Implementing a systemic approach to water management: piloting a novel multi-level collaborative integrated water management framework in east London

To meet challenges of urban growth, competition for resources and environment deterioration there is a need to address issues of fragmentation in governance and planning for water, and therefore a need to implement more integrated water planning and management approaches that support cities' growth and the health of the environment. In the United Kingdom, the 25-Year Plan for the environment called for the environment to be ‘mapped and managed more as a system’ (DEFRA 2018). The National Framework for Water Resources (Environment Agency 2020) reflected this ambition with emphasis on multi-sector planning, and integration of flooding, wastewater and drainage issues to unlock broader benefits. To achieve this ambition, the development and application of new conceptual frameworks for integrated water management will provide an important platform of tools and real-world experience for practitioners and policymakers to share and use in the future.

To understand and manage complex water systems, a range of frameworks have been developed. Some of them focus on modelling, for example, of water resources and allocation (Letcher et al. 2007) or water quality (Rode et al. 2010). Others address the complexity of water management from stakeholder perspectives only (Thoradeniya & Maheshwari 2018), emphasising the need to involve decision-makers in strategic planning processes. However, recent publications emphasise the need to integrate the physical and human aspects of the water system. At the high-level, frameworks such as meta-models of coupled human-water systems (CHWS) can give an overview of the complexity of the water systems (Mijic et al. 2023). When applied to more practical challenges such as regional water planning, frameworks that support integration between qualitative (e.g., system mapping) and quantitative (system performance simulation) evidence provide a basis for a holistic approach to integration (Mijic et al. 2022). However, there is still a need to test and refine systems-level frameworks within real-world applications.

In achieving an integrated water system model, the complexity of linking multiple components is an open scientific question (Voinov & Shugart 2013; Iwanaga et al. 2021). Traditionally, water systems (supply, wastewater, basin) have been modelled separately, using detailed (e.g., Gironás et al. 2010; Jaber & Shukla 2012) or conceptual (e.g., Lindström et al. 2010; Coxon et al. 2019) physically based models. Although giving valuable information for design and scenario analysis, these models cannot assess interdependences between the system components. To address this challenge, the scientific community has developed integrated modelling tools, one such being the Water Systems Integration Modelling (WSIMOD) Framework (Dobson et al. 2023a). The WSIMOD provides an integrated evaluation of a water system by linking components of the urban and rural water cycle in a flexible arrangement, that represents the physical connectivity of a system, its operational management, and its governance. The value of the model has been presented in previous scientific publications, assessing the complexity of urban (Dobson et al. 2021) and basin (Liu et al. 2023a) water systems, the role of Nature-Based Solutions (NBSs) (Liu et al. 2023b) and links between the water system and planning (Puchol-Salort et al. 2022). However, WSIMOD has not yet been used to provide the evidence required for decision-making and effective governance.

Even when frameworks and models that provide analysis and evidence around water system performance are available, stakeholders' collaboration to reach agreement and coordinated plans remains a challenge. This is emphasised by the concepts of participatory and polycentric (Shunglu et al. 2022) governance approaches. Both frameworks, while emphasising the benefits for future water planning by accounting for the multiple perspectives of the management problem (Manny 2023), still face the issues of unintended consequences occurring when the scope of the analysis is too narrow compared to the scale of physical and management interactions. These challenges have been documented in the overview of so-called socio-hydrological phenomena (Di Baldassarre et al. 2019). Theoretical studies have shown that to anticipate and prevent an occurrence of a phenomenon in a coupled human-water system, there is a need for collaborative and coordinated decision-making (Mijic et al. 2023), but examples of successful real-world examples are still missing.

This paper develops a novel framework for regional water planning that brings together participatory engagement, integrated water systems modelling and collaborative decision-making.

While we describe primarily the technical aspects of the Sub-Regional Integrated Water Management Strategy (SIWMS) for East London, we consider institutional challenges as a sub-theme that is assumed to be central to a strategy's success and its impact. We hypothesise that, for successful integrated water management, the process (engagement and collaboration) is equally important as the output (the evidence and plans defined) to achieve the desired long-term outcomes.

The value of this study is the real-world application and demonstration of the novel framework developed and tested for London's first SIWMS pilot study, described in Section 2. The framework development and its novel aspects with respect to process definition, model development and application to SIWMS evidence generation and interpretation are described in Section 3, followed by key results discussed in Section 4. In Section 5, we reflect on the implications of the work and provide recommendations for other researchers and practitioners on the practical delivery of SIWMS using systems thinking.

The novel framework described in this paper is based on the policy need for more effective integrated water planning. In recent years, London's vulnerability to water-related climate impacts has been brought into greater public consciousness by flash flooding events in July 2021 followed by the severe drought of 2022. Both events affected London and the south-east of England bringing major disruption to the city with flooding affecting 24 boroughs and over 1,000 properties. The drought resulted in a temporary use notice restricting water use for domestic consumers between August and November 2022 (London Councils 2022, p.2). In this section, we make the case for the integration of water planning in London as well as outline previous attempts at integrated water management, which create the context and need for the new framework.

The institutional arrangements for water policy, planning and governance in London are highly fragmented, with different actors holding responsibilities for the management of different parts of the water cycle with limited coordination. Compliance of the water industry with environmental and drinking water legislation is regulated by the Environment Agency (EA) and the Drinking Water Inspectorate (DWI), respectively. Water supplies are provided to London by four companies (Thames Water, Affinity Water, SES Water, and Essex and Suffolk Water). Local Planning Authorities have an influential role in the management of the water cycle through the creation of spatial development frameworks; setting of policies for new housing development; and responsibilities for highway drainage and surface water management. Natural England, as a government advisory body for the natural environment, sets the agenda for environmental management and protection and is increasingly influential on key river basin issues, such as nutrients. Beyond this, numerous other actors hold influence on policies and plans, including major industries, landowners and farmers, pressure groups and NGOs.

