Sustainable Flood Risk and Stormwater Management in Blue-Green Cities; an Interdisciplinary Case Study in Portland, Oregon

Research Impact Statement : Interdisciplinary research in Portland, Oregon USA, provides an enhanced evidence base to justify adoption of Blue-Green Infrastructure (BGI) for sustainable ﬂood risk and stormwater management. ABSTRACT: Blue-Green Infrastructure (BGI) is recognized as a viable strategy to manage stormwater and ﬂood risk, and its multifunctionality may further enrich society through the provision of multiple cobeneﬁts that extend far beyond the hydrosphere. Portland, Oregon, is an internationally renowned leader in the implementation of BGI and showcases many best practice examples. Nonetheless, a range of interdisciplinary barriers and uncertainties continue to cloud decision making and impede wider implementation of BGI. In this paper, we synthesize research conducted by the “Clean Water for All” (CWfA) research project and demonstrate that interdisciplinary evaluation of the beneﬁts of Portland’s BGI, focusing on green street bioswales and the East Lents Floodplain Restoration Project, is essential to address biophysical and sociopolitical barriers. Effective interdisciplinary approaches require sustained interaction and collaboration to integrate disciplinary expertise toward a common problem-solving purpose, and strong leadership from researchers adapt at spanning disciplinary boundaries. While the disciplinary differences in methodologies were embraced in the CWfA project, and pivotal to providing evidence of the disparate beneﬁts of multifunctional BGI, cross-disciplinary engagement, knowledge coproduction, and data exchanges during the research process were of paramount importance to reduce the potential for fragmentation and ensure research remained integrated.


INTRODUCTION
Municipalities across the United States (U.S.) are increasingly using Green Infrastructure (GI) as a cost-effective measure to manage stormwater and improve water quality (McPhillips and Matsler 2018; Shandas et al. 2020). Several factors have helped drive the transition from a traditional approach using centralized gray stormwater management infrastructure toward more decentralized GI facilities that retain or reuse stormwater onsite. This includes; the federal endorsement of GI as a primary stormwater management technique (USEPA 2008); the recognition that GI can deliver a variety of environmental, societal and economic cobenefits and ecosystem services (Kremer et al. 2016), and; the need for water resource management systems that are adaptive to change and resilient to extremes (Rijke et al. 2013). Exemplar GI programs further demonstrate the potential for GI to meet a range of challenges, for example, in Portland, Seattle, New York City, North Carolina, and San Francisco (City of Seattle 2015; Kremer et al. 2016;Trogrli c et al. 2018;Shandas et al. 2020) and Philadelphia, the first city in the U.S. to attempt an entirely green approach to meeting federal regulations (Fitzgerald and Laufer 2016).
The city of Portland, Oregon, has one of the oldest and most successful GI programs in the U.S. and is an internationally renowned leader in the implementation of GI to manage stormwater for the promotion of sustainable development practices, climate change adaptation, and improved liveability (Lukes and Kloss 2008;Rottle 2015). An excessive burden on the City's drainage system and repeat discharge of sewage into the Willamette River due to frequent combined sewer overflow (CSO) events in the 1990s led the Portland municipal authority to search for an alternative to using conventional "gray" infrastructure to supplement the "Big Pipe" Project needed to substantially reduce the frequency of CSOs (BES 2019a, accessed November 2019). In its "Grey to Green" initiative, the City invested widely in GI implementation to help alleviate loadings on the piped infrastructure system and reduce adverse impacts on urban watercourses. These ongoing efforts have, to date, delivered over 2,000 street bioswales, more than 600 ecoroofs and tens of thousands of street trees. The City of Portland, and notably the Bureau of Environmental Services (BES), have promoted a variety of best management practices and embraced Blue-Green principles of reintroducing elements of the natural water cycle into urban environments. BES have invested in widespread culvert removal, purchasing of properties at high flood risk from willing sellers, reconnecting and restoring urban streams and floodplains, and reintroducing native vegetation and wildlife (including beaver [Castor canadensis] and several species of Pacific salmonids [Oncorhynchus]) (BES 2019b, accessed November 2019. Portland is clearly progressing toward becoming a Blue-Green City; where a naturally oriented water cycle that mimics predevelopment hydrology is created through reduced imperviousness, increased infiltration, enhanced surface storage, reintroduction of native water retentive plants, restoration of urban watercourses, and improvements to water quality and aquatic environments (Novotny et al. 2010). In a Blue-Green City, the hydrological and ecological values of the urban landscape are protected while providing resilient and adaptive measures to deal with flood events and improve the amenity of the city (Hoyer et al. 2011). Blue infrastructure includes the flowing waterways, ponds, wetlands, and wet detention basins that exist within the drainage network. GI is an approach to wet weather management that uses soils and vegetation to utilize, enhance and/or mimic the natural hydrological cycle processes of infiltration, evapotranspiration and reuse (USEPA 2019, accessed December 2019. In this context, we define the Blue-Green Infrastructure (BGI) adopted in pursuit of Blue-Green ideals as a combination of the two; an interconnected network of natural and designed landscape components that includes blue and green spaces, and green spaces designed to turn "blue" during rainfall and flood events.
BGI is increasingly recognized as enriching society through the provision of multiple cobenefits beyond water and flood risk management, and its multifunctionality is fundamental to its growing appeal. In addition to stormwater abatement, BGI can support climate change adaption (to, e.g., extreme storms, heatwaves, and droughts); lengthen the useful life of aging, gray infrastructure; improve wildlife, biodiversity, air, and water quality; increase landscape connectivity and access to greenspace; improve health and well-being, and; create attractive landscapes that enhance quality of place (Demuzere et al. 2014;Netusil et al. 2014;Kabisch et al. 2016;Fenner 2017;Hoang et al. 2018;Venkataramanan et al. 2019). Despite this, BGI is typically implemented from the perspective of a single benefit, often stormwater management (Kabisch et al. 2016).
