourth, the high level of variability across alternative sampled hydrologic scenarios demonstrates the importance of considering internal hydroclimatic variability in infrastructure partnership evaluation, while the significant partner-level heterogeneities highlight the critical role of highly detailed water supply models that can resolve complex infrastructural and institutional constraints and multiscale system dynamics. Both uncertainty and heterogeneity provide highly decision-relevant information that would not be available using traditional aggregated mean benefit-cost analyses. Fifth, given the preponderance of partnerships investing in both canal expansion and groundwater banking within the optimal tradeoff set, there is clearly a need for more integrated planning efforts that consider the interactive effects of multiple simultaneous infrastructure investments spanning local to regional scales within the larger water supply network. Finally, the severe performance tradeoffs that emerge from relatively subtle changes in partnership design across the optimal tradeoff set and the dominance of the Compromise Partnership over the Status Quo Partnership both demonstrate the value of multiobjective intelligent search for improving the design of infrastructure partnerships in institutionally complex water supply networks relative to traditional ad hoc planning processes.

In this study, we explore hydrologic uncertainty using a broad ensemble sampling of plausible 30-year daily streamflow sequences from a synthetic generator trained on historical observations. This represents an actively optimistic framing because we only account for stationary hydroclimatic variability while neglecting nonstationary climate change. Despite the optimistic framing, we nevertheless find that uncertainties in assessing water supply benefits and financial risks are highly consequential as a result of California’s highly variable climate. Future work will consider the impact of a wider range of plausible nonstationary hydrologic futures to account for the impact of climate change on uncertainty and decision-making. Future work will also consider additional uncertainties related to future water demand, reservoir operations, infrastructure performance, and cost. These factors are expected to further widen the envelope of uncertainty around aggregate and partner-level impacts and exacerbate the challenge of designing robust partnerships. Future extensions of this work should also consider the role of contractual innovations, environmental impact bonds, and adaptive risk management strategies in redistributing and reducing the partner-level financial risks associated with cooperative infrastructure investments^34,71–74. Lastly, it is important to note that different water providers will have different abilities to pay for infrastructure and different vulnerabilities to financial risk based on their underlying customer bases and political contexts. Although we focus on equality between partners for this study, future work should extend our analysis by considering equity and asymmetries in economic and political power, customer vulnerability, racial injustice, and historical responsibility for groundwater overdraft^58,75–78.

Candidate infrastructure investments

Following Hamilton et al. (2022)³³, we consider two candidate infrastructure projects in this study. First is the rehabilitation of the Friant-Kern Canal. The canal has been severely damaged by subsidence related to groundwater overdraft, which has reduced its conveyance capacity by up to 60% in some regions^48,49. A collection of water providers known as the Friant Contractors (Fig. 1) is currently investing roughly $50 million in rehabilitating the canal, with the remainder of the $500 million cost to be covered by various federal, state, and regional organizations⁵⁰. For this study, we consider only the $50 million cost share allocated to the local water providers. We assume for simplicity that the modeled investment will restore the entire length of the canal to its original conveyance capacity.

The second candidate infrastructure project is a new groundwater banking facility near the confluence of the Friant-Kern Canal and the Tule River. This hypothetical project is not based on any particular existing or planned groundwater bank, but is largely consistent with other recent and planned investments in the region⁴². We estimate that the bank’s cost to project partners will be $50 million, which is on par with other recent and planned investments in groundwater recharge facilities⁴³. There are three major parameters governing the capacity of groundwater recharge facilities in our water resource simulation model: the infiltration pond storage volume, the dry soil infiltration capacity, and the recovery pumping capacity. These are set to 0.74 GL/day (0.6 kAF/day), 0.37 GL (0.3 kAF), and 0.25 GL/day (0.2 kAF/day), approximately 25% of the parameter values for the largest groundwater bank in the region, the Kern Water Bank⁷⁹.

Both infrastructure projects are assumed to be financed using 30-year revenue bonds with 3% interest, broadly consistent with revenue bonds issued in 2019 by water districts in California⁵³. Assuming equal annual payments over time, this translates to a debt payment of $2.55 million/year for partnerships investing in either canal expansion or groundwater banking, or $5.1 million/year for partnerships investing in both projects. All other infrastructure investment details are taken to be consistent with Hamilton et al. (2022)³³.