Silos of institutional arrangements and planning frameworks pose a challenge to planners seeking to deliver outcomes across the water system, at a basin scale. Solutions are promoted and prioritised based on their performance against a narrow subset of water outcomes relevant to a single planning framework (e.g., water resources, surface water flooding, fluvial flooding), rather than performance across all aspects of the water cycle. More integrated planning allows the delivery of wider, cross-cutting benefits and provides an opportunity to manage second-order impacts of interventions better. In many cases, interventions planned on benefits analysed in one planning framework will have second-order impacts affecting other planning frameworks, such as impacts on flooding of a reservoir planned for its water resource benefit. The current structure of planning frameworks means that these second-order impacts are not always fully considered. Where they are identified, misalignment of planning timelines across frameworks poses a challenge to ensuring that second-order impacts are managed in other plans. The case for more integrated water planning is dictated by the opportunity to address these issues and promote interventions that create multiple benefits across the water system, enhancing the overall return on investment of those interventions.

Recognising the importance of good planning for London, The London Plan identifies the need for integrated water management to enable ‘Good Growth’ defined as ‘… growth that is socially and economically inclusive and environmentally sustainable’ (GLA 2021, para. 0.0.18). Specifically, the London Plan policy on water infrastructure (SI 5) specifies the requirement for an Integrated Water Management Strategy (IWMS) to be delivered in growth areas where there are flood risk constraints or insufficient water infrastructure capacity (GLA 2021, p. 356).

These previous IWMSs had success in providing integration by creating a common evidence base and plan for joined-up action across planning frameworks, underpinned by an understanding of geospatially specific contexts. This was coupled with an awareness of key water issues and consensus around the level of ambition needed to address them among stakeholders, developed over the course of the IWMS planning activities. These elements provide a tested basis for the development of integrated water planning activities going forwards. However, the limited and local geographical scope of these studies ignored the dependencies existing in the water system and identified the need to develop a more system-based understanding of challenges and opportunities in the river basin. Doing so would allow for better integration of action across multiple stakeholders to address specific strategic challenges observed in the river basin, such as climate change or major urban development. Furthermore, the experience in the implementation of these local IWMSs has identified the potential for further focus on the development of collective capacity and capability across stakeholders to support ongoing action and implementation of water management measures. This would help to build a robust platform for long-term action beyond the initial development of the IWMS.

In 2022, the GLA decided to pilot a sub-regional approach to integrated water management to respond to the need for a more basin focussed and systemic approach to water management. The pilot built on the successes of previous IWMSs, aligning evidence and ambition for action across stakeholders, while scaling up to a sub-regional area to enable a better understanding of river basin complexity. The pilot objectives are summarised as:

• Provide a basin-level understanding of the water cycle and infrastructure systems, including any strategic challenges and interdependencies.

• Provide a clear evidence base for action and consensus across stakeholders.

• Develop agreement across stakeholders on the level of action needed to enable housing growth sustainably.

• Develop consensus across stakeholders of measures needed and commitments required for implementation.

• Ensure the adaptive capacity of the strategy in the future and identify governance and data-sharing requirements.

• Develop capacity for implementation and delivery of the strategy.

In determining the study area, several factors were taken into consideration. The two determining criteria for the study area selection are listed in the following.

• Match between hydrological basin and administrative boundaries – to provide outputs and prioritised actions across the whole basin and to provide a direct link between strategy recommendations and stakeholders.

• The limited scale of the area – to increase collaboration and co-learning, to provide detailed recommendations for implementation and to allow for more in-depth analysis and modelling of the basin.

The sub-regional IWMS pilot was delivered as a joint project between Mott MacDonald (MM) consultancy and the Imperial College London (ICL) and was governed by a collaborative steering group comprised of organisations with key water management responsibilities in the East London sub-region. This included London Borough (LB) of Enfield, LB Waltham Forest, LB Haringey, LB Hackney, LB Tower Hamlets, LB Newham, City of London, Natural England, Environment Agency, Thames Water, and Greater London Authority.

In this section, we provide an overview of the conceptual model underpinning the programme of work for the SIWMS (the ‘Multi-Level Collaborative Integrated Water Management Framework), a description of WSIMOD (which was used for the main analytical element of the SIWMS), followed by a summary of activities delivered over the work programme. This will set the scene for a discussion of selected results and finally the benefits of the framework in delivering an improvement in integrated water planning.

The SIWMS was dictated by the three overarching stages of the Multi-Level Collaborative Integrated Water Management Framework, laid out as an inverted triangle to indicate the convergence in analysis from broad perspective thinking in the early stages to narrow, targeted analysis subsequently. This framework is the result of the learning gathered through the experience of MM and ICL applying WSIMOD in other river basin contexts such as phase 1 of the OxCam Arc Integrated Water Management Framework for the Environment Agency (Mijic et al. 2022; Mott MacDonald 2022). The three stages are described in detail in section 2.2 and summarised here. The framework begins with a Conceptual Analysis phase in which the insights of river basin stakeholders are gathered alongside collection of data and inputs from existing plans. The goal of the conceptual analysis is to arrive at a coherent set of scenarios, modelling objectives, and performance metrics that can be targeted by the second phase of the framework, which is Integrated Modelling. The integrated modelling stage numerically implements the setting provided by the conceptual stage. The implementation begins by establishing a baseline by which the identified scenarios and proposed options can be measured. The outputs of this stage are performance metrics (formulated as WFD indicators) and associated simulation time series. The metrics are used to provide evidence of the expected and unanticipated behaviours associated with the scenarios and options, while the simulations provide an in-depth cause-and-effect understanding of the integrated mechanisms at play. Ultimately, this evidence and understanding is packaged for decision-makers in the final, Integrated Planning stage. Option portfolio selection and identifying a path to implementation are carried out in collaboration between the system actors and steering group.