An interdisciplinary approach is therefore needed to comprehensively evaluate the multiple cobenefits of Blue-Green approaches to flood and water management that extend beyond the hydrosphere, and facilitate the development of strategies to overcome the barriers to implementation that are known to encompass the biophysical, social, and political spheres . The natural and societal components of stormwater management interact in complex ways and desired societal results are rarely achieved by addressing the scientific aspects in isolation (Morss et al. 2005). Interdisciplinary research emphasizes interaction and joint working to integrate disciplinary expertise toward a common problem-solving purpose and avoids partial framing of key societal challenges (Bruce et al. 2004;Lowe and Phillipson 2006). The objective is to create knowledge that is solution-oriented, socially robust and more easily adopted by policy makers and practitioners (Gibbons 1999).
However, the development of an interdisciplinary approach challenges our disciplinary training and ways of thinking (Klein 2008). In this paper, we introduce the interdisciplinary approach adopted by the "Clean Water for All" (CWfA) research project (2014)(2015), that brought together disciplinary scientists from academic institutions in the UK, U.S., and China, under the themes of Water Engineering, Resilience, and Sustainability, to tackle questions posed by the interactions between humans and landscapes. Through collaborative research, this project coproduced solution-oriented, transferrable knowledge, focusing on evaluating the multiple cobenefits delivered by BGI in Portland, Oregon. As a result, longterm collaborative partnerships were formed between academics working on; the UK Engineering and Physical Sciences Research Council (EPSRC) funded "Blue-Green Cities Research Project" (www.bluegree ncities.ac.uk) (Lawson et al. 2014); the U.S. National Science Foundation (NSF) funded "Portland-Vancouver Urban Long-term Research Area research project" (www.fsl.orst.edu/eco-p/ultra) (Chang et al. 2014), and; academics at the University of Nottingham Ningbo China, researching sustainable flood risk and urban water management in Chinese "Sponge Cities" .
The overarching aim of the interdisciplinary CWfA research project was to build a substantive evidence base to justify adoption of BGI, while developing a greater understanding of design modification to co-optimize the multiple benefits of a conurbation choosing to become a Blue-Green City. The hypothesis underpinning the research is that implementing multifunctional BGI that delivers cobenefits can help multiple stakeholder organizations and departments meet their strategic objectives not only in relation to flood and water management, but also climate change adaption, urban heat island reduction, greenspace development, biodiversity improvement, public health and well-being, and recreation and amenity.
We frame this paper around the interdisciplinary barriers, or "Relevant Dominant Uncertainties" (RDUs), which continue to cloud decision making and impede wider implementation of BGI in Portland (Theme 1), and show how research into flow and suspended sediment dynamics (Theme 2);sediment-related, heavy metal contaminants, and river habitats (Theme 3); community perceptions and acceptance of BGI (Theme 4), and; multiple cobenefit evaluation (Theme 5), can reduce some of these uncertainties. The insights of our disciplines were used in the context of stormwater management in Portland to coproduce the evidence base to support BGI as a means of enhancing the environment and society. The disciplinary differences in methodologies were embraced but of primary importance was the cross-discipline engagement and data exchanges during the research process to marry the interdisciplinary appeal with the disciplinary mastery (Klein 2008). We synthesize the key research findings of the CWfA research project and illustrate the importance of multidirectional communication, sustained interactions among researchers, and an iterative approach to coproduce knowledge and tools in delivering interdisciplinary research outputs (Morss et al. 2005). We close with recommendations for researchers and practitioners involved in interdisciplinary flood risk and stormwater management projects.

STUDY LOCATION
Portland is situated near the foothills of the Tualatin Mountains, at the confluence of the Columbia and Willamette Rivers, west of the Cascade Range, Oregon. The maritime climate is typical of the Pacific Northwest, being marked by cool, wet winters and hot, dry summers. Average annual precipitation in Portland is about 1,400 mm (Velpuri and Senay 2013), which generates approximately 450,000 m 3 /yr of stormwater runoff (BES 2015).
Clean Water for All research focused on the City of Portland administrative area and, particularly, Johnson Creek, a 42 km long tributary of the Willamette River ( Figure 1). The 134 km 2 Johnson Creek watershed is subdivided into numerous, smaller rural, peri-urban, and urban catchments of varying size (0.006-0.7 km 2 ) and comprises mixed land-use, with forest, agriculture, and rural residential characterizing the upper watershed and urban development in the middle and lower reaches (Sonoda et al. 2001). In approximately 75% of the catchment, stormwater is conveyed to Johnson Creek through a piped network and a series of outfalls, without treatment. During the 1980s, high concentrations of heavy metals, the presence of E. coli (Escherichia coli), and elevated stream temperatures led to many reaches of Johnson Creek being classified as "impaired" under article 303 (d) of the Clean Water Act (USEPA 1972). BES, with assistance from the Johnson Creek Watershed Council and multiple other private organizations, subsequently invested in extensive river restoration, riparian planting, and installation of BGI to gradually restore and reconnect the creek to its remaining floodplains, reduce local flood risk (JCWC 2012), and improve water quality through, for example, installation of set-back stormwater outfalls and pocket wetlands designed to trap pollutants (Janes et al. 2016). A total of 209 restoration projects were undertaken between 1990 and 2014, producing positive impacts ranging from the return of native salmon (Oncorhynchus) to increases in the values of nearby residential properties (Jarrad et al. 2018). The East Lents Floodplain Restoration Project is a good example of JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION urban stream restoration and floodplain re-creation. This project regenerated the 28 ha "Foster Floodplain Natural Area" in lower Johnson Creek (Figure 1) (BES 2013, accessed November 2019. Briefly, this extensive project (implemented over a 15 year period) was designed to reduce "nuisance flooding" (by events with return periods between <1 and 10 years) that had for decades inundated low-lying homes and businesses adjacent to Johnson Creek and blocked Foster Road, causing widespread traffic disruption and associated impacts on local residents, commuters, and businesses (The Oregonian 2012, accessed March 2019). Families were moved out of 66 private properties within the 1-in-100-year floodplain, under the Willing Seller Land Acquisition Program, to re-create a natural basin for flood storage, conservation, and recreation.