The elements of partnership design to be optimized can be understood as answering three questions. First, which infrastructure project should be built: canal expansion, groundwater banking, or both? Second, out of 42 candidate water providers in the region (Fig. 1), which subset should work together as partners? Third, once the set of partners has been established, how should the “ownership shares” be distributed amongst them? Each partner’s ownership share dictates its share of priority capacity in the new infrastructure as well as its share of annual debt payment obligations. Importantly, the annual debt payment cost considered here does not include any operating or maintenance costs associated with the infrastructure or the cost of procuring additional water. It thus represents an incomplete and optimistic view of the potential for financial risk associated with these infrastructure partnerships. The minimum allowable ownership share is set to 1%, so that each partner’s share is between 1-100%. All of the partner ownership shares must sum to 100%.

Infrastructure partnership evaluation

This study employs the California Food-Energy-Water System (CALFEWS), a free, open-source, Python/Cython-based simulation model of California’s water resource system dynamics with a particular focus on the San Joaquin Valley²³. The model combines detailed representation of reservoir operations, water conveyance, interbasin transfer projects, water rights, municipal and agricultural demand, and conjunctive surface and groundwater supply management including groundwater banking operations. CALFEWS operates on a daily time step and resolves water deliveries and operations at the level of individual water providers (e.g., municipal utilities, irrigation districts, and water storage districts). This spatiotemporal resolution combined with detailed representation of water management institutions is unique among models for California and leads to improved performance in reproducing historical system behavior, from reservoir storages and releases to interbasin transfer project pumping to groundwater banking balances²³. Most existing water supply planning models employ longer timesteps (e.g., daily or monthly), larger spatial aggregation (e.g., regional), and simplified representation of water management institutions and/or infrastructure constraints. CALFEWS’ high-fidelity system representation allows for improved evaluation of infrastructure partnerships compared to lower fidelity models that cannot resolve the multi-timescale dynamics of managed aquifer recharge or the multi-spatial scale distribution of water supply benefits and financial risks for local project partners (see Zeff et al. (2021)²³ and Hamilton et al. (2022)³³ for detailed discussions).

Hydroclimate in California is characterized by extreme interannual variability^51,52. Over the last 20 years, the region has experienced three periods of persistent droughts that have been ended by extreme wet periods driven by atmospheric rivers. Facilitating robust management of the region’s water supply system requires simulating operations under an array of scenarios that can represent current and plausible future hydroclimatic sequences in the region. To support this modeling effort, we generate synthetic full natural flows using a two-state Gaussian Hidden Markov Model (HMM), which has been shown to be an appropriate method to capture long-term persistent drought conditions^80,81. The 110-year multisite historical full natural flow reanalysis dataset described in Zeff et al. (2021)²³ is used for training and verification of the synthetic hydrologic scenario generator. The HMM consists of two “hidden” climate states representing wet and dry conditions. Within each of these states, log-space flows are sampled from respective Gaussian distributions. Following Thyer & Kuczera (2003)⁸², we contribute a multi-site extension of the two-state HMM-based model to simultaneously generate full natural flows for 15 major watersheds in the Central Valley system. We generate 32 different 30-year sequences of multi-site correlated daily full natural flows to use within the multiobjective intelligent search step, along with another larger set of 64 additional 30-year daily streamflow sequences for the reevaluation step (Supplementary Fig. S1). The number of sequences simulated in the search (32) and reevaluation (64) steps are selected to balance the fidelity of the noisy performance metrics against the significant computational expense of the search. The synthetic full natural flow records are well matched to the historical record in terms of correlation, distribution, and flow duration (Supplementary Figs. S7, S10, S11, S14). However, the synthetic ensemble also widens the envelope of extreme high and low flows beyond the historical record, allowing for a deeper understanding of the impacts of internal hydroclimatic variability on infrastructure performance uncertainty. More details on the hydrologic scenario generator and its verification can be found in Supplementary Note S1. Given a multi-site full natural flow record (e.g., from our synthetic scenario generator), CALFEWS uses an internal statistical reconstruction module to extrapolate inflows, snowpack levels, and other relevant hydroclimatic variables based on monthly linear regression relationships derived from historical observed data (see Zeff et al. (2021)²³ for a full description).

