Abstract

Designing stormwater drainage systems considering deep uncertainty is a task that has no correct solution, rather, it can only be addressed by managing the system in a smart, robust way. Over recent decades, robust decision-making has been promoted as a solution to planning systems that are vulnerable to deep uncertainty. In this paper, we adopt a robust decision-making methodology to propose a stormwater drainage system in a coastal catchment in an arid region which is vulnerable to sea-level rise and increased precipitation intensity. We used bias-corrected precipitation and sea-level rise projections from a regional climate model, in addition to analyzing observed data. The decision-making methodology adopted is the Dynamic Adaptive Policy Pathways (DAPP). It involves building a decision tree with probable actions to consider when the stormwater system is expected to fail. The success of DAPP relies on continuous and extensive monitoring of the system and all components/factors that form risk or add to the vulnerability of the system, in addition to extensive simulations of the pre-identified actions that enable quick implementation of the solutions preceding failure of the system. This gives an early warning and aids the proactive execution of actions, hence making the system resilient to deep uncertainty. The DAPP for the study site is presented, and the advantages of relying on robust decision-making for arid regions are discussed.

1 Introduction and background

Long-term decision-making in the water policy domain is not a straightforward task due to the multitude of cascading factors that lead to deep uncertainty. These factors include uncertainty in climate change projections, which in turn rely on socio-economic projections, demographic projections, new technologies, and the expected extent of their implementation which affects emission scenarios, probability distributions of key parameters/variables, etc. (Haasnoot et al. 2011; Haasnoot et al. 2013; Kalra et al. 2014). Hence, high-risk levels are involved in the selection of the ‘most likely future scenario’ based on which of the ‘static optimal’ actions are implemented. However, problems occur when the future unfolds into a different reality (Haasnoot et al. 2013). 

Adaptive management or policymaking is a tool that helps manage a system that is subject to deep uncertainty by allowing the flexibility of revising/improving actions by constant monitoring and evaluation, as opposed to traditional ‘static’ decision-making (Chaffin et al. 2016; Haasnoot et al. 2011). Adaptive management involves extensive monitoring and feedback mechanisms, including both technical and governance procedures, which are naturally accompanied by additional costs compared to non-adaptive management practices. However, the expenses for learning in an active-adaptive management system are potentially less than expenditures in non-adaptive management due to the potential adverse effects inflicted by unknown unknowns (Hung et al. 2022).

The trend of moving from reactive to proactive decision-making has gradually gained interest over the past 2-3 decades. It involves an inverted decision-making process which encompasses policy-makers and practitioners to “agree-on-decisions” that would be implemented for all possible scenarios, rather than “agree-on-assumptions” and selecting a most-likely future scenario (Kalra et al. 2014). Upon agreeing on the decisions, continuous data monitoring and close interaction between practitioners and decision-makers are necessary to select from the agreed-upon actions as the future unfolds and renders current actions ineffective (Hallegatte et al. 2012; Kalra et al. 2014; Albrechts 2004; Ranger et al. 2010; Swanson et al. 2010). This framework is often referred to as robust decision-making.

In this paper, a robust decision-making framework known as the Dynamic Adaptive Policy Pathways (DAPP) (Haasnoot et al. 2013) is developed for a coastal catchment in an arid climate.

2 Study site

2.1 General description

The study site is a planned residential development with an area of 700 ha. It is a coastal catchment located on Bahrain’s east coast (Figure 1) which slopes at 2–3% towards the sea and has water production plants located to its north and south. Bahrain is an arid country with an average annual total rainfall of about 80 mm and temperatures ranging from about 15°C in the winter to over 40°C in the summer.

Figure 1 Location of the study site in Bahrain. The picture to the right shows an aerial view of the proposed main channels of the stormwater drainage system.