During the last two decades, the City of Portland invested $1.4 billion in CSO control through "Big Pipe" projects (East Side, West Side and Columbia Slough), aimed at improving water quality in the Willamette River. Pipe construction completed in 2011 and has eliminated 94% of CSOs to the Willamette River and 99% of CSOs to the Columbia Slough (BES 2019a, accessed November 2019). The monetary cost, and indeed the size of the required pipes, would have been much larger were it not for parallel investment in numerous GI projects, especially those implemented under the $55 million "Grey to Green" program conducted between 2008(BES 2010. This program included several Cornerstone Projects that help keep billions of gallons of stormwater runoff out of the combined sewer system annually, significantly contributing to the huge reduction in CSOs (BES 2019c, accessed November 2019), while generating a wide range of environmental, ecological, and social cobenefits. For example, approximately 1.2 billion gallons of stormwater runoff reduction can be attributed to the Downspout Disconnection project (1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011) (Figure 2a), which provided financial incentives to residents to encourage them to disconnect their roof drainage from the combined sewer system and redirect runoff into their gardens. In total, 56,000 disconnections were made.
The City of Portland has an impressive portfolio of GI assets, including street trees, ecoroofs, deculverted watercourses, areas of re-introduced native vegetation, and restored and reconnected streams and floodplains. Over 2,000 "green streets" transform impervious street surfaces into landscaped, green spaces featuring stormwater curb extensions, bioswales, and street planters that manage stormwater runoff at source (BES 2019c, accessed November 2019) (Figure 2b-2d). Through these and other efforts, Portland has begun to make the urban water cycle more like the natural hydrological cycle, while meeting targets for urban regeneration, growth, and promoting the esthetic appeal of the GI.

INTERDISCIPLINARY RESEARCH THEMES AND PROCESS
The CWfA research project adopted an interdisciplinary approach with the integration of disciplines driven by interactions and joint-working amongst CWfA consortium members motivated by a common, problem-solving intent. Relationships between the research themes were identified as being interactive and complex, and so a range of methodologies and routes to problem solving were developed and applied. However, as the goals were to jointly address the research questions and coevolve understanding and knowledge, it was essential to maintain close integration between members of the consortium. As the project progressed, the team developed a research program in which disciplines were not the focal points. Instead, the inputs of the researchers were distributed and blended to ensure that each theme had the disciplinary-based experiences, skills, and competencies necessary to address the specific research question at hand.
The CWfA research was performed through a carefully sequenced series of actions, arranged in five themes that ran concurrently, providing a stable framework for executing the project (Figure 3). The framework was designed to address discrete elements of the complex issues surrounding urban flood risk and stormwater management. It further demonstrates the range of disciplines necessarily involved in BGI projects that move beyond analyses of their hydrologic, hydraulic, and engineering performance to identify and evaluate the multiple cobenefits that may accrue. The specific methods used to meet the objectives of each research theme were not necessarily wholly innovative within a disciplinary context; rather when taken collectively they result in innovative research (i.e., the sum is greater than the parts).
Theme 1 sought to identify barriers that continue to hinder implementation of BGI, a topic central to the project as uncertainties impact all aspects of BGI research. Geographically, this theme spans the entire Portland administrative area. Themes 2, 3, and 5 focused on the East Lents Floodplain Restoration Project, which was selected because its implementation included large-scale floodplain reconnection, urban stream restoration, and creation of multifunctional BGI. Theme 4 investigated public attitudes to bioswales, which are a fundamental component of Portland's Green Streets Policy.
Interdisciplinary research was made possible by strong linkages between research themes through which knowledge and expertise were shared. For example, particle size distribution data collected in Johnson Creek as part of Theme 3 (sediment, contaminants, morphology, and riparian restoration) were also used in Theme 2's sediment transport model, to simulate sediment dynamics in Johnson Creek and the reconnected East Lents floodplain. The questions developed by Theme 4 to determine community perceptions of bioswales received input from the interdisciplinary team who highlighted functional and esthetic aspects that may influence public attitudes. We continue to refer back to the importance of interdisciplinary research as we synthesize the key outputs from each research theme. Brief descriptions of the methods are given, and the source databases, reports, and publications are signposted. The synthesis presented here shows the necessity of interdisciplinary research to tackle the myriad barriers to widespread implementation of BGI, and to codevelop the evidence base that demonstrates the multiple benefits of BGI.
Theme 1. Identifying and Dealing with Uncertainty as a Barrier to the Adoption of BGI Implementation of BGI projects requires alignment of planning frameworks, engineering design, construction practices, maintenance strategies, ownership/adoption agreements, and community acceptability. Despite the successes of the "Grey to Green" transition, diversity of BGI already built in Portland, and proven advantages of Blue-Green over equivalent gray infrastructure (Casal-Campos et al. 2015), uncertainties remain within each stage of the implementation process that collectively hamper further innovation in stormwater management. In Portland, known uncertainties relate to the hydrologic performance of BGI and a lack of confidence among decision makers that BGI will be publicly acceptable. Typically, socioinstitutional barriers are believed to exert more of a negative influence on sustainable drainage decision making than hydrological factors (Carlet 2015). These specific uncertainties nest within broader, interdisciplinary challenges that affect all infrastructure projects, including the impact of climate change, delivery of socially equitable schemes, and communicating complex technical and planning issue to communities.