Overall, the framework enables technical system analysis to be situated alongside engagement and participation with key system actors to capture the plural perspectives in the river basin. It represents a practical attempt to integrate the physical and human aspects of the water system by providing a logical structure around which to organise analytical and engagement activities. With the inclusion of integrated system modelling by WSIMOD, this also addresses the need for a system-based understanding of the river basin while building consensus, capability and capacity for action collectively across stakeholders.

The WSIMOD Framework is a comprehensive modelling framework designed to capture interactions between different parts of the water system and therefore addresses the complexities of integrated water management. It provides a set of self-contained objects, including nodes, arcs, water stores, and model orchestration, making it suitable for a wide range of water systems. A key feature of WSIMOD is its acknowledgment that water systems are not universally textbook in nature. Recognising the diversity of non-textbook water systems, WSIMOD adopts a customisable modelling approach. It is not intended as a one-size-fits-all solution, but rather as a flexible tool that can be tailored to specific needs and characteristics of different water systems. WSIMOD is implemented in Python 3, a widely adopted language in the environmental modelling community. This choice facilitates quick setup and easy customisation, allowing users to adapt the framework to their specific requirements. A notable aspect of WSIMOD is its implementation of model orchestration. Users can customise the high-level control of interactions within a time-step, providing a level of adaptability that sets WSIMOD apart. The framework allows for a wide variety of representations of water systems, showcasing its versatility and applicability to different scenarios.

Physical representations of the different components in the water cycle are typically implemented as model nodes and arcs. Nodes are representations of the various components of the water system, including urban and rural land, water infrastructure systems (urban and rural) and environmental systems (such as rivers). The components are then integrated through a concept of arcs, which can be used for basic transmitting of flow and pollution through the system and for more complex management decisions such as water abstraction and discharge rules. The unique feature of the model is its ability to simulate both flow and water quality, enabling a comprehensive assessment of integrated water systems performance. The simulation of pollution concentration in rivers enables an explicit alignment with the WFD chemical indicator metrics (in particular, the phosphate and ammonia classifications which are common chemicals of concern in British rivers), which in turn improves the relevance of simulation results for decision-makers.

System mapping was undertaken to identify system interdependencies at a conceptual level. The mapping was informed by engagement with system operators and the SIWMS steering group. Discussions with operators were particularly insightful in understanding the operational dynamics of the system, such as the approaches to managing water quality challenges in the River Lea and how water resources and supplies are managed in conjunction with the rest of the London water supply system.

The system mapping identified the following metrics: properties at risk from flooding; flood placemaking; river flow data (Q5, Q95, QMED, environmental flow); river WFD status (chemical); river WFD status (ecological); supply/demand balance; river morphology; invasive non-native species, carbon sequestration; embodied carbon, operational carbon, biodiversity; soil health; mental health; physical health; urban heat; air quality; social connectivity and networks. As not all metrics could be assessed quantitatively, a limited number were taken forward for modelling prioritising river flow data (Q95, Q5) and river WFD status (ammonia, nitrate and phosphate).

Data were collated from a wide range of sources to inform the SIWMS, including unstructured data from steering group organisations and data from open sources. Water plans and planning frameworks were the main sources of information in enabling the identification of options in task 5 ‘water management option identification’. The data included in plans were limited and therefore, for the purpose of modelling, additional assumptions were made related to their representation within WSIMOD.

Open-source data were used, where possible, to inform baseline assumptions for the Integrated Modelling phase of work (task 7–9), while the model data are available in Appendix I of MM for GLA 2023, the original data sources are described in Liu et al. 2022 and Dobson et al. 2023b. Further information related to the water resource system connectivity, operational controls, monitored reservoir levels and raw water quality sampling was provided by Thames Water, along with more detailed outputs from their Drainage and Wastewater Management Plan (DWMP) to inform asset location and flood risk information.

Population (including current and future projections), land use and agricultural data were used to parameterise nodes within the WSIMOD framework. The model was then validated against open-source observed river flow (time series) and river quality data (spot sampling) over a 20-year continuous time series simulation with daily rainfall totals and daily model outputs.

A series of discrete future scenarios were developed to represent different plausible futures in the sub-region and to be modelled using WSIMOD. The scenarios were built up from a series of common variables representing different exogenous factors which may change to a greater or lesser degree in the future. These variables included: growth and urbanisation forecasts, environmental adaptation including water quality improvements delivered via the Water Industry National Environment Programme (WINEP), including treatment works upgrades and abstraction regime changes; Schedule 3 implementation of the Flood and Water Management Act (FWMA) 2010 regarding the implementation of SuDS for new developments, and level of demand for water.

We used the FUTURES process to collate the different plausible combinations of factors into five likely future development scenarios: ‘Country Life’, ‘Unrealised Urbanisation’, ‘City Living’, ‘Prosperous Growth’ and ‘Environmental Priority’ (Mott MacDonald and UWE Bristol). These future development scenarios were discussed and validated with the project steering group.