The proposed stormwater drainage system is comprised of open channels as indicated in Figure 1. The channels collect stormwater and slope with the natural topography at about 2% towards the sea. A typical cross-section of the site from west to east (left to right) is presented in Figure 2, and it shows that along the east coast, a natural depression occurs, which will be designed as detention basins that will collect the stormwater from the channels and direct it to the four planned outfalls.

Figure 2 A rough schematic of a cross-section of the study site (cross-section AA in Figure 1).

2.2 Vulnerability to climate change – Regional climate model projections

Based on the results of a regional climate model (Belušić et al. 2020) which is nested within six different global climate models (Seland et al. 2020; Yukimoto et al. 2019; Müller et al. 2018; Wyser et al. 2020; Séférian et al. 2019; Cherchi et al. 2019) and based on the SSP5-8.5 scenario, the volume of precipitation in Bahrain is projected to increase by about 4 mm/month in the near-term (2021–2040) and about 3 mm/month during the mid-term (2041–2060) (Swedish Meteorological and Hydrological Institute (SMHI) and United Nations Economic and Social Commission for Western Asia (ESCWA) 2022). Therefore, the currently observed average monthly precipitation of about 15 mm/month is expected to increase to about 19 mm/month by 2040, and to 22 mm/month by 2060. In addition to the increase in volume, the spatiotemporal variability of rainfall is projected to increase, and this would result in longer dry periods followed by intense rainfall (SMHI and ESCWA 2022). Being an arid country, bare soils in Bahrain are known to form a thin impermeable crust on the surface which leads to high runoff rates at the start of rain events. The thin crust eventually breaks due to the collision of rain droplets with the crust (Wheater 2007). Therefore, the increase in precipitation variability will lead to higher runoff rates generated from bare soils owing to the geological and climate characteristics of Bahrain, and in the case of unurbanized areas, which are typically not developed for sustainable stormwater drainage owing to the current policies, will similarly experience higher runoff rates. Therefore, increased precipitation variability will lead to increased vulnerability to climate change if urbanization continues without considering the expected changes in climate. 

Furthermore, the sea-level is projected to rise by 0.81 m by 2100 based on the SSP5-8.5 scenario relative to the 1995–2014 observed sea levels (Garner et al. 2021a; Garner et al. 2021b; Fox-Kemper et al. 2021; Masson-Delmotte et al. 2021). However, there is low agreement for projections up to 2100, but high agreement for the 2050 sea-level rise projections which is an expected rise of 0.4 m (Garner et al. 2021a; Garner et al. 2021b; Fox-Kemper et al. 2021; Masson-Delmotte et al. 2021). The 0.4 m sea-level rise corresponds to about an 11 mm/year rise.

2.3 Objectives of the planned stormwater drainage system

The proposed stormwater drainage system is designed using a 5-year return period, 30-minute duration design storm of 21 mm/hr (referred to as the 5-year, 30-minute storm hereinafter), which is in accordance with Bahrain’s Road Drainage Design Manual (AECOM 2017). The channels are sized using Manning's equation, and the detention basins are sized based on the volume naturally available from the topography. Since the catchment is steep (about a 2% slope), energy dissipators are included in the channels to reduce the velocity of flow. These are modeled using Manning’s roughness coefficient of 0.07, which corresponds to a channel lined with boulders (Arcement and Schneider 1989).

The stormwater drainage system is designed bearing two objectives in mind:

• that no stormwater flooding should be observed within the catchment, and
• that no tidal flooding should be experienced.

2.4 Future scenarios

Bearing in mind the potential vulnerabilities of the study site (Section 2.2), the resilience of the designed stormwater drainage system will be tested for projected increases in rainfall intensity and sea-level rise. First, historical rainfall and sea-level data will determine higher return period rainfall and sea-level rise. The corresponding rainfall and sea levels based on projected data will be used to determine an approximate timeline. This would indicate when the stormwater drainage system is expected to be functional and when it may fail. An increase in rainfall intensity is projected to be a total of about 7 mm/month by 2060; for the sake of this work a linear increase in rainfall intensity is assumed starting with a 28 mm/hr rainfall intensity in 2023 (which corresponds to a 5-year, 30-minute design storm determined from historical data) and increasing to 29 mm/hr in 2060. This assumes only one rain event per month which is very likely to occur in Bahrain based on historical data. A sea-level rise of 11 mm/year will be used as a downstream boundary at the outfalls. A linear trend will be assumed starting with a 0 m sea level in 2023.