Theme 1 investigated these uncertainties through a series of semi-structured interviews with institutional stakeholders in the City of Portland (hereafter referred to as "City of Portland stakeholders"). Twelve respondents in total were interviewed, from the BES, Bureau of Planning and Sustainability, Bureau of Transport, and Metro. The Relevant Dominant Uncertainty (RDU) approach was applied to identify known unknowns most likely to limit the capacity for decision makers to make more informed choices (Smith and Petersen 2014), in this case, around BGI. The full methods and outputs are presented in Thorne et al. (2018) and briefly described here. First, the CWfA researchers involved in this theme individually listened to the recorded interviews and/ or read the transcripts in order to identify and rank the RDUs (based on frequency and intensity with which each interviewee referred to each RDU). The findings were then shared and discussed through a Delphi-like sequence of exchanges and debates until consensus regarding the RDU classifications and rankings were reached. In addition to identifying 15 RDUs, subdivided into biophysical and sociopolitical RDUs (Figure 4), 13 concerns (issues respondents are worried about) and 11 challenges (which result from aforementioned concerns) were identified. Sociopolitical RDUs were found to exert the strongest negative influences on BGI decision making in Portland. The strongest RDU resulted from lack of confidence on the part of City officials and technical leaders that the public and their elected representatives would continue to understand, support, and be willing to pay for BGI. Fewer biophysical RDUs were identified by the stakeholders interviewed, with the most prominent being the impact of climate change on the current and future performance of BGI.
Overcoming Uncertainty as a Driver for Interdisciplinary Research. The range of biophysical and sociopolitical RDUs identified by the City of Portland stakeholders show the importance of an interdisciplinary approach to generate the evidence that is clearly needed to reduce these uncertainties and support widespread implementation of BGI. This is because the RDUs are not unique to a single research discipline. Decision makers must have greater confidence that BGI assets are both scientifically sound, and understood, accepted, and supported by communities and their elected representatives. This means addressing the biophysical RDUs related to asset performance and maintenance, modeling, and downscaling climate projections, through collaborative research conducted by teams of professional researchers and BGI practitioners ). Central research themes should be the development of improved technical and scientific analyses of BGI functionality, coupled with long-term, postproject monitoring, and evaluation to establish the maintenance regimes and adaptive management needed to ensure that BGI functions at its design capacity throughout its useful life. Research conducted under Themes 2, 3, and 5 addressed some of these biophysical RDUs. In parallel, research is required to address socioeconomic RDUs related to, for example, public preferences, stewardship, effective inter-agency working and economic resilience. Determining public perceptions of BGI, as investigated under Theme 4 using FIGURE 4. Biophysical and sociopolitical challenges and concerns that constitute significant "Relevant Dominant Uncertainties" (RDUs) impeding implementation of BGI in Portland. RDUs are presented in order of importance, as identified by the interview respondents, beginning at the top and moving clockwise. Sociopolitical RDUs (e.g., public preferences, stewardship, and equitable delivery of BGI) have a greater impact on Portland decision making than biophysical uncertainties. CC = climate change.
Point of Opportunity Interaction methods to identify attitudes toward bioswales, is the first step to understanding local awareness and satisfaction which ultimately impact the potential for effective and longterm local stewardship.

Theme 2. Floodplain Restoration in East Lents
A biophysical RDU that impedes implementation of BGI in Portland (and other cities) relates to concerns about long-term performance and service provision, particularly as assets age and environmental loadings change. Using the East Lents project as an example, Theme 2 used a two-dimensional hydraulic, sediment, and morpho-dynamic model (Guan et al. 2015) to investigate concerns expressed by City of Portland stakeholders regarding uncertainties surrounding how the lowered, reconnected floodplain functions during a range of storm events with different peak discharges. The model also forecast sediment deposition and retention within the floodplain, evaluating the risk that gradual accumulation of storm-generated sediments over a period of decades might significantly reduce the future capacity of the floodplain to detain stormwater (Ahilan et al. 2018).
Although designed primarily to reduce the frequency of flooding by events with short return periods, the restored East Lents floodplain was found to reduce downstream flood peaks for the modeled 500year flood event (115 m 3 /s) by over 25% (Figure 5). This challenges the common assumption that floodplain reconnection is only effective during low and medium return period flow events (Woltemade and Potter 1994). The floodplain was also found to act as an effective sediment sink, trapping~20%-30% of the sediment generated further upstream and considerably reducing the cumulative sediment loading into the Willamette River. However, the volume of deposited sediment is small when compared with the large storage capacity of the floodplain, showing that there are minimal long-term impacts on downstream flood resilience.
Theme 2 research addressed a further concern of the City of Portland stakeholders regarding infrastructure performance and maintenance requirements. The fact that the elevation of the floodplain will not substantially change over a range of flood events suggests that minimal maintenance is required to maintain functionality to the design standard, allowing the scheme to deliver its environmental and social cobenefits without requiring significant investment in frequent or complex maintenance activities.
The disciplinary expertise of the CWfA consortium in hydraulic, sediment and morpho-dynamic modeling has provided information to overcome several biophysical RDUs and demonstrated the ability of BGI, represented here by a restored floodplain, to manage local flood risk and retain sediment, without requiring extensive maintenance regimes. A final concern stated by City of Portland stakeholders referred to the risk that floodplains may become pollutant hotspots owing to the deposition of pollutants (e.g., heavy metals) attached to fine suspended sediment particles. Ahilan et al. (2018) found no evidence to suggest that pollutant hotspots occur within the East Lents floodplain, and shared these modeling results with the Theme 3 team, who investigated this further by contributing knowledge from a different disciplinary standpoint (water quality and ecology).

Theme 3. Sediment, Contaminants, Morphology and Riparian Restoration
The interdependencies among stream hydrology, hydraulics, sediment transport, and water quality are a prime example of where interdisciplinary research is necessary to fully explain processes and trends observed in the field and comprehensively evaluate the environmental costs and benefits of floodplain reconnection. The sources and delivery of contaminated sediment to Johnson Creek, and the ability of natural flood management approaches (such as setback outfalls that discharge into BGI features prior to entering the channel) to remove in-channel pollutants, were further investigated in Theme 3.