Climate change was included as a factor, using Representative Concentration Pathway (RCP) projections with RCP2.6 as a lower-bound scenario, and RCP8.5 as an upper-bound scenario. In line with the water regulator's (OFWAT's) guidelines, we selected the 50th percentile from the UKCP18 probabilistic forecasts. Furthermore, an additional scenario where we assumed wet summers with higher intensity convective storms and drier winters, referred to as ‘unseasonable’ was sampled from the 25th (winter) and 75th (summer) percentiles of RCP2.6. However, as the impact of climate change is largely independent of economic and legislative impacts, we modelled three climate change scenarios in conjunction with the five development scenarios.

Following data collection from existing water plans and planning frameworks (task 3) a series of water management options were compiled to inform option modelling (task 9) based on existing plans relating to water in the study area. This included: Thames Water's Drainage and Wastewater Management Plan (DWMP) and Water Resource Management Plan (WRMP); the Environment Agency's Flood Risk Management Plan (FRMP) for the Lea Valley and the River Basin Management Plan (RBMP); Lead local flood authorities' Local Flood Risk Management Strategies (LFRMS) and subsequent Surface Water Management Plans (SWMPs); and the London Plan and Local Plans for each planning authority.

The options from across the planning frameworks were collated into a long list, categorised and combined where they overlapped to enable ease of modelling. A qualitative screening exercise was undertaken to capture the likely benefits of different option types. This analysis identified several intervention types that could be grouped together for simplicity. Ultimately, to inform the integrated modelling phase (task 7–9), all options were categorised as one of the following:

• Options for modelling to assess their performance against the metrics identified (Table 1).

• Options not suitable for modelling. These were often more related to enabling behaviour or institutional change in the system and so could not be assessed for performance against metrics reliably. These were instead considered during the integrated planning phase of work (tasks 10–11).

Over the course of two steering group meetings, a collective ambition for the pilot was created. The first meeting established the context of the project, and the second discussed and captured desired outcomes collectively. This helped establish a connection between the pilot and the context of individual steering group organisations. As a result of the diversity of roles of the steering group members, a broad range of themes emerged, including the need for alignment, collaboration, clarity, and a focus on implementation. This ambition was used to inform the integrated planning phase of work qualitatively and ensure the final recommendations responded directly to stakeholder ambitions.

Using the core components of WSIMOD and the scope definition set out in the conceptual analysis phase, a model representing the upper and lower Lea basins was established. Nodes are used to represent the key population, land use, treatment, runoff and pollutant loading and supply/demand elements and arcs to represent the transfer of water between these elements, such as river or sewerage flow paths.

To relate the modelling framework to the river basin, the model outputs were measured at the downstream of the Water Framework Directive (WFD) surface bodies. Population and agricultural land use data were used as part of the initial model conceptualisation, along with runoff parameters and demand figures. Key assumptions were then adjusted to better match the sample and flow data.

Nash–Sutcliffe and percentage-bias values were used to evaluate the overall model performance and model confidence for certain indicators: measured flow over a 20-year time series; and intermittent spot samples with ammonia, phosphate and nitrate measures were used for water quality validation. Key assumptions were then adjusted to better match the sample and flow data, such as runoff coefficients and pollutant loadings.

The model represented low-flow conditions well, and there were some limitations related to the influence of external water resource transfers during high-flow conditions which could not be validated at this stage. These were found to be a result of the complex water supply operating regime in the London region. As this was not a focus of the study, the model was considered suitable to demonstrate the influence of scenarios and options on the key components of water quality, water stress and flooding.

Using WSIMOD, five development scenarios and three climate change scenarios identified in task 4 were modelled. To incorporate scenarios into the model, we considered a 2050 time horizon over which various river basin changes would take place. The scenario modelling measured potential changes in the river basin associated with these scenarios across the following indicators: phosphate, nitrate, ammonia, fluvial flood risk, drought resilience of flows, and water supply stress.

However, available information for scenario modelling did not always align with the 2050 time horizon or was subject to uncertainty. For example, population growth projections extend to 2041; and there is uncertainty relating to future potential policy or legislation relating to storm overflow reductions and biodiversity net gain requirements. The objective of the scenario modelling was to indicate a range of future challenges or risks for the basin and water system's operating environment. We used assumed figures where information did not exist or was uncertain, recognising that in future rounds of planning or modelling this may be updated.

On undertaking this modelling, we found that two pairs of the development scenarios were not distinctively different in the changes observed in the indicators after modelling, providing unnecessary complexity to the analysis without additional insight. We streamlined the scenario modelling to focus on the two development scenarios with the greatest differences between them in terms of modelled changes to indicators: City Living and Country Life. This made the analytical work more efficient and provided a simpler set of outputs for stakeholders to understand and engage with. In addition, the climate scenarios were rationalised such that the RCP2.6 pathway was used in option modelling for analysis, as there was limited difference in overall results across the three climate scenarios.

The five options listed in Table 1 were modelled in WSIMOD to measure their impact on the six key basin indicators (phosphate, nitrate, ammonia, fluvial flood risk, drought resilience of flows, and water supply stress) under the City Living and Country Life development scenarios. Our options were shortlisted based on whether they could be represented within WSIMOD, and options such as river restoration were excluded as the river channel itself is not explicitly modelled.