3 Methodology

3.1. Dynamic Adaptive Policy Pathways

Dynamic Adaptive Policy Pathways (DAPP) is first introduced by Haasnoot et al. (2013). It combines the concepts of adaptive pathways and adaptation policymaking. The procedure, which is summarized in Figure 3, starts by identifying the attributes of the system, which in this case, is the stormwater drainage system of the study site. The objectives that, when achieved, would render the system successful, are clearly identified at the start. In this case, the objectives (described in Section 2.3) are to avoid roads and property inundations due to heavy rains, and to avoid tidal flooding along the coast. Next, the current and future states of the system are evaluated to determine whether the objectives are met in light of climate change and any uncertainties that may cause the system to fail. In this case, the vulnerabilities of the system (described in Section 2.4) are an increase in the rainfall intensity and sea-level rise. If, due to certain future states, the objectives are not achieved, this is addressed by taking action. Some of the actions proposed involve implementing sustainable drainage options and some are conventional solutions. The proposed actions are listed below:

• Adding green roofs to residences;
• Adding in-site runoff storage/retention elements such as permeable pavements, infiltration trenches, and bioretention basins;
• Increasing the surface area of the detention basins by 30–35% based on the available area on-site along the coast;
• Increasing the size of the drainage system; and
• Adding pumps to pump the stormwater from the detention basins to the sea.

Figure 3 Flow chart of the procedure for developing Dynamic Adaptive Policy Pathways.

The actions are modeled to ensure their effectiveness and determine for how long they are expected to meet the objectives. When an action no longer meets the objectives, it reaches its ‘tipping point’. When an action reaches its tipping point it can be combined with other actions to increase the lifespan of the system. The sequence of actions is called ‘adaptation pathways’, and it shows decision-makers the actions that can be taken to alleviate flooding. Finally, all adaptation pathways that are analyzed are then presented in a decision tree that communicates to decision-makers all actions that can be taken for different future states.

3.2 Analysis of sea-level data

Sea-level data are available at a time resolution of 10 minutes for a period of 8 years (from 2012 to 2020) from Bahrain’s Hydrographic Survey Directorate at a monitoring station close to the study site (Figure 4). The Vulntoolkit (Hill and Anisfeld 2021) tool written in R is used to perform the frequency analysis for the sea-level data.

Figure 4 Location of the station at which sea level is monitored relative to the study site.

The frequency analysis result is presented in Figure 5. The 2-, 50-, 100-, and 200-year return period tide levels (which correspond to 4, 0.16, 0.08, and 0.04 frequencies, respectively) are 0, 0.6, 0.7, and 0.75 m sea-level, respectively. These elevations will be used as boundary conditions at the outfalls.

Figure 5 Frequency analysis of 10-minute sea-level data at Sitra station, near the study site.

3.3 Validation of regional climate model precipitation projections

Prior to the use of the precipitation projections of the regional climate model (which are presented in Section 2.1), the daily historical simulations of the regional climate model (6 data sets are available from 6 different global climate models in which the regional climate model is nested, as described in Section 2.1) are compared to daily observed precipitation for the period 1961 to 2012. The comparison is performed by studying the empirical cumulative distributions (Figure 6) of all the datasets and the annual frequency analysis (Figure 7). The results show good agreement between simulated and observed datasets, hence, the projections are relied on when determining the adaptive pathways.

Figure 6 Empirical cumulative distribution function of daily observed and simulated rainfall for the period 1961 to 2012. The colored lines are simulated rainfall data from the regional climate model which is nested in 6 different global climate models.