It is estimated that~85% of heavy-metal pollutants derived from urban surface (e.g., from vehicles, construction, commercial, and industrial sources, and degradation of old infrastructure) are conveyed into urban watercourses adsorbed to fine (D < 500 µm) sediment particles (Wei and Yang 2010). In mixed rural-urban catchments, heavy-metal concentrations typically increase with distance downstream and the concomitant increase in urban development, traffic, and piped drainage infrastructure that generate and transport pollutants into watercourses (Sharley et al. 2016). However, the relationship between sediment heavy-metal concentrations and subbasin characteristics in Johnson Creek are more complex. Research conducted during the CWfA project, and detailed in full in Chang et al. (2018), has shown that land-use is less clearly segregated; a range of point and nonpoint sources and delivery paths of sediment-associated pollutants are influential.
Sediment samples were collected from 37 outfalls along the main stem of Johnson Creek, and analyzed by Inductively Coupled Plasma Optical Emission Spectrometry for five key heavy metals; copper, zinc, lead, chromium, and cadmium (full methods are given in Chang et al. 2018). Contributing subcatchments for the outfalls were delineated using the ArcHydro tool in ArcGIS 10.4 (Environmental Systems Research Institute, Redlands, California). Catchment variables derived for each subcatchment were based on 2011 National Land Cover Data, hydrologic soil groups, and selected topographic parameters (e.g., elevation and slope) derived from a 10-m Digital Elevation Model (USGS 2015).
Contrary to expectations, heavy-metal concentrations in stream bed samples were not found to increase downstream ubiquitously. In fact, copper, chromium, and lead concentrations were elevated in both upstream reaches and in downstream reaches near listed clean-up sites (Figure 6a-6c). This suggests that point industrial sources play a key role in heavy-metal pollution in Johnson Creek, in addition to nonpoint sources such as traffic. Only zinc concentrations demonstrated a statistically significant relationship with variables defining catchment characteristics (Figure 6d). Chang et al. (2018) infer that the complicated processes of mobilizing and delivering sediment-bound heavy metals in catchments like Johnson Creek, where subcatchment landuse is spatially heterogeneous often down to subdecimeter scales, requires a more nuanced relationship than is currently assumed. This research has illustrated to City of Portland stakeholders that subtle changes in land-use patterns, stormwater runoff pathways (and proportion of runoff drained via pipe networks), outfall designs and BGI interventions can significantly affect the sources, pathways and delivery of heavy metals in mixed use watersheds such as Johnson Creek. Sharing of data and effective multidirectional communication, between researchers working under Themes 2 and 3, aided the CWfA research team in their evaluation of the delivery paths of sediment-associated pollutants, and subsequent interpretation of the efficacy of Johnson Creek's BGI to improve water quality.
Mitigating the Impact of Contaminated Sediment in Stormwater Discharge through Stormwater Management Practices. Stormwater management practices that promote infiltration and pollutant removal by routing stormwater through BGI, natural flood management structures or restored riparian zones are widely used in Johnson Creek yet many City of Portland stakeholders expressed uncertainty over how (and by whom) BGI performance and service provision would be maintained in order to deliver these benefits to water quality. Janes et al. (2016) investigated the effectiveness of setback outfalls, an example of BGI whereby piped stormwater is discharged into a wetland or swale before reaching the watercourse, to remove stormwater pollutants. This combines disciplinary research into ecological performance with water quality and hydrology.
Pollutant concentrations were found to be higher in setback outfalls when compared with background levels (upstream) due to the increased deposition of contaminated sediment and greater uptake of dissolved, or bioavailable bound contaminants, by aquatic organisms. However, recent research into phytoremediation in swales (e.g., Leroy et al. 2016) suggests that there is limited potential for long-term concentration of contaminated sediment in setback wetlands and swales, potentially mitigating the need for extensive maintenance of setback outfalls by the City of Portland and project partners. This, in line with findings reported under Theme 2 regarding the minimal maintenance required to maintain functionality of the East Lents floodplain, suggests that BGI assets in Portland have the potential to effectively manage stormwater and improve water quality, without significant investment in complex maintenance regimes. This negates some of the sociopolitical uncertainty surrounding costs of maintaining BGI (identified under Theme 1).
Channel modification and quality of the river habitat were also found to influence levels of sediment contamination in Johnson Creek by impacting pollutant removal efficiency. The UK River Habitat Survey (RHS) assessment method was used to determine the level of habitat diversity, and habitat modification (or artificiality), of four diverse reaches (Figure 7). Utilizing the RHS scoring methodology to assess the benefits of channel restoration on water quality at stormwater outfalls is a novel application of this method. It provides a semi-quantitative assessment of the level of modification that may be used to infer the ability of the reach to reduce pollutant concentrations around stormwater outfalls. In practice, this assessment may highlight locations where restoration works may be the most effective, helping river managers focus their investments to achieve the best possible outcomes, which helps address some of the uncertainties outlined in Theme 1. Reaches adjacent to stormwater outfalls with lower habitat modification scores were significantly correlated with greater removal efficiency of several key pollutants (Fe, Ba, Sn, Mg, P, K) (Janes et al. 2016). This reiterates the key relationship between ecological, hydrological and hydrodynamic processes in Johnson Creek in determining water and sediment quality.