The function and performance of different option types reflect that they are unlikely to operate within the same space: water resource options are imported from external regions and benefits supply, and water reuse from Deephams STW is taken independently from other external water resources. Options were modelled independently; however, in some cases, such as SuDS and natural capital options, it is likely that benefits are similar depending on their typology and geography. As a result, we limited natural capital options to the upper Lea basin only; and the SuDS options to the lower Lea basin only, reflecting the urban/rural differentiation in land use between the two operational areas. An additional benefit was to demonstrate the influence that increased SuDS programmes might have across various boroughs that take a lead risk management role in surface water flooding.

For ease of comparison of results, the outputs were also presented as postage stamps in a matrix structure, as described in Figure 5, section 3.1

Based on the option modelling of task 9, options were categorised based on whether they provided an improvement or detriment to one or more of the metrics.

A priority category of options (‘Least Regrets’) was established for options which provided system benefits across one or more modelled metrics and no deterioration or reduction to any other metrics. These represent the water management options with no trade-offs and therefore can be delivered as a priority.

A secondary category of options (‘Principal Options’) was established for options which provided significant benefits to one or more modelled metrics but showed deterioration or reduction to other metrics. These represent options which should be delivered for their benefits but which will need to have any negative second-order impacts mitigated.

On analysis of modelling results for the ‘natural capital’ option, it was identified as needing further work to represent the potential benefits of natural capital enhancements in the upper basin.

The prioritised options from task 11 were analysed in combination with the options which were categorised as not suitable for modelling in task 5 to form the overall recommended options for delivery. Based on these recommended options the steering group was consulted and asked to propose short-term actions to enable delivery of these options in the 90-day period following the completion of the SIWMS. These were organised by theme: flooding, water quality, water stress, and planning and governance. The 90-day action plan ensured that the momentum for collaborative action across the steering group was maintained in the creation of real change in practice among project partners.

A core benefit of using WSIMOD in this instance is to demonstrate the interdependencies at play between the various systems and allow for more holistic water management options to be promoted. The model may be used to identify in which basins (defined at WFD waterbody scale) interventions may have the most benefit but does not include assessment of key mechanisms, such as local hydraulic behaviour, that are required to satisfy detailed planning requirements for interventions. The scale of the model assists in speeding up simulations so that many more options and scenario combinations can be simulated, but also means that further work is required to develop water management options to the point where they can be delivered.

In this section, selected outputs (from the tasks defined in section 2) are presented to provide a view of the type of information produced at different stages of the application of the Multi-Level Collaborative Integrated Water Management Framework.

The findings of the SIWMS are discussed thematically in the following section, section 5, to give an assessment of the effectiveness of the Multi-Level Integrated Water Management Framework.

An example of a system map created during task 2 of the Conceptual Analysis phase of work for the water resource system is shown in Figure 4.

This map includes the key relationships in the basin in water quality and the operation of Coppermills WTW. The negative second-order impacts of PCC reductions explored in the modelling are shown: more ‘Adoption of demand-side measures’ > lower ‘Water usage (PCC)’ > lower ‘Treated WW discharge’ > more ‘Low flow (frequency/duration)’ > lower ‘Water quality’ > potentially increased ‘interrupted abstractions’. WINEP interventions and natural capital interventions can be seen as measures that mitigate the negative impact of PCC reductions on water quality.

The map indicates key ways in which climate change will affect the system and it shows the metrics used to model system interventions.¹ The map shows how the major category of water resource zone options (e.g., major infrastructure schemes) address water resource shortfalls, allowing a balance in supply and demand to be maintained alongside the introduction of upstream abstraction reductions. The water resource zone interventions and the ‘dry year supply–demand balance’ metric are shown as having complex links because they operate over a wider geographical scope than the SIWMS boundaries and the modelling done.

Figure 5 is an example of ‘postage stamp’ diagrams created to visualise scenario modelling results of task 8 in the Integrated Modelling phase of work.

A key characteristic of WSIMOD compared to system-focused models is the capability to evaluate the impact of the different scenarios in changes to the baseline classifications across the integrated system. For example, results in Figure 5 indicate that the most significant observed changes are the impacts of abstraction licence reductions on flood risk, characterised by a failing flood classification in the main River Lea (Figure 5 in row 4). At the same time, the increased flows that result from the licence changes are sufficient to improve the water quality classifications in the Lea (e.g., row 1). The model enables comparison between different scenarios, showing, for example, negligible changes compared to the baseline evaluation for all selected indicators for the Unrealised Urbanisation scenario and more significant changes in case the Country Life future gets realised.

Analysis of the WSIMOD results for the basin identified a series of findings which have relevance to the discussions of this paper. Analysis of all of the WSIMOD results is contained in the SIWMS final report (MM for GLA 2023), the section below discusses only the findings with relevance to the discussion in section 5 of this report relating to integration and implementation of systems-based approaches. The findings discussed include Abstraction reductions, sewage treatment works upgrades, sustainable drainage systems, and reducing per capita consumption.

The outputs from the modelling demonstrated that planned reductions in groundwater abstraction in the upper basin (modelled as a development scenario variable) have significant and widespread impacts on the basin.

Planned sewage treatment work upgrades in the upper basin to reduce discharges containing phosphate as part of WINEP were modelled as variables in the development scenarios as they are exogenous to the SIWMS study area but may have an impact on it. The analysis found that the upgrades do not have a widespread or transformative impact on nutrients in the basin as a whole.