Figure 7 Annual peak frequency analysis of daily observed and simulated rainfall for the period 1961 to 2012. The colored lines are simulated rainfall data from the regional climate model which is nested in 6 different global climate models.

3.4. Hydraulic modeling

The hydraulic modeling is performed using PCSWMM by Computational Hydraulics International (CHI). PCSWMM is a semi-distributed hydrological model that simulates rainfall-runoff using a series of compartments as shown in Figure 8. In this work, we used the land surface compartment and the conveyance compartment. The land surface compartment is responsible for converting rainfall into runoff. In PCSWMM, a catchment is treated as a non-linear reservoir, with precipitation and evaporation as inputs, and runoff, depression storage, and infiltration as outputs from each catchment. These quantities are updated at every time step. In the conveyance compartment, the runoff entering is routed using the Saint Venant flow equation which uses the conservation of mass and momentum equations to simulate gradually varied and unsteady flows.

Figure 8 Schematic presenting the conceptual model of PCSWMM software.

The proposed stormwater drainage system is modeled using the 2-year return period, 30-minute duration design storm (referred to as the 2-year, 30-minute storm hereinafter) which has an intensity of 21 mm/hr. The stormwater drainage channels (shown in yellow in Figure 1) are sized using Manning’s equation for the 2-year, 30-minute storm. The detention basin along the coast is modeled as 17 storage components connected via weirs. The dimensions of the storage components are taken from the available space, and their elevation from the natural topography. The runoff in the storage tanks will eventually flow to the sea via four wide channels (30 m × 0.5 m) as indicated in Figure 1. This simulation represents the proposed stormwater drainage system, i.e., the current scenario.

Following the simulation of the proposed system, a series of scenarios are created to test if the system would withstand the effect of climate change. The increase in rainfall and sea-level rise for the same return period are simulated. The sea-level rise and rain intensities for the simulated return periods are shown in Table 1.

Table 1 Summary of forcing and boundary conditions used to drive the hydraulic model.

Return Period (years)	2	50	100	200
30-minute duration design storm (mm/hr)*	21	57	65	73
Sea-level Rise (m above mean sea-level)	0	0.6	0.7	0.75

*Rainfall is assumed uniformly distributed

In the case that the current system fails, i.e., in-site floods and/or coastal flooding is experienced, one of the actions (explained in Section 3.1) is implemented and a new simulation is run. Logical combinations of the different actions are implemented. The actions are implemented as follows (Meierdiercks and McCloskey 2022; McCutcheon and Wride 2013; McCutcheon et al. 2012; Jeffers et al. 2022):

• In-situ storage/retention: this is implemented by calculating the total area covered by roads, playgrounds, and public open spaces. It is assumed that up to 80% of this area can be converted into permeable pavement or sustainable drainage units. Therefore, two scenarios are simulated; the first is that 50% of the area is converted to units that allow infiltration and storage, and the second is that 80% of the available area is converted. The storage units are distributed equally over the study site.
• Green roofs: this is implemented by calculating the total area covered by buildings (residences, public institutions, etc.). It is assumed that 80% of this area is covered by roofs and they can be converted into green roofs. Two scenarios are considered; the first is if regulations will dictate converting 40% of the total roof area into green roofs. The second is that regulations will dictate converting 80% of the total roof area into green roofs.
• Increase the volume of the detention basins: The detention basins are initially sized at 6 m × 200 m × 2 m, which is within the available space along the east coast of the study site. However, the surface area may be increased by about 30% based on the available space. The new dimensions are 8 m × 200 m × 2 m.
• Adding pumps at vulnerable outfalls: With repeated simulations, we noticed that one of the storage tanks is the most vulnerable owing to its volume and the incoming flow rate; therefore, a pump will be added to increase the rate of discharge from the vulnerable storage tank.
• Increasing the size of the proposed SW system: This involves increasing the dimensions of the stormwater drainage channels.