Theme 4. Community Perceptions of BGI -Bioswales
Theme 4 moved away from biophysical uncertainties to investigate some of the key sociopolitical issues concerning BGI design, implementation, and maintenance, focusing on bioswales. Sociopolitical RDUs, identified by City of Portland stakeholders and acknowledged as inhibiting further implementation of BGI, include uncertainties in forecasting future social conditions, public preferences, and the lack of confidence among decision makers that beneficiary communities will value, accept, and support BGI schemes (see Theme 1). The perceptions that residents and communities have of BGI has a direct impact on their understanding of the functionality and their willingness to pay for such assets (Everett et al. 2015). Residents' preferences for bioswales, as an example of a highly visible BGI intervention and key component of the citywide GI program (Figure 8a, 8b), were evaluated in Theme 4, using Point of Opportunity Interaction methods to collect information from 45 respondents (Everett et al. 2018). Briefly, this involved approaching people directly outside their house in order to include individuals who might not otherwise volunteer to be interviewed. Respondents were asked to talk freely around twelve key questions in a relaxed, conversational environment. Several factors were found to influence residents' appreciation and acceptance of bioswales, including their awareness of the asset and its functionality; their community values (e.g., environmental attitudes), and; site-specific physical and esthetic characteristics including plant choice, maintenance regime and level of perceived "mess" and littering (Everett et al. 2018). Maintenance was a primary concern, echoing the City of Portland stakeholders with regard to questions of maintenance cost and who pays; If you want the public to embrace them you need to keep them looking good and keep them functional, but in order to do that you need to spend money (Water Resource Engineer, BES, Thorne et al. 2018).
Several residents expressed willingness to maintain the bioswales themselves (i.e., to improve esthetics), but such uncoordinated efforts may reduce the ability of the assets to manage water quantity and quality to the design standard. This demonstrates disconnect between bioswale characteristics (e.g., plant choice) that residents prefer and the characteristics that deliver the required (optimal) functionality. Interdisciplinary partnerships between public and private stakeholders that consider both residents' preferences regarding landscape design, and knowledge of how to design bioswales to ensure maximum water retention and bioremediation potential, could address this disconnect and develop multifunctional, mutually beneficial assets. Improving local awareness and satisfaction of bioswales through greater consultation and coconstruction of solutions during the development phase will help overcome the challenge of effective stewardship (Everett et al. 2018). Engagement was a key component of the City's Tabor to the River program and positively impacted residents' perceptions and support for bioswales (Shandas et al. 2010;Church 2015). Such collaborations between residents, communities, nonprofit organizations and City of Portland agencies, such as BES, could greatly increase local stewardship, as exemplified by the volunteer Green Street Stewards Program. This encourages people to clear out trash and debris to improve BGI functionality, reduces the likelihood of untrained residents inadvertently damaging the bioswales (Figure 8c), and facilitates a cost reduction for the City through reducing maintenance costs. Signs explaining when incomplete bioswales will be planted (Figure 8d) also help manage residents' expectations and keeps them informed throughout the construction process.
Residents are often aware of the stormwater management and water quality improvement functions of BGI; typically, the functions that the assets were primarily designed to deliver (Kabisch et al. 2016).
Awareness of other, less visible, cobenefits, such as carbon sequestration (as part of climate change adaptation objectives), air quality improvement, and noise reduction, are less widely understood, suggesting that the multifunctionality of BGI is not routinely acknowledged. This may be due to challenges in identifying and quantifying the range of BGI benefits that extend beyond water and flood risk management, which requires an interdisciplinary team to fully address. This was explored in the final CWfA research theme (Theme 5) that built on the disciplinary knowledge gained by research Themes 1-4 to develop an innovative approach to illustrate the spatial distribution of key BGI benefits. economic valuation and monetization of benefits, the CWfA research team developed a Geographic Information System (GIS) approach to illustrate and evaluate the spatial distribution of several BGI benefits. The GIS approach was developed to aid City of Portland stakeholders demonstrate multifunctionality in BGI performance, and identify the dominant, relevant benefits for a given location (Hoang et al. 2018). It complements the monetization of BGI benefits provided by other benefit calculators by illustrating the relative uplift in the case study location after an intervention, showing how the benefits may have an impact at the site of intervention and much more widely into adjacent neighborhoods. This ultimately facilitates the identification of beneficiaries of BGI schemes, and in particular, which benefits they may primarily benefit from. Full details of the method are provided in Hoang et al. (2018) and in the brief summary that follows. The spatial distribution and intensity of six biophysical benefits (habitat connectivity, recreational accessibility, traffic movement, noise propagation, carbon sequestration, and NO 2 trapping) generated by the East Lents Floodplain Restoration Project (the "after" condition) were evaluated under flood and nonflood condition states, and compared with the benefits that would have accrued before completion of the floodplain restoration scheme (the "before" condition). The benefits were then normalized using a piecewise linear transformation to a common, dimensionless, scale (0 representing no benefit and 10 representing the maximum benefit) which allowed benefits to be compared across categories. The total, cumulative benefit in each grid cell, for the before and after conditions, is calculated by summing the benefit intensity in each grid cell for each benefit. This highlights where multiple benefits may accrue and whether the BGI intervention leads to specific benefit hotspots, where several of the benefits show a high intensity. By comparing the before and after condition, a value is derived that reflects whether the overall accrual of benefits at the site, following BGI implementation, has improved or worsened. This comparative approach between the before and after condition also allows disbenefits to be identified, for instance, a temporary reduction in carbon sequestration and NO 2 trapping will occur when the restored floodplain is inundated with water. This shows a temporary tradeoff between benefits; when the floodplain is fulfilling its primary function of storing floodwater other environmental processes become less effective and result in a net disbenefit (Figure 9). Interdisciplinary knowledge of the site and the functions that the restored floodplain provides is essential in interpreting the results of the GIS evaluation; creation of flood storage was the primary aim of the scheme and is not a disbenefit provided that the flood water remains within the designated, flood-suitable, area. In addition to mitigating flood risk, the benefits of the East Lents restored floodplain extend beyond the project area by improving recreational accessibility and habitat connectivity, and reducing traffic across the site (due to the removal of several streets as part of the restoration) when the system is in a nonflood state (Figure 9). Carbon sequestration and noise reduction were temporarily reduced due to removal of some vegetation during construction, but should recover and then increase as the new vegetation matures.