Each candidate infrastructure partnership is evaluated according to five decision-relevant metrics: (1) the number of water provider partners participating in the partnership; (2) the captured water gain, defined as the expected increase in total surface water deliveries across all project partners; (3) the pumping reduction, defined as the total expected reduction in groundwater pumping across all project partners; (4) the captured water gain for non-partner water providers in the region; and (5) the cost of gains for the worst-off partner, defined as the annual debt payment divided by captured water gain for the worst-off partner in a partnership. The ideal partnership would maximize Objective Metrics 1–4 and minimize Objective Metric 5, but given their conflicting nature tradeoffs exist. These tradeoffs emerge because performance gains in single objective comes at the cost giving up performance in one or more of the remaining objectives. Each candidate infrastructure partnership is simulated and evaluated across 32 different 30-year daily hydrologic sequences from the synthetic generator in the multiobjective intelligent search step, and optimal tradeoff partnerships are then reevaluated across 64 additional 30-year sampled sequences to ensure robust generalization (see next section). Objective Metrics 2–4 are calculated by taking the expected value across sampled hydrologic scenarios and the sum across water provider partners (or non-partners in the case of Metric 4). Objective Metric 5 is calculated using the 90th percentile across sampled hydrologic scenarios and the max (i.e., worst-case) across partners in order to favor partnerships that can provide robust affordable water supply benefits across a wide range of plausible hydrologic conditions^83,84. This problem formulation draws from previous efforts to define robust multiobjective problem formulations in noisy water resources simulation-optimization contexts^35,83,84. More details on objective formulations can be found in Supplementary Note S2 and Supplementary Table S4.

Multiobjective intelligent search

The Borg Multiobjective Evolutionary Algorithm (MOEA) is a hyper-heuristic intelligent search algorithm employing auto-adaptive search operator selection, epsilon-dominance archiving, epsilon-progress stagnation detection, adaptive population sizing, and other beneficial features⁵⁴. The Borg MOEA has been successfully employed across a range of difficult problems in water resources and engineering design characterized by high-dimensional, noisy, multimodal search spaces^85–90. We employ the master-worker parallel version of the Borg MOEA, which is highly scalable on high performance computing clusters^91,92.

The Borg MOEA is used to discover the set of optimal tradeoff partnerships (i.e., non-dominated or approximately Pareto optimal solutions) within our multiobjective intelligent search procedure (Supplementary Fig. S1). We run six independent searches with different random seeds to account for the stochastic path-dependent nature of metaheuristic search algorithms⁵⁵. We develop two alternative formulations with different strategies for mapping from the numeric decision vector that is optimized to the underlying infrastructure partnership that is simulated (Supplementary Note S3). Because it is unknown a priori which decision vector formulation will exhibit superior learning, we assign three of the six random seeds to each formulation. For each seed, we run the parallelized Borg MOEA for 120 hours across 128 Intel Xeon Skylake nodes (6,144 cores) on the Texas Advanced Computing Center’s Stampede2 supercomputer. In total, we test 629,312 candidate infrastructure partnerships across the 6 seeds. The number of seeds and wall clock per seed are selected to balance the noise across seeds vs. the convergence, consistency, and diversity of solutions within each seed according to common multiobjective search metrics such as hypervolume, additive epsilon indicator, and generational distance^55,85,86. Each seed exhibits acceptable convergence, consistency, and diversity given the constraints of our computing budget (Supplementary Fig. S15). The second decision vector formulation exhibits clearly stronger learning compared to the first formulation.

Upon completion, we calculate the overall non-dominated reference set across all seeds, yielding 1291 unique infrastructure partnerships. Each of these partnerships is then reevaluated across 64 additional 30-year synthetic hydrologic sequences outside of the training set to ensure that partnerships are robust in the face of a range of plausible hydroclimatic conditions. The final optimal tradeoff set consists of the 374 partnerships that remain non-dominated after reevaluation. We also evaluate the Status Quo Partnership (see next section) with respect to the same 64 hydrologic scenarios. All performance metrics reported in this study refer to performance in the reevaluation step. The epsilon-dominance parameters that define the meaningful resolution for each objective with respect to the non-dominance calculation in the MOEA search and reevaluation steps can be found in Supplementary Table S4.

Selection of Status Quo & Compromise Partnerships

We highlight two alternative partnerships in more detail in the results. The Status Quo Partnership (Fig. 5) is based on the real-world coalition of Friant Contractor water providers (Fig. 1) that are currently investing roughly $50 million to rehabilitate the Friant-Kern Canal^49,50. We design this partnership to mirror the existing institutional arrangement rather than being designed by our multiobjective intelligent search process. Given that the formal plans for apportioning cost shares across the partnership are not available, for the purpose of this analysis the ownership share for each of the 16 partners is assumed to be proportional to its average historical delivery of Central Valley Project Friant water, following Hamilton et al. (2022)³³.