This resulted in various simulations which will be used to draw the decision tree that summarizes all acceptable adaptation pathways. The DAPP is presented in Section 4.

4 Results and discussion

The proposed stormwater drainage system is simulated with the combination of rainfall intensities and sea levels presented in Table 1. The proposed stormwater system is found to withstand a 30-minute, 50-year rain event and sea-level rise of the same return period. To accommodate higher rain intensities and sea-level rise, a series of actions, as explained in Section 3.3, are simulated. All logical combinations are simulated and represented in the DAPP in Figure 9, however, only successful pathways, i.e., those that satisfy the objectives of the system (explained in Section 2.3) are presented in the DAPP (Figure 10).

Figure 9 All simulated pathways, including both successful (opaque lines) and unsuccessful pathways (transparent lines). 

Figure 10 Dynamic Adaptive Policy Pathways for the study site.

The DAPP has all the simulated actions identified on the y-axis, and the rainfall and sea-level rise are represented on the x-axis. The two uppermost x-axes present the 30-minute duration rainfall and sea-level rise for different return periods. The corresponding return periods are shown on the third-from-the-top x-axis. These values are determined using historical sea-level and rainfall data. The two bottommost x-axes show the predictions of when sea levels and 30-minute duration, 5-year RP rainfall is expected to exhibit the values shown in the uppermost two axes. For example, a 5-year, 30-minute storm with an intensity of 57 mm/hr is expected to happen in the year 2200 based on regional climate model projections.

The DAPP is drawn as follows. Since the proposed storm water system can withstand a 50-year return period, but cannot withstand a 100-year return period, the proposed actions are implemented. If an action can extend the lifespan of the drainage system to the 100-year return period, a line is drawn that extends from the 50-year to the 100-year return period. It is observed from the multiple simulations that flooding occurs within the catchment, hence, actions such as increasing the capacity of the stormwater drainage channels or increasing in-site runoff storage and attenuation solutions are effective (refer to Figure 9 and Figure 10). For the 200-year return period, the main issue is flooding at the detention basins along the coast; therefore, solutions such as pumping water out of vulnerable basins, or increasing the size of basins are found to be the most effective. The colours of the lines drawn between the 100- and 200-year return period are representative of the preceding action that was implemented between the 50- and 100-year return period. For example, increasing the capacity of stormwater channels (represented in yellow) is successful between the 50- and 100-year return period, however, it does not solve the flooding issue between the 100- and 200-year return period, hence there is no yellow line drawn between the 100- and 200-year return period in Figure 10. Instead, adding pumps at detention basins (represented in light blue) is effective in solving the flooding issue between the 100- and 200-year return period, therefore, a pathway that starts with the yellow line and then continues with the light blue line is an example of an adaptive pathway that is predicted to maintain a system that is resilient to sea-level rise predicted to occur in around 2090, and rainfall intensity expected to occur at around 2300. By similarly arranging successful actions, various adaptive pathways are developed, as shown in Figure 10.

From all the simulations performed, the vulnerable elements of the study site are identified. This enhances our understanding of the study site and enables a quicker response to potential failures of the system.

An essential component of robust decision making is extensive, and continuous monitoring of the system and factors or parameters that pose risk to failure of the system. In the case of the study site, continuous sea-level and rainfall intensity at or near the site must be monitored and regularly analyzed to detect changes in trends. Additionally, runoff levels in the channels and detention basins must be monitored to allow for early warning. This illustrates how robust decision-making is, in fact, a continuous process, rather than a one-off solution, which is the typical approach of conventional decision-making.