This conceptual GIS approach to multiple benefit evaluation, and the corresponding creation of a spatial benefit intensity profile, also demonstrates the aggregated impact of BGI projects on a range of stakeholder groups and beneficiaries. The results indicate that, while most of the beneficiaries are local, the scope of beneficiaries is increased through wider (though modest) contributions to improved air quality, recreational access, and habitat connectivity at the city scale. Nonetheless, there are limitation to this approach and uncertainties associated with the scale of analysis.
Limitations. Data resolution and data availability are limiting factors in the GIS multiple benefits evaluation method; the resolution of the overall benefit intensity is restricted by the finest resolution that all benefit layers could be aggregated without further processing. The benefit determination also only represents a snapshot in time and longer-term benefit accrual during flood and nonflood conditions are not accounted for. The six benefit categories that are included in the GIS assessment are also only illustrative of the potential benefits, and disbenefits, of this project; in reality, the East Lents Floodplain Restoration Project will generate a range of additional impacts, some negative. While not concentrating specifically on East Lents, Hagerman (2007) found that the production of new greenspaces in Portland has previously forced low income housing and service agencies to fight their own displacement. BGI and Green Stormwater Infrastructure are often associated with gentrification; poor and vulnerable communities will not receive the intended benefits of BGI if they are moved out of areas due to greenspace development (Anguelovski et al. 2019). Investigating issues of gentrification and environmental justice both in our case study areas, and more widely across Portland, were beyond the scope of this study which was designed to address discrete elements of the complex issues surrounding stormwater management. Nonetheless, these issues, and the equitable delivery of BGI, should be key considerations for planners, designers, and decision makers if Portland is to become a sustainable Blue-Green City.

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
Geographical Scope of CWfA Research Project. Our evaluation of the multiple benefits of BGI is also limited in geographical scope. The geographically targeted approach employed by the CWfA research project means that the exact benefits associated with the East Lents Floodplain Restoration Project (e.g., related to flow and suspended sediment dynamics [Theme 2], or river habitats [Theme 3]) are specific to this site. However, the type of benefits will likely be delivered by similar projects, such as the Schweitzer restoration project that facilitates the deposition and storage of sediment and sediment-bound pollutants (Janes et al. 2016), or the future West Lents Floodplain Restoration Project that is expected to reduce local flood risk, restore fish and wildlife habitat, and improve water quality in Johnson Creek (BES 2019d, accessed December 2019. Similarly, the community perceptions of bioswales identified under CWfA research are specific to the neighborhoods surveyed. Nonetheless, common issues that we have identified are expected to affect residents' appreciation and acceptance of Portland's bioswales more widely, including environmental attitudes, awareness and understanding of functionality, and plant choice and maintenance.

INTERDISCIPLINARY RESEARCH TO OVERCOME BARRIERS TO BGI
Interdisciplinary research encourages the bridging of dominant paradigms from individual disciplines to generate new understanding. Thoms and Parsons (2002) use the example of eco-geomorphology as an interdisciplinary focus to develop new understandings of river systems. Responses to the question "what characteristics are desirable for urban rivers?" from hydraulic, geomorphological, and ecological perspectives will differ; to support river flows during a 1:100 year event; maintain the structure and function of natural features in the river channel, and; maintain individuals, populations, and ecosystem processes, respectively (Thoms and Parsons 2002). In order to protect and conserve the river function while mitigating against future flooding all of these answers need to be incorporated into strategies to improve urban rivers. Parallels can be drawn when asked to define characteristics that are desirable of (multifunctional) BGI systems. The key advantages of Blue-Green over traditional "gray" infrastructure FIGURE 9. Cumulative benefits distribution across the East Lents floodplain restoration site, calculated according to the methods detailed by Hoang et al. (2018). The dashed outline shows the extent of the floodplain restoration. The highest benefits (red) are where flood mitigation, landscape connectivity, and amenity accessibility have been improved by removal of streets and houses within the reconnected floodplain. Apparent flood disbenefits (blue) occur within the restored floodplain due to higher water levels during floods (as intended), and include reduced carbon sequestration and NO 2 trapping due to temporary loss of some trees during construction. Insert: the location of the East Lents floodplain restoration site (black rectangle) in relation to the Johnson Creek Watershed (green) and City of Portland boundary (yellow). Image adapted from BGC (2016). approaches relate to the creation of the environmental, ecological, social and economic benefits, in addition to meeting flood risk and stormwater management objectives (Fenner 2017).

Multiple Benefit Evaluation that Unifies the CWfA Research Themes
Effective integration of BGI into the existing urban fabric requires detailed understanding of the hydrological, ecological, and social benefits, and tradeoffs, plus knowledge of where they interact with other infrastructure systems (Hoang et al. 2018). Demonstration that the multiple benefits of Portland's floodplain reconnection and BGI projects extend far beyond the hydrosphere is a focal point for the CWfA interdisciplinary research project, and unifies the five research themes.
As an example, the East Lents Floodplain Restoration Project generates flood risk reduction benefits (flood peak attenuation, increased water storage and infiltration, reduced runoff velocities, sediment trapping) and the river restoration and setback outfalls in Johnson Creek have been shown to improve sediment quality and habitats, and reduce pollutant concentrations (Themes 2 and 3). Awareness of the flood risk management and water quality improvement benefits of street bioswales was common among surveyed residents (Theme 4) yet knowledge of the wider, less visible benefits was limited. Similarly, the lack of recognition and valuation of the multiple benefits of BGI in policy and practice was highlighted as a sociopolitical RDU by City of Portland stakeholders that inhibits progress with citywide BGI implementation (Theme 1).
The contrasting nature of the biophysical and sociopolitical uncertainties that limit BGI implementation highlights the importance of interdisciplinary teams. These teams are needed to design BGI that is adaptable and resilient to future (uncertain) changes in climate and land-use, while ensuring uninterrupted service delivery. They are also essential for developing and raising awareness of BGI, often by coproducing new knowledge with beneficiary communities and policy makers. Disciplines must not be the focus of research programs, and instead, research inputs should be distributed and blended to ensure that the specific research questions can be addressed by a group with the necessary disciplinary-based experiences, skills and competencies. Collectively, the interdisciplinary CWfA research project created an evidence base for a range of environmental and social benefits of BGI as a means of helping City of Portland stakeholders overcome the barriers to BGI.