By comparison, the Compromise Partnership is selected as a representative solution from the broader 374-member optimal tradeoff set that outperforms the Status Quo Partnership and performs relatively well across all five objectives. This partnership is selected through a two-step process (Supplementary Fig. S16). First, we find the subset of optimal tradeoff partnerships that perform equal to or better than the Status Quo Partnership on all five aggregate objective metrics. The Compromise Partnership is then selected out of this subset by finding the partnership with the lowest worst-partner cost of gains.

Data availability

All data and code needed to generate the analysis and figures in this paper can be found in the GitHub repository (https://github.com/ahamilton144/CALFEWS/tree/MORDM_experiment_paper1) as well as in a permanent archive on Zenodo⁹³.

Code availability

Acknowledgements

Funding for this work was provided by the National Science Foundation (NSF), Innovations at the Nexus of Food-Energy-Water Systems, Track 2 (Award 1639268). The authors also acknowledge support from the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by NSF grant ACI-1548562. The multiobjective intelligent search and reevaluation simulation steps were carried out on Stampede2 at the Texas Advanced Computing Center (TACC) through XSEDE allocation TG-EAR090013. The authors also acknowledge the Cornell University Center for Advanced Computing (CAC) for additional HPC resources and support. The views expressed in this work represent those of the authors and do not necessarily reflect the views or policies of the NSF, TACC, or CAC.

Author contributions

A.H. conducted experiments, analyzed data, and wrote the first draft of the manuscript, except for figures and text contributed by R.G. related to the synthetic hydrologic scenario generator. A.H., R.G., and H.Z. contributed to code development. P.R. and G.C. supervised the project and contributed to the development of the paper’s central ideas along with A.H. All authors participated in the final version of the manuscript.

Competing interests

The authors declare no competing interests.