Overall, the planned treatment works upgrades have a modest positive impact across nutrient indicators in the basin. For example, in absolute terms across the basin, there is a 28% decrease in phosphorous concentrations in Country Life. However, this falls to a 6% decrease in phosphate in City Living (MM for GLA 2023, p. 31). Moreover, examination of spatial impacts on WFD classifications in sub-basins for nutrients reinforces the modesty of impact. For example, in Country life despite a 28% reduction in phosphate, only one WFD sub-basin incurred an improvement in WFD classification (MM for GLA 2023, Figure 3.3, P30).

SuDSs were modelled as a water management option in WSIMOD and were found to have basin-wide benefits across a number of modelled indicators including flood risk, water quality and drought resilience.

Reducing per capita consumption was modelled as a water management option applying a general reduction to 105 l/h/d from a baseline of 150 l/h/d. PCC reductions have expected positive impacts in reducing water stress but also have negative second-order impacts on water quality and drought risk due to reduced effluent at Deephams STW.

As an example, the modelling results show that PCC reductions reduce the number of days of water stress by 39% in the baseline and 36% in the City Living scenario, with no change in days of water stress in the Country Life scenario (MM for GLA 2023, Tables 4.4–4.6, pp. 47–48).

As mentioned, PCC also has trade-offs with water quality metrics across all scenarios. In Country Life, this trade-off is significant enough to cause a negative threshold change in nitrate in one sub-basin (MM for GLA 2023, p. 46). This option reduces effluent entering the river from Deephams STW, which results in a drop in low flows under all scenarios. In both City Living and Country Life, this is significant enough to cause a negative threshold change for drought risk in one sub-basin (MM for GLA 2023, pp. 45–46). These complex trade-offs can be exemplified further by looking at a specific sub-basin, Pymmes Brook and Salmon Brooks – Deephams STW to Tottenham Locks. Here PCC reduction provides a 10% reduction in high flows, reducing flood risk, but it also reduces low flows by almost 40%, increasing drought risk. Moreover, it increases the level of ammonia by around 30%, and nitrate and phosphate levels by almost 50% (MM for GLA 2023, p. 55).

We have found that the multi-level integrated water management framework provides an effective and practical framework for applying systems thinking to a river basin. An immediate benefit of this framework is that the logical progression from broad to targeted perspective enables a high degree of tailoring of analysis to the context of the basin. In addition, this logical progression also supports effective steering group engagement allowing for key systems thinking concepts to be sensitised with steering group members in the practical context of the basin as they arise.

Undertaking conceptual system mapping, developing future scenarios, and engaging with the project steering group created a shared understanding of the basin's history and context, including key interdependencies. These activities also included tailoring the modelling and other technical analysis, to the context of the basin. For example, this included key decisions on scope such as deciding to model the upper basin (outside the administrative boundaries of this work) as well as the lower basin. Conversely, the group limited the scope of drainage system modelling in the lower basin to reduce complexity and avoid duplication of efforts with pre-existing work.

Integrated modelling added numerical analysis, validating and refining the conceptual understanding of the basin. Importantly, this included the identification of the scale of challenges and interventions in the basin as well as any second-order impacts. This numerical understanding was crucial in enabling the development of a systemically efficient and effective strategy, based on the specific context of the basin. This is discussed in more detail in Section 5.2. A secondary effect of the numerical evidence was that it also enabled the identification of follow-up studies to address gaps in understanding. For example, the observed flood risk increases driven by abstraction reductions have identified the need for a detailed study to understand groundwater and surface water interactions in the upper basin. This will support a more informed understanding of the flood risk impacts of abstraction reductions going forward. Additionally, the identification of the locally specific relationship between PCC reductions and water quality in the Lea (via Deephams STW) prompted the need for a follow-up PCC study to better understand trade-offs associated with different levels of PCC reduction across different drainage basins (MM for GLA 2023, pp. 76–77).

Finally, the broad to narrow perspective progression of the multi-level integrated water management framework provides an effective framework to situate participatory and engagement activities alongside technical analysis and enable further tailoring of analysis. The benefits of this to policy, planning and decision-making more widely are discussed in 5.2. Establishing a broad understanding of the wider river basin context through early engagement and system mapping provided context to integrated modelling findings. For example, the system-wide benefits of SuDSs combined with the identification of significant existing capabilities across stakeholders led to the inclusion of an ambitious SuDS delivery recommendation. This was repeated through the use of participation. Steering group members were asked to contribute suggestions for actions they could undertake to advance the delivery of strategy recommendations within 90 days of the publication of the strategy.

The engagement of the steering group throughout the project was an essential element of the project and drove a high degree of co-ownership of the work, creating the platform for successful engagement in the development of recommendations. We believe that by nurturing collaborative behaviours throughout the process a culture of collaboration is developed, which will, we trust, pay dividends in the forthcoming implementation of the strategy.

As discussed in this section, the multi-level collaborative integrated water management framework was essential to enable close tailoring of analysis and the strategy to the basin context. The system mapping, scenario development and steering group engagement informed the refinement and tailoring of the modelling. In turn, the integrated modelling provided crucial numerical evidence which, combined with the wider understanding of the river basin, was used to develop an effective strategy with buy-in for delivery across stakeholders.

We have identified three key aspects of the system evidence produced by WSIMOD which make the case for further use of integrated models to enable better integration of planning and governance for water. The benefits of these characteristics in supporting better policy, planning and decision-making are discussed in more detail in Section 1. In this section, we focus on discussing the data itself.