One of the challenges of interdisciplinary approaches is ensuring that projects are truly interdisciplinary, rather than allowing fragmentation into multidisciplinary research undertaken from a number of perspectives, with individuals working on their own subtopics within a common framework, and outputs only being synthesized toward the end of the project (Massey et al. 2006). Some degree of fragmentation was unavoidable during the multifaceted CWfA research project, which involved field-based research performed using different techniques and conducted simultaneously at multiple locations. Nevertheless, we found that our commitment to mutually agreed, interdisciplinary goals, adoption of a common terminology, and establishments of a project intranet to facilitate data and information sharing, together with regular meetings, enabled us to keep our research coordinated and integrated.

Changing the BGI Paradigm Toward Greater Multifunctionality
In order to realize the multiple benefits of BGI, the "urban drainage" paradigm must be supplanted by one whereby BGI is seen as the starting point for urban planning, development and regeneration. Opportunistic implementation of BGI should be at the forefront of all new city projects, in addition to the use of BGI to solve specific problems. In practice, interdisciplinary collaboration across departments and organizations on BGI projects intended to meet the respective policy and strategic objectives of those involved has several advantages. Financially, the burden on individual organizations would be reduced if collaborative BGI projects were collectively funded, as illustrated in Seattle and partnerships between Seattle Public Utilities and King County Wastewater Treatment Division (City of Seattle 2015). Furthermore, increased value-for-money and improved costbenefit analyses of BGI projects are assured once the multiple benefits are included in calculations (Jones et al. 2017).
Moving the focus of BGI projects away from solely "stormwater management" will increase the likelihood that organizations not typically involved in BGI projects will get involved, offering specialist knowledge and resources that may ultimately increase the realization of multiple benefits derived from a wider array of ecosystem services provided by urban streams and urban stream corridors (Yeakley et al. 2016). These types of interdisciplinary projects are more easily adopted by policy makers (Donaldson et al. 2010) and deliver benefits to a wider range of beneficiaries. Such collaborative schemes would help address wider issues of interagency fragmentation JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION and ineffective communication that are common within and between institutions across the globe (O'Donnell et al. 2017;Thorne et al. 2018). Regular meetings are a prerequisite to limit potential challenges due to gaps in understanding at the interface between disciplines, and facilitate the potential evolution of subdisciplines that may develop their own framing of a common problem (Qin et al. 1997).
Leadership in interdisciplinary projects relies on individuals who are able to span disciplinary boundaries, build trust and relationships, and successfully facilitate cognitive (integrating the different epistemics of the different team members) and structural (need for timely knowledge and information exchange) tasks (Margerum and Robinson 2015). The CWfA research project was led by Theme 1 researchers who have an excellent understanding of the physical and social science components of stormwater management, and expertise in both quantitative and qualitative data analyses. They were able to identify where cross-theme research was needed, encourage communication and data sharing, and keep the team focused on the interdisciplinary goals. Theme 1 researchers were also responsible for identifying the key contributions from each research theme that help address the overarching interdisciplinary aim involving the codevelopment of an evidence base to demonstrate the multiple benefits of BGI. Strategies for resolving procedural tasks, for example, conflicts during any stage of the process, reluctance, inertia, and structural obstacles, are also essential (Gray 2008) and were assigned to Theme 1 researchers. Tools, such as GIS, were also used to facilitate learning and communication activities across boundaries due to their effective visualization capabilities, providing shared visions and understandings for disciplines to converge toward, as illustrated here with the development of the GIS-based approach to evaluating the multiple benefits of BGI.

CONCLUSIONS
Designing and delivering effective flood risk and stormwater management strategies is more than a purely engineering challenge; indeed, addressing the scientific aspects in isolation can rarely achieve the desired societal results owing to the complexity of interactions between natural and societal components of stormwater management (Morss et al. 2005). BGI is recognized as a viable strategy to manage stormwater and flood risk, and its multifunctionality may further enrich society through the provision of multiple cobenefits that extend beyond the hydrosphere, including urban regeneration, climate change adaptation, recreation and public amenity, health and well-being, open space improvements, and enhanced biodiversity. The incorporation of multifunctional BGI into the design of urban landscapes is essential if cities are to achieve higher goals for sustainable development.
Despite an abundance of BGI and international recognition as a leader in green stormwater management, a range of barriers, or "RDUs," continue to cloud decision making and impede wider implementation of BGI in Portland, Oregon. In this paper, we demonstrate how the "CWfA" research collaboration addressed many of the biophysical and sociopolitical RDUs through an interdisciplinary study of the benefits of Portland's BGI, focusing on green street bioswales and the East Lents Floodplain Restoration Project in lower Johnson Creek.
An interdisciplinary approach is advocated to comprehensively evaluate the multiple cobenefits of BGI and facilitate the development of strategies to overcome the barriers to implementation. The approach adopted by the CWfA research project was made possible by strong linkages between research themes, sharing of knowledge and expertise, and leadership by researchers who were adapt at spanning disciplinary boundaries and maintaining focus on the overarching interdisciplinary research goal. The importance of sustained interaction and collaboration to integrate disciplinary expertise in, for example, hydrology, hydraulics, ecology, hydrochemistry, sociology, geography, and economics, is essential for an interdisciplinary evaluation of BGI benefits and disbenefits. While the disciplinary differences in methodologies were embraced in this project, and pivotal to providing evidence of the disparate benefits of multifunctional BGI, cross-disciplinary engagement and data exchanges throughout the research process were of paramount importance to reduce the potential for fragmentation and ensure research remained integrated. The interdisciplinary research process that was adopted is highly transferable to other urban environments facing similar development pressures and increasing demands for natural resources.