Caretta, Martina Angela et al. Water. in Climate Change 2022: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change 551–712 (Cambridge University Press).
Lall, U. et al. Water. in Impacts, Risks, and Adaptation in the United States: Fourth National Climate Assessment (eds. Reidmiller, D. R. et al.) vol. II 145–173 (U.S. Global Change Research Program, 2018).
He, C. et al. Future global urban water scarcity and potential solutions. Nat. Commun. 12, 4667 (2021).
Dolan, F. et al. Evaluating the economic impact of water scarcity in a changing world. Nat. Commun. 12, 1915–1915 (2021).
American Society of Civil Engineers & Value of Water Campaign. The Economic Benefits of Investing in Water Infrastructure. https://www.uswateralliance.org/sites/uswateralliance.org/files/publications/The%20Economic%20Benefits%20of%20Investing%20in%20Water%20Infrastructure_final.pdf (2020).
Greer, R. A. A review of public water infrastructure financing in the United States. Wiley Interdiscip. Rev. Water 7, (2020).
Hanak, E. et al. California’s Water: Paying for Water. 1–4 https://www.ppic.org/wp-content/uploads/content/pubs/report/R_1016EH2R.pdf (2018).
Hansen, K. & Mullin, M. Barriers to water infrastructure investment: Findings from a survey of U.S. local elected officials. PLOS Water 1, e0000039–e0000039 (2022).
Hughes, J. et al. Defining a Resilient Business Model for Water Utilities. https://efc.sog.unc.edu/sites/default/files/4366_Exec_Summary_0.pdf (2014).
Chapman, T. A. & Breeding, J. M. U.S. Public Finance Waterworks, Sanitary Sewer, and Drainage Utility Systems: Methodology And Assumptions. 1–39 https://www.spratings.com/documents/20184/908554/US_PF_Event_RFCRndTblsJan2015_Article1/30d125eb-1066-4730-8ab1-f2cd6a6d6f9a (2014).
Smull, E., Patterson, L. & Doyle, M. Rising Market Risk Exposure of Municipal Water Service Providers in Distressed Cities. J. Water Resour. Plan. Manag. 148, (2022).
Patterson, L. A. & Doyle, M. W. Measuring water affordability and the financial capability of utilities. AWWA Water Sci. 3, (2021).
Rachunok, B. & Fletcher, S. Socio-hydrological drought impacts on urban water affordability. Nat. Water 1, 83–94 (2023).
Cardoso, D. S. & Wichman, C. J. Water Affordability in the United States. Water Resour. Res. 58, (2022).
Bell, E. V., Fencl, A. & Mullin, M. External drivers of participation in regional collaborative water planning. Policy Stud. J. (2022) doi:10.1111/psj.12473.
Hansen, K., Mullin, M. & Riggs, E. K. Collaboration Risk and the Choice to Consolidate Local Government Services. Perspect. Public Manag. Gov. 3, 223–238 (2020).
Gorelick, D. E., Zeff, H. B., Hughes, J., Eskaf, S. & Characklis, G. W. Exploring Treatment and Capacity-Sharing Agreements Between Water Utilities. J. - Am. Water Works Assoc. 111, 26–40 (2019).
Riggs, E. & Hughes, J. Crafting Interlocal Water and Wastewater Agreements. https://efc.sog.unc.edu/wp-content/uploads/sites/1172/2021/06/Crafting20Interlocal20Agreements_Final_01.pdf (2019).
Cypher, T. & Grinnell, C. Governments Working Together. A Citizen’s Guide to Joint Powers Agreements. https://sgf.senate.ca.gov/sites/sgf.senate.ca.gov/files/GWTFinalversion2.pdf (2007).
Escriva-Bou, A., Sencan, G., Hanak, E. & Wilkinson, R. Water Partnerships between Cities and Farms in Southern California and the San Joaquin Valley. https://www.ppic.org/wp-content/uploads/water-partnerships-between-cities-and-farms-in-southern-california-and-the-san-joaquin-valley.pdf (2020).
Newsom, G. et al. California Water Resilience Portfolio - Governor’s Executive Order N-10-19. https://waterresilience.ca.gov/wp-content/uploads/2020/01/California-Water-Resilience-Portfolio-2019-Final2.pdf (2020).
Norriss, J., Cunningham, M., DeRosa, A. R. & Vedachalam, S. Too Small to Succeed: State‐Level Consolidation of Water Systems. J. AWWA 113, 8–15 (2021).
Zeff, H. B. et al. California’s food-energy-water system: An open source simulation model of adaptive surface and groundwater management in the Central Valley. Environ. Model. Softw. 141, 105052–105052 (2021).
Trindade, B. C., Gold, D. F., Reed, P. M., Zeff, H. B. & Characklis, G. W. Water pathways: An open source stochastic simulation system for integrated water supply portfolio management and infrastructure investment planning. Environ. Model. Softw. 132, (2020).
Escriva-Bou, A. et al. Water Accounting in Western US, Australia, and Spain: Comparative Analysis. J. Water Resour. Plan. Manag. 146, 04020004–04020004 (2020).