Firstly, WSIMOD enabled us to assess simultaneously the impacts of scenarios and interventions across a range of flow (high and low) and water quality indicators. This is critical as the intended consequences of interventions could be analysed alongside secondary (or unintended) consequences. The system data produced on the multi-benefits of delivering SuDS is an example of this (0). This provides a critical additional layer of system-wide analysis to complement the more narrowly focused assessments of planning frameworks.

Connected to this, WSIMOD enabled the evaluation of benefits and trade-offs across a range of climate, development, and policy scenarios. This allowed a more nuanced understanding of the benefits of different options based on variability in performance between scenarios. An example is the performance of sewage treatment work upgrades which deliver a 28% reduction in phosphates in the Country Life scenario dropping to a 6% reduction in City Living. This also enabled identification of most influential scenario factors driving challenges in the basin, such as the abstraction reductions discussed in 4.0. In the future, this could be improved even further by restricting scenarios to purely exogenous factors and modelling policy factors as interventions for better comparability and attribution of impacts.

Finally, WSIMOD supported analysis across the upper and lower basins of the river basin. This allowed for analysis of effects manifesting in the lower basin which were driven by different scenario factors and interventions in the upper basin. This was shown to be important in the case of both abstraction reductions (4.0) and sewage treatment works upgrades (4.1) in the upper basin, which led to impacts on flooding, water stress and water quality metrics within the study area (lower basin).

The development of the study has uncovered opportunities to address the fragmentation of water governance and planning using system evidence produced by WSIMOD. In this section, we describe the ways in which the evidence discussed in section 4 can support more systemically effective decision-making based on the experience of the SIWMS. We also reflect on the importance of consultative and participatory activities to ensure the value of systems evidence is exploited.

Production of system evidence enabled the identification of priorities and scale of action required to address challenges in the river basin. Specifically, the integrated approach to modelling allowed for comparison of the scale of different challenges and the sensitivity of the system to different factors. This evidence provided a strategic system context to evaluate existing and planned initiatives. This allowed us to identify effectively where ambition or action was needed in the planning and strategy development stages of work. An example is the evidence of the impacts of planned upstream sustainability reductions on future fluvial flood risk and water supply stress in the basin (section 4.0). Numerical evidence of the scale of the impact drove the prioritisation of key interventions such as PCC reductions, major new regional water resources schemes, and SuDS in the strategy (MM for GLA 2023, pp. 63–68).

System evidence allowed for better identification of solutions to deal with challenges at source as opposed to reactive solutions. As mentioned, to offset the future water supply stress increase identified by WSIMOD, significant investment in new supply sources and PCC reduction would be required. However, the system evidence identifies that planned abstraction reductions for sustainability reasons in the upper basin are a significant driver of this increase in risk. A more efficient strategy is likely to be to work with planners in the upper basin to manage better the second-order impacts on water stress in the lower basin and therefore mitigate the need for new supply schemes or PCC reduction. As there is currently no coordination on this specific issue between the upper and lower basin, the SIWMS recommended coordination with regulators and upstream abstractors to ensure that second-order impacts are fully addressed in future planning (MM for GLA 2023, p. 76–77). This new capability supplements the approach of previous IWMS, which had a local focus on end-of-pipe solutions.

The integrated modelling also provided critical evidence to support the prioritisation of multi-benefit schemes and support the development of a systemically efficient strategy. SuDS is an example whereby the integrated modelling provided numerical evidence of benefits generated across multiple indicators in the river basin beyond flooding (section 4.2). The opportunity to unlock k multi-benefits through SuDS was a driving factor in their prioritisation of the strategy and the development of an ambitious delivery recommendation. Conversely, the integrated modelling found that PCC reductions had negative second-order impacts on drought resilience and water quality, despite reducing water stress. In response, the strategy identified the need for additional investigations to proactively mitigate the negative second-order impacts on the system (MM for GLA 2023, pp. 76–77).

Situating the analytical capability of WSIMOD alongside participatory and consultative activities (through the multi-level integrated water management framework outlined in section 0) built the capability of the steering group to deliver strategy recommendations effectively. This is critical as although WSIMOD provides a useful tool to prioritise action across stakeholders, investment and policy delivery remains within existing statutory planning frameworks. For example, ‘awareness’ sessions with subject matter experts as part of the steering group programme focussed on creating a baseline of awareness and understanding of key concepts related to integrated planning.² Doing this helped to establish the common benefits of integration across steering group members as well as a consistent understanding of the roles of individual organisations. Furthermore, participatory activities, such as the identification of 90-day actions, helped create common ownership of the strategy recommendations and consequently commitment to delivery.

Ultimately, the combination of the analytical capability of WSIMOD with its application using the multi-level integrated water management framework enabled the delivery of a robust strategy with strong buy–in from stakeholders. The application of both dimensions (analytical and participation) and their interaction is essential to derive the most benefit from integration in the context of the current governance and planning system for water. For example, without the early buy-in, support and common understanding of key stakeholders, a sub-regional strategy supported by system-scale numerical evidence is unlikely to be delivered effectively through existing silo planning frameworks. Conversely, a strategy with buy-in from stakeholders but not underpinned by system evidence is unlikely to respond to the specific context and interdependencies of the basin. This point is highlighted by the discovery during the SIWMS of critical basin trade-offs associated with pre-existing planned activities in planning frameworks which had not been identified or quantified.

We have identified a series of recommendations, based on the experience of the Lea pilot, to improve the delivery of future SIWMS using WSIMOD and the Multi-Level Integrated Water Management Framework outlined in this paper.