Since the study site of concern is in the planning phase, issues such as land ownership, access for maintenance, etc., which will cause challenges for implementing some of the actions are brought up at this stage. This allows decision makers to update building guidelines in an appropriate way that will allow the implementation of some of the actions, such as converting part of the residences’ roofs into green roofs and granting access for maintenance, or general guidelines such as limiting runoff outflow from private property, which can only be implemented by sustainable drainage or rain harvesting solutions. Furthermore, since sustainable drainage solutions are relatively new ideas in this region, raising public awareness is necessary for their successful implementation (Chaffin et al. 2016; Flynn and Davidson 2016). Hence, starting at an early stage is crucial to avoid facing social barriers, and for the proper execution of some of the solutions that may be necessary to ensure a successful system in the future.

Governance-wise, it is essential that multiple stakeholders, such as environmental agencies, ministry of works, municipalities, non-government organizations, media agencies, etc., are involved to enable a holistic and proactive approach that will alleviate problems in the future. This will only be considered if a robust decision-making approach is adopted.

5 Conclusion

Dynamic Adaptive Policy Pathways (DAPP) for a coastal catchment in an arid region is presented in the current work to provide critical information for decision-makers when planning under deep uncertainty. Rainfall and sea-levels from a regional climate model are used to predict increases in rainfall intensities and sea-level rise. This enabled estimating the lifespan of the drainage system, which is designed based on regular engineering procedures. Once the expected lifespan is determined, actions which will extend the lifespan of the system are simulated, and multiple alternative plans (known as adaptive pathways) are established. This involves rigorous analysis; however, it helps responsible personnel identify early on the most vulnerable locations/components of the system that will be affected by climate change. It also demonstrates at an early stage the potential solutions that may be implemented in the future and, hence, it is possible to prepare for it either governance-wise by early stakeholder engagement, through public awareness campaigns, and by changing necessary building guidelines, etc. Early stakeholder engagement is essential to alleviate complex future problems related to the implementation of potential solutions.

Robust decision-making involves computationally extensive study of the system, in addition to continuous monitoring and involvement of stakeholders, which imposes a financial burden at the planning and implementation phases, in addition to the complexity of liaising with different stakeholders; however, the benefits of an extended lifespan of the system in addition to a lower risk of failure are among the benefits of robust decision making.

Acknowledgements

The authors utilized the educational version of PCSWMM software, which is provided by Computational Hydraulics International (CHI) to researchers and academics under an educational grant agreement.