Hanak, E. et al. Replenishing Groundwater in the San Joaquin Valley. https://www.ppic.org/wp-content/uploads/r-0417ehr.pdf (2018).
Alam, S., Gebremichael, M., Li, R., Dozier, J. & Lettenmaier, D. P. Can Managed Aquifer Recharge Mitigate the Groundwater Overdraft in California’s Central Valley? Water Resour. Res. 56, (2020).
Dillon, P. et al. Sixty years of global progress in managed aquifer recharge. Hydrogeol. J. 27, 1–30 (2019).
Kocis, T. N. & Dahlke, H. E. Availability of high-magnitude streamflow for groundwater banking in the Central Valley, California. Environ. Res. Lett. 12, 084009–084009 (2017).
Gold, D. F., Reed, P. M., Gorelick, D. E. & Characklis, G. W. Power and Pathways: Exploring Robustness, Cooperative Stability, and Power Relationships in Regional Infrastructure Investment and Water Supply Management Portfolio Pathways. Earths Future 10, (2022).
Yoon, J. et al. A Typology for Characterizing Human Action in MultiSector Dynamics Models. Earths Future 10, (2022).
Srikrishnan, V. et al. Uncertainty Analysis in Multi‐Sector Systems: Considerations for Risk Analysis, Projection, and Planning for Complex Systems. Earths Future 10, (2022).
Hamilton, A. L., Zeff, H. B., Characklis, G. W. & Reed, P. M. Resilient California water portfolios require infrastructure investment partnerships that are viable for all partners. Earths Future 10, e2021EF002573-e2021EF002573 (2022).
Gorelick, D. E., Gold, D. F., Reed, P. M. & Characklis, G. W. Impact of Inter‐Utility Agreements on Cooperative Regional Water Infrastructure Investment and Management Pathways. Water Resour. Res. 58, (2022).
Herman, J. D., Zeff, H. B., Reed, P. M. & Characklis, G. W. Beyond optimality: Multistakeholder robustness tradeoffs for regional water portfolio planning under deep uncertainty. Water Resour. Res. 50, 7692–7713 (2014).
Bird, R. M. & Slack, E. Financing Infrastructure: Who Should Pay? SSRN Electron. J. (2017) doi:10.2139/ssrn.3083743.
Madani, K. Game theory and water resources. J. Hydrol. (2010) doi:10.1016/j.jhydrol.2009.11.045.
De Souza, S., Medellín-Azuara, J., Lund, J. R. & Howitt, R. E. Beneficiary Pays Analysis of Water Recycling Projects. http://www.waterboards.ca.gov/water_issues/programs/grants_loans/water_recycling/docs/econ_tskfrce/beneficiarypays.pdf (2011).
Giglio, R. J. & Wrightington, R. Methods for apportioning costs among participants in regional systems. Water Resour. Res. 8, 1133–1144 (1972).
Watkins, A. R. Cost Allocation in Urban Infrastructure Funding. J. Urban Plan. Dev. 124, 44–53 (1998).
Office of Governor Gavin Newsom. Governor Newsom Announces $5.1 Billion Package for Water Infrastructure and Drought Response as Part of $100 Billion California Comeback Plan. (2021).
Hanak, E., Jezdimirovic, J., Escriva-Bou, A. & Ayres, A. A Review of Groundwater Sustainability Plans in the San Joaquin Valley (Public comments submitted to the California Department of Water Resources). 1–13 https://www.ppic.org/wp-content/uploads/ppic-review-of-groundwater-sustainability-plans-in-the-san-joaquin-valley.pdf (2020).
Jezdimirovic, J., Hanak, E. & Escriva-Bou, A. What’s the Plan to End Groundwater Overdraft in the San Joaquin Valley? https://www.ppic.org/blog/whats-the-plan-to-end-groundwater-overdraft-in-the-san-joaquin-valley/ (2020).
Hanak, E. et al. Water Stress and a Changing San Joaquin Valley. https://www.ppic.org/content/pubs/report/R_0317EHR.pdf (2017).
Faunt, C. C. & Sneed, M. Water availability and subsidence in California’s Central Valley. San Franc. Estuary Watershed Sci. 13, 0–8 (2015).
Levy, Z. F. et al. Critical aquifer overdraft accelerates degradation of groundwater quality in California’s Central Valley during drought. Geophys. Res. Lett. 1–10 (2021) doi:10.1029/2021gl094398.
Scanlon, B. R., Reedy, R. C., Faunt, C. C., Pool, D. & Uhlman, K. Enhancing drought resilience with conjunctive use and managed aquifer recharge in California and Arizona. Environ. Res. Lett. 11, 035013–035013 (2016).
Friant Water Authority. Subsidence: A critical challenge to Friant-Kern Canal water deliveries. https://static1.squarespace.com/static/58c2eccc15d5db46200ea426/t/5df2e69ea705f61846a258bd/1576199845717/FWA_Subsidence_Challenge_V3_web.pdf (2019).
U.S. Bureau of Reclamation. Friant-Kern Canal Middle Reach Capacity Correction Project. https://www.usbr.gov/mp/docs/fkc-feasibility-report.pdf (2020).
Whisnand, C. Friant-Kern Canal repair process continues with repayment contract. The Porterville Recorder (2021).