• (1) Disaggregation of modelled options and information from the sub-regional scale to the scale of the different planning frameworks. As existing statutory planning frameworks are the main mechanism for action, interventions identified in strategies such as the SIWMS need to be delivered through these (WRMP, DWMP, FRMP). In practice, this requires some aggregation or disaggregation of interventions to match the scale of the planning framework responsible for their delivery. To do this effectively, it's important to ensure that interventions eventually promoted through planning frameworks remain aligned with the ambition and context of the specific sub-regional analysis delivered. Closely connected to this is the challenge of spatial evidence and scale. This is especially true for Local Plans where spatially specific evidence forms the basis of policy ambition. The spatial requirements of planning frameworks' evidence bases need to be identified at an early stage to ensure that sub-regional evidence produced can support robust policies and decisions.

• (2) Ongoing oversight and management of planning information and evidence. Ongoing resource is required at the sub-regional scale to enable the successful transfer of evidence and planning information into statutory planning frameworks at the right time. This requires a permanent organisation to coordinate stakeholders in the sub-region and the evidence created by the strategy once it has been completed. In the case of the SIWMS, this role was naturally suited to the GLA, which as the combined authority for London already holds responsibilities and interest in coordinating action across local authorities in London. If applied to other basin contexts, the organisation taking on this coordinating role will need to be defined and agreed. Connected to this, more work is required to understand how automation and digital tools can be used to enable this ongoing coordination function by streamlining information flow and presentation, in particular.

• (3) Leveraging ongoing commitment for action across participating organisations. Steering group participation was a success on the SIWMS, providing a significant opportunity to help address structural challenges in water planning and governance. To improve this even further, future applications of the approach could consider formal commitments for action across all stakeholders based on basin findings and evidence. The Greater Manchester Combined Authority, United Utilities and EA Integrated Water Management Plan have set an important precedent in this regard (GMCA et al. 2023). Among other things, the plan identifies seven workstreams identified for action and collaboration underpinned by a Memorandum of Understanding between the organisations (GMCA et al. 2023). Replicating a similar approach in the future could help to ensure long-term commitment to action and ensure that the benefits of participatory development of the strategy are exploited into the future.

The approach piloted in this project has set a precedent to deliver an integrated and systemic approach to water management. It was grounded in a robust technical analysis of the river basin, supported by a focus on the development of institutional and organisational capability. From this experience, we have identified three areas of reflection with relevance for future attempts at the application of systems-based approaches to river basins.

We have found that the application of the Multi-Level Integrated Water Management Framework provides a useful framework for applying systems thinking to a river basin. The general progression from broad to narrow perspective enables the refinement of modelling which produces evidence that dictates the development of targeted interventions, strategies and further work based on the specific issues identified in the basin. Moreover, the overall framework enables the effective play out of participatory and analytical dimensions of work between project participants, supporting further tailoring of recommendations and benefits in enabling better policy, planning and decision-making.

Integrated modelling, using WSIMOD in this case, provides the basis for improvements in policy, planning and decision-making through the production of numerical system evidence. We have found three key aspects of this evidence which are of relevance to this. Integrated modelling enabled the simultaneous assessment of scenario and intervention impacts across a range of flow and water quality indicators. Similarly, integrated modelling enabled the evaluation of the performance variability of interventions across a range of climate, development, and policy scenarios. Finally, integrated modelling enabled analysis of effects manifesting in the lower basin which were driven by different scenario factors and interventions in the upper basin.

The numerical system evidence provided by integrated modelling provides opportunities to significantly improve policy, planning and decision-making. System evidence allows for the alignment of interventions and strategies to the specific scale of challenges identified in the basin, providing a strategic context for planners operating within separate planning and governance frameworks. Linked to this, system evidence opens the opportunity for the identification of interventions which proactively address challenges at source, as opposed to less systemically efficient reactive solutions. Furthermore, the integration provided by modelling also supports the prioritisation of multi-benefit schemes and the identification of critical second-order impacts of interventions. Finally, we have found that the situation of analysis alongside participatory and engagement activities, using the Multi-Level Integrated Water Management Framework is essential to build consensus and commitment, as well as collective organisational capability to deliver on the evidence produced.

Collaborative working across the project delivery team (GLA, MM, ICL) and steering group (boroughs, EA, NE, Thames Water) were essential in negotiating the interaction between these analytical and participatory dimensions of the project. The different perspectives and fields of expertise allowed for the framing and presentation of results in terms relevant to the organisational contexts of the system as well as the direction of focus of discussions to critical issues identified.

The three areas identified for further improvement in future applications of the Multi-Level Integrated Water Management Framework focus on continuing to improve the value of modelling activities as well as the collective organisational capability to deliver on the integration.

• (1) Disaggregation of modelled options and information from the sub-regional scale to the scale of the different planning frameworks to better support the inclusion of integrated strategy recommendations in existing planning frameworks of different scales.

• (2) Ongoing oversight and management of planning information and evidence to enable continued coordination of planning and delivery, enabled by data and digital tools.

• (3) Leveraging ongoing commitment for action across participating organisations through the use of more formalised working agreements.

Going forwards, the interplay of analytical and participatory dimensions is likely to grow in importance in the application of system-based approaches to water management. As advancements in AI and machine learning drive continued increases in analytical capability, demands on institutional arrangements and capabilities to deliver on this will change and develop. In future applications of the Multi-Level Integrated Water Management Framework, additional focus should also be given to this interplay in the context of water's fragmented institutional arrangements and the need to transition to more integrated forms of planning and governance to meet the challenges of climate change.

Planning and implementation of water efficiency programmes.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.