References

1. AECOM. 2017. Bahrain Roads Design Manual – Part 9: Road Drainage. Kingdom of Bahrain: Ministry of Works.
2. Albrechts, L. 2004. "Strategic (Spatial) Planning Reexamined." Environment and Planning B: Urban Analytics and City Science 31 (5): 743–758. https://doi.org/10.1068/b3065 
3. Arcement, G.J., and V.R. Schneider. 1989. Guide for Selecting Manning's Roughness Coefficients for Natural Channels and Flood Plains. United States Geological Survey: Geological Survey Water-Supply Paper. Vol. 2339. Washington: U.S. Gov. Print. Off.
4. Belušić, D., H. de Vries, A. Dobler, O. Landgren, P. Lind, D. Lindstedt, R.A. Pedersen, et al. 2020. "HCLIM38: A Flexible Regional Climate Model Applicable for Different Climate Zones from Coarse to Convection-Permitting Scales." Geoscientific Model Development 13 (3): 1311–1333. https://doi.org/10.5194/gmd-13-1311-2020
5. Chaffin, B.C., W.D. Shuster, A.S. Garmestani, B. Furio, S.L. Albro, M. Gardiner, M. Spring, and O.O. Green. 2016. "A Tale of Two Rain Gardens: Barriers and Bridges to Adaptive Management of Urban Stormwater in Cleveland, Ohio." Journal of Environmental Management 183 (Pt 2): 431–441. https://doi.org/10.1016/j.jenvman.2016.06.025
6. Cherchi, A., P.G. Fogli, T. Lovato, D. Peano, D. Iovino, S. Gualdi, S. Masina, et al. 2019. "Global Mean Climate and Main Patterns of Variability in the CMCC‐CM2 Coupled Model." Journal of Advances in Modeling Earth Systems 11 (1): 185–209. https://doi.org/10.1029/2018MS001369
7. Flynn, C.D., and C.I. Davidson. 2016. "Adapting the Social-Ecological System Framework for Urban Stormwater Management." Ecology and Society 21 (4): 19. https://doi.org/10.5751/ES-08756-210419
8. Fox-Kemper, B., H.T. Hewitt, C. Xiao, G. Aðalgeirsdóttir, S.S. Drijfhout, T.L. Edwards, N.R. Golledge, et al. 2021. Ocean, Cryosphere and Sea Level Change. In Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, pp. 1211–1362. https://doi.org/10.1017/9781009157896.011
9. Garner, G.G., T. Hermans, R.E. Kopp, A.B.A. Slangen, T.L. Edwards, A. Levermann, S. Nowicki, et al. 2021a. "IPCC AR6 Sea Level Projections". Accessed online: Aug 2022, Zenodo. https://doi.org/10.5281/zenodo.5914709
10. Garner, G.G., R.E. Kopp, T. Hermans, A.B.A. Slangen, G. Koubbe, M. Turilli, S. Jha, et al. 2021b. "Framework for Assessing Changes to Sea-Level (FACTS).” Geoscientific Model Development 2023-14.
11. Haasnoot, M., H. Middelkoop, E. van Beek, and W.P.A. van Deursen. 2011. "A Method to Develop Sustainable Water Management Strategies for an Uncertain Future." Sustainable Development 19 (6): 369–381. https://doi.org/10.1002/sd.438
12. Haasnoot, M., J.H. Kwakkel, W.E. Walker, and J. ter Maat. 2013. "Dynamic Adaptive Policy Pathways: A Method for Crafting Robust Decisions for a Deeply Uncertain World." Global Environmental Change 23 (2): 485–498. https://doi.org/10.1016/j.gloenvcha.2012.12.006
13. Hallegatte, S., A. Shah, C. Brown, R. Lempert, and S. Gill. 2012. “Investment Decision Making Under Deep Uncertainty – Application to Climate Change.” World Bank Policy Research Working Paper no. 6193. Available at SSRN: https://ssrn.com/abstract=2143067
14. Hill, T., and S.C. Anisfeld. 2021. "VulnToolkit: Analysis of Tidal Datasets." Accessed online: Sep 2021, Version 1.1.4. Available at: https://doi.org/10.6084/m9.figshare.14161202.v1
15. Hung, F., B.F. Hobbs, A. McGarity, and X. Chen. 2022. "A Modeling Framework for Assessing the Value of Learning in Dynamic Adaptive Planning: Application to Green Infrastructure Investment Evaluation." Water Resources Research 58 (8). https://doi.org/10.1029/2021WR031622
16. Jeffers, S., B. Garner, D. Hidalgo, D. Daoularis, and O. Warmerdam. 2022. "Insights into Green Roof Modeling using SWMM LID Controls for Detention-Based Designs." Journal of Water Management Modeling 30:C484. https://doi.org/10.14796/JWMM.C484