Lehner, F. & Deser, C. Origin, importance, and predictive limits of internal climate variability. Environ. Res. Clim. 2, 023001 (2023).
Gershunov, A., Shulgina, T., Ralph, F. M., Lavers, D. A. & Rutz, J. J. Assessing the climate-scale variability of atmospheric rivers affecting western North America. Geophys. Res. Lett. 44, 7900–7908 (2017).
California State Treasurer’s Office. All Issuance CY2019 on 12-22-20. https://www.treasurer.ca.gov/cdiac/datafile/2019.xls (2020).
Hadka, D. & Reed, P. Borg: An auto-adaptive many-objective evolutionary computing framework. Evol. Comput. 21, 231–259 (2013).
Coello Coello, C. A., Lamont, G. B. & Van Veldhuizen, D. A. Evolutionary Algorithms for Solving Multi-Objective Problems. (Springer Science+Business Media, LLC, 2007). doi:10.1046/j.1365-2672.2000.00969.x.
Ghasemizade, M. et al. An Integrated Approach Toward Sustainability via Groundwater Banking in the Southern Central Valley, California. Water Resour. Res. 55, 2742–2759 (2019).
University of California Agriculture and Natural Resources. Kern County: Irrigation Management & Agronomy. https://cekern.ucanr.edu/Irrigation_Management/ (2021).
Fernandez-Bou, A. S. et al. Underrepresented, understudied, underserved: Gaps and opportunities for advancing justice in disadvantaged communities. Environ. Sci. Policy 122, 92–100 (2021).
Hanak, E. et al. Water and the Future of the San Joaquin Valley. https://www.ppic.org/wp-content/uploads/water-and-the-future-of-the-san-joaquin-valley-february-2019.pdf (2019).
Szeptycki, Leon F., Forgie, Julia, Hook, Elizabeth, Lorick, Kori, & Womble, Philip. Environmental water rights transfers: A review of state laws. https://waterinthewest.stanford.edu/sites/default/files/WITW-WaterRightsLawReview-2015-FINAL.pdf (2015).
Womble, P. & Hanemann, W. M. Legal Change and Water Market Transaction Costs in Colorado. Water Resour. Res. 56, (2020).
Gonzalez, P. et al. Southwest. in Impacts, Risks, and Adaptation in the United States: Fourth National Climate Assessment, Volume II (eds. Reidmiller, D. R. et al.) 1101–1184 (U.S. Global Change Research Program, 2018). doi:10.7930/NCA4.2018.CH25.
AghaKouchak, A. et al. Anthropogenic Drought: Definition, Challenges, and Opportunities. Rev. Geophys. 59, e2019RG000683-e2019RG000683 (2021).
Moallemi, E. A., Kwakkel, J., de Haan, F. J. & Bryan, B. A. Exploratory modeling for analyzing coupled human-natural systems under uncertainty. Glob. Environ. Change 65, 102186–102186 (2020).
Lempert, R. J. A new decision sciences for complex systems. Proc. Natl. Acad. Sci. U. S. A. 99, 7309–7313 (2002).
Kasprzyk, J. R., Nataraj, S., Reed, P. M. & Lempert, R. J. Many objective robust decision making for complex environmental systems undergoing change. Environ. Model. Softw. 42, 55–71 (2013).
Basdekas, L. & Hayslett, R. Improving Tradeoff Understanding in Water Resource Planning Using Multi-Objective Search. https://www.waterrf.org/research/projects/improving-tradeoff-understanding-water-resource-planning-using-multi-objective (2021).
Smith, R., Kasprzyk, J. & Dilling, L. Testing the potential of Multiobjective Evolutionary Algorithms (MOEAs) with Colorado water managers. Environ. Model. Softw. 117, 149–163 (2019).
Smith, R. et al. Decision Science Can Help Address the Challenges of Long‐Term Planning in the Colorado River Basin. JAWRA J. Am. Water Resour. Assoc. (2022) doi:10.1111/1752-1688.12985.
Moallemi, E. A. et al. Evaluating Participatory Modeling Methods for Co-creating Pathways to Sustainability. Earths Future 9, (2021).
Gorelick, D. E. et al. Water Supply Infrastructure Investments Require Adaptive Financial Assessment: Evaluation of Coupled Financial and Water Supply Dynamics. J. Water Resour. Plan. Manag. 149, (2023).
Hamilton, A. L., Characklis, G. W. & Reed, P. M. From Stream Flows to Cash Flows: Leveraging Evolutionary Multi‐Objective Direct Policy Search to Manage Hydrologic Financial Risks. Water Resour. Res. 58, e2021WR029747-26 (2022).
Brand, M. W. et al. Environmental Impact Bonds: a common framework and looking ahead. Environ. Res. Infrastruct. Sustain. 1, 023001–023001 (2021).
Larson, W. M., Freedman, P. L., Passinsky, V., Grubb, E. & Adriaens, P. Mitigating corporate water risk: Financial market tools and supply management strategies. Water Altern. 5, 582–602 (2012).