17. Kalra, N., S. R. Hallegatte, C. Lempert, C. Brown, A. Fozzard, S. Gill, and A. Shah. 2014. “Agreeing on Robust Decisions: New Processes for Decision Making under Deep Uncertainty.” Policy Research Working Paper No. 6906. World Bank, Washington, DC. http://hdl.handle.net/10986/18772
18. Masson-Delmotte, V., A. Pirani, P. Zhai, S.L. Connors, C. Péan, S. Berger, N. Caud, Y. Chen, et al. 2021. "Climate Change 2021: The Physical Science Basis.” Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. IPCC 6th Assessment Report. Cambridge University Press, Cambridge, United Kingdom.
19. McCutcheon, M., and D. Wride. 2013. "Shades of Green: Using SWMM LID Controls to Simulate Green Infrastructure." Journal of Water Management Modeling R246-15. https://doi.org/10.14796/JWMM.R246-15  
20. McCutcheon, M., D. Wride, and J. Reinicke. 2012. "An Evaluation of Modeling Green Infrastructure using LID Controls." Journal of Water Management Modeling R245-12. https://doi.org/10.14796/JWMM.R245-12
21. Meierdiercks, K. and N. McCloskey. 2022. "The Effectiveness of Centralized Versus Decentralized Green Infrastructure in Improving Water Quality and Reducing Flooding at the Catchment Scale." Journal of Water Management Modeling 30: C490. https://doi.org/10.14796/jwmm.c490
22. Müller, W.A., J.H. Jungclaus, T. Mauritsen, J. Baehr, M. Bittner, R. Budich, F. Bunzel, et al. 2018. "A Higher‐resolution Version of the Max Planck Institute Earth System Model (MPI‐ESM1.2‐HR)." Journal of Advances in Modeling Earth Systems 10 (7): 1383–1413. https://doi.org/10.1029/2017MS001217
23. Ranger, N., A. Millner, S. Dietz, S. Fankhauser, A. Lopez, and G. Ruta. 2010. "Adaptation in the UK: A Decision-Making Process." Centre for Climate Change Economics and Policy 9: 1–62.
24. Séférian, R., P. Nabat, M. Michou, D. Saint‐Martin, A. Voldoire, J. Colin, B. Decharme, et al. 2019. "Evaluation of CNRM Earth System Model, CNRM‐ESM2‐1: Role of Earth System Processes in Present‐Day and Future Climate." Journal of Advances in Modeling Earth Systems 11 (12): 4182-4227. https://doi.org/10.1029/2019MS001791
25. Seland, A., M. Bentsen, D. Olivie, T. Toniazzo, A. Gjermundsen, L. Seland Graff, J. Boldingh Debernard, et al. 2020. "Overview of the Norwegian Earth System Model (NorESM2) and Key Climate Response of CMIP6 DECK, Historical, and Scenario Simulations." Geoscientific Model Development 13 (12): 6165-6200. https://doi.org/10.5194/gmd-13-6165-2020
26. Swanson, D., S. Barg, S. Tyler, H. Venema, S. Tomar, S. Bhadwal, S. Nair, D. Roy, and J. Drexhage. 2010. "Seven Tools for Creating Adaptive Policies." Technological Forecasting and Social Change 77 (6): 924–939. https://doi.org/10.1016/j.techfore.2010.04.005
27. Swedish Meteorological and Hydrological Institute (SMHI) and United Nations Economic and Social Commission for Western Asia (ESCWA). 2022. Future Climate Projections for the Mashreq Region: Summary Outcomes. RICCAR Technical Report, Beirut. https://www.unescwa.org/publications/riccar-future-climate-mashreq 
28. Wheater, H.S. 2007. "Modelling Hydrological Processes in Arid and Semi-Arid Areas: An Introduction." In Hydrological Modelling in Arid and Semi-Arid Areas, Cambridge University Press, Cambridge, UK. https://doi.org/10.1017/CBO9780511535734
29. Wyser, K., E. Kjellström, T. Koenigk, H. Martins, and R. Döscher. 2020. "Warmer Climate Projections in EC-Earth3-Veg: The Role of Changes in the Greenhouse Gas Concentrations from CMIP5 to CMIP6." Environmental Research Letters 15 (5): 54020. https://doi.org/10.1088/1748-9326/ab81c2
30. Yukimoto, S., H. Kawai, T. Koshiro, N. Oshima, K. Yoshida, S. Urakawa, H. Tsujino, et al. 2019. "The Meteorological Research Institute Earth System Model Version 2.0, MRI-ESM2.0: Description and Basic Evaluation of the Physical Component." Journal of the Meteorological Society of Japan 97 (5): 931-965. https://doi.org/10.2151/jmsj.2019-051