Dobbin, K. B. & Lubell, M. Collaborative Governance and Environmental Justice: Disadvantaged Community Representation in California Sustainable Groundwater Management. Policy Stud. J. 49, 562–590 (2021).
Fletcher, S. et al. Equity in Water Resources Planning: A Path Forward for Decision Support Modelers. J. Water Resour. Plan. Manag. 148, (2022).
Osman, K. K. & Faust, K. M. Toward Operationalizing Equity in Water Infrastructure Services: Developing a Definition of Water Equity. ACS EST Water 1, 1849–1858 (2021).
Avelino, F. Theories of power and social change. Power contestations and their implications for research on social change and innovation. J. Polit. Power 14, 425–448 (2021).
Kern Water Bank Authority. Frequently Asked Questions. https://www.kwb.org/faqs/ (2021).
Bracken, C., Rajagopalan, B. & Zagona, E. A hidden Markov model combined with climate indices for multidecadal streamflow simulation. Water Resour. Res. 50, 7836–7846 (2014).
Hadjimichael, A. et al. Defining Robustness, Vulnerabilities, and Consequential Scenarios for Diverse Stakeholder Interests in Institutionally Complex River Basins. Earths Future 8, 1–22 (2020).
Thyer, M. & Kuczera, G. A hidden Markov model for modelling long-term persistence in multi-site rainfall time series 1. Model calibration using a Bayesian approach. J. Hydrol. 275, 12–26 (2003).
McPhail, C. et al. Robustness Metrics: How Are They Calculated, When Should They Be Used and Why Do They Give Different Results? Earths Future 6, 169–191 (2018).
Quinn, J. D., Reed, P. M., Giuliani, M. & Castelletti, A. Rival framings: A framework for discovering how problem formulation uncertainties shape risk management trade-offs in water resources systems. Water Resour. Res. 53, 7208–7233 (2017).
Reed, P. M., Hadka, D., Herman, J. D., Kasprzyk, J. R. & Kollat, J. B. Evolutionary multiobjective optimization in water resources: The past, present, and future. Adv. Water Resour. 51, 438–456 (2013).
Hadka, D. & Reed, P. Diagnostic assessment of search controls and failure modes in many-objective evolutionary optimization. Evol. Comput. 20, 423–452 (2012).
Gupta, R. S., Hamilton, A. L., Reed, P. M. & Characklis, G. W. Can modern multi-objective evolutionary algorithms discover high-dimensional financial risk portfolio tradeoffs for snow-dominated water-energy systems? Adv. Water Resour. 145, 103718–103718 (2020).
Zatarain Salazar, J., Reed, P. M., Herman, J. D., Giuliani, M. & Castelletti, A. A diagnostic assessment of evolutionary algorithms for multi-objective surface water reservoir control. Adv. Water Resour. 92, 172–185 (2016).
Singh, L. A. et al. Low cost satellite constellations for nearly continuous global coverage. Nat. Commun. 11, 1–7 (2020).
Woodruff, M. J., Reed, P. M. & Simpson, T. W. Many objective visual analytics: Rethinking the design of complex engineered systems. Struct. Multidiscip. Optim. 48, 201–219 (2013).
Giuliani, M., Quinn, J. D., Herman, J. D., Castelletti, A. & Reed, P. M. Scalable Multiobjective Control for Large-Scale Water Resources Systems under Uncertainty. IEEE Trans. Control Syst. Technol. 26, 1492–1499 (2018).
Zatarain Salazar, J., Reed, P. M., Quinn, J. D., Giuliani, M. & Castelletti, A. Balancing exploration, uncertainty and computational demands in many objective reservoir optimization. Adv. Water Resour. 109, 196–210 (2017).
Hamilton, Andrew L., Reed, Patrick M., Gupta, Rohini S., Zeff, Harrison B., & Characklis, Gregory W. ahamilton144/CALFEWS: Release for paper submission (MORDM_experiment_paper1_v1.0.1). https://doi.org/10.5281/zenodo.8118092 (2023).

There is NO Competing Interest.

hamiltonCApartnershipDesignSI06302023.docx

Download PDF

Journal Publication

published 27 Aug, 2024

Read the published version in Nature Communications →

Version 1

posted

You are reading this latest preprint version

Resilient water infrastructure partnerships in institutionally complex systems face challenging supply and financial risk tradeoffs

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Results

Water supply infrastructure partnership design

Understanding diverse partnership design options

Key performance tradeoffs for water supply investment partnerships

Navigating uncertainty and heterogeneity in tradeoffs

Regret of the current baseline partnership design

Discussion

Methods

Candidate infrastructure investments

Infrastructure partnership evaluation

Multiobjective intelligent search

Selection of Status Quo & Compromise Partnerships

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1