Flood risk assessment methodology for planning under climate change scenarios and the corresponding change in land cover

No organization or local government can be on a long-term path to more sustainable and resilient development without first addressing climate change. Without taking the impacts of climate change into consideration, today's development gains may be wiped out tomorrow (UN Habitat 2014). Climate change has become one of the most significant challenges of the 21st century and while there may be uncertainty surrounding the scale, scope, and pace of climate change, one thing is clear – cities and towns everywhere are exposed to some level of climate change-driven impacts. The scientific evidence of climate change is overwhelming, and the global impacts of climate change may be severe. Although climate change is often seen as a long-term challenge, the impacts are being experienced currently, through unprecedented global trends and through more localized severe weather events (TCPA 2018). Today, planners have the opportunity and obligation to address the challenge of global and regional climate change and any responsible governing authority requires planning for a wide range of contingencies (APA 2011). The third National Climate Assessment notes that certain types of weather events have become more frequent and/or intense, including heat waves, heavy downpours, and, in some regions, floods and droughts. Sea levels are rising, oceans are becoming more acidic, and glaciers along with the Arctic sea ice are melting (Melillo et al. 2014) and scientists are predicting that these changes will continue and even increase in frequency or duration over the next 100 years. The first step in planning for these challenges is to identify the effects of climate change with tangible and specific metrics, using the best available science.

We have recently witnessed how natural disasters cost lives and destroyed urban spaces and communities (Boyd 2017). The issue at stake is not just the climate change impacts but also a whole spectrum of environmental changes that interplay with interdependent and rapidly globalizing human societies (Folke et al. 2011) and the resulting risks that human settlements and humanity in general may face (Jabareen 2013). It is clear that in order to reduce the risk and impact of these threats and to increase the safety and well-being of residents, cities and their communities must be more resilient and prepared to address the threats head-on. If they are not, their urban communities will live under continuous threat, and more and more become vulnerable to risks (UNISDR 2010). While climate change will have a lasting impact on people and the environment, it will also define future economic progress. Only those places that can demonstrate climate resilience will be able to secure insurance and investment (TCPA 2018). In order to manage the effects of climate change on local operations and infrastructure in both the short and long term, a development of a strategic policy for climate change adaptation will necessitate adjustments to facilities, infrastructure, and operational activities. By integrating climate change considerations into various planning processes, society will be better prepared to effectively respond and to ensure continued safety of communities and infrastructure.

In some regions, climate change is expected to increase both the magnitude and frequency of extreme precipitation events, which may lead to more intense and frequent river flooding. The general problem is that we can project the effects of climate change on precipitation both temporally and spatially, but do not translate this into longer duration storms and the associated runoff in individual watersheds. Since there is spatial and temporal variability in rainfall and we are projecting these values under a nonstationary climate, it is difficult to observe trends in rainfall, because we often rely on data and model outputs that are at the wrong scales (Singer & Michaelides 2017). Global rainfall datasets and output from climate models are typically resolved on timescales of days or months and at spatial scales larger than most river basins. It is unknown how particular scenarios of climate change might affect rainfall patterns and runoff generation across a basin. In general, scientists have a poor understanding of how a warming climate will affect the magnitude, timing, and spatial patterns of rainfall. Yet these aspects of the climate system are fundamental to assess the sustainability of water resources, especially flood risks. This information gap creates challenges for predicting expected hydrologic patterns and processes under shifting climate drivers. It therefore limits progress in research and management of environmental flood risks in regional communities.

This study has developed a methodology to assess the riverine flood hazard under climate change scenarios in 43 regions worldwide, with the contributing watershed areas ranging from approximately 1.3 square kilometers to 2,738 square kilometers. The scope of flood modeling was limited to stream channel networks and did not consider flooding of independent surface bodies, stormwater systems, or surface ponding. The methodology developed in this study can assist in simulating watershed rainfall and calculating the associated run off under climate change scenarios worldwide, additionally under the associated changes in land cover in the contiguous United States; thus, enabling successful future regional flood risk assessment. This methodology involves the process of coupling the identified storm to a runoff model to explore scenarios of climate change and how they might affect the magnitude and the frequency of runoff and the corresponding flood hazard. Once the flood hazard has been mapped, this methodology serves as additional guidance to conduct vulnerability assessments to identify adaptation strategies, thus providing a framework for decision-makers to plan for a changing climate.

The science behind weather systems and climate change is complex, with many interacting factors and processes. For the project described in this paper, a team comprising climate scientists, civil and environmental engineers, ecologists and natural resources experts worked together to generate site-specific climate data, conduct site-specific flood modeling and the associated vulnerability assessments. Two future time horizons, the year 2030 (near-term) and year 2050 (longer-term), along with two future emission scenarios denoted as Representation Concentration Pathway (RCP) 4.5 (moderate emissions scenario) and 8.5 (high emissions scenario) were chosen. Emission scenarios are based on assumptions about future worldwide changes in demographic development, socio-economic development, and technological change that result in different greenhouse gas concentrations in the atmosphere. These are non-probabilistic scenarios; in other words, no one scenario is more likely than another. They are useful for the purposes of conducting what-if exercises to determine a range of possible futures.

Climate information for each site is generated to capture the average 30-year historical time period from 1976 through 2005 using daily information located over each site. Climate projections are based on recent global climate model (GCM) simulations developed for the International Panel on Climate Change (IPCC), Coupled Model Intercomparison Project (CMIP) Phase 5 (Hibbard et al. 2007; Moss et al. 2008, 2010). Under the CMIP protocol (CMIP5 Data Search | CMIP5 | ESGF-CoG n.d.), specified radiative forcing of atmospheric warming was simulated using 32 global climate models to provide scenarios associated with RCP 4.5 and 8.5 (van Vuuren et al. 2011). Climate variables extracted from these emission scenarios include information on average daily temperature (°C), maximum daily temperature (°C), minimum daily temperature (°C), and daily precipitation amount (mm). Climate projections are generated for a decade around the target years 2030 and 2050. For 2030, climate data from 2026 to 2035 are generated to acquire the decadal average for the year 2030 and for 2050, and climate data from 2046 to 2055 are generated to acquire the decadal average for the year 2050.

Climate data sources used are different for sites within the contiguous 48 states versus elsewhere.

For sites within the contiguous United States: Historical data used are DAYMET (Thornton et al. 2012) of approximately 1 km spatial resolution. Climate projections are derived from climate model simulations from the US National Center for Atmospheric Research Community Climate Model (CCSM4) simulations prepared for the IPCC-AR5 (CESM Model n.d.). CCSM4 was chosen since it provides very consistent and moderate climate representation across various climate regions. The data source for projections is derived from the Localized Constructed Analogs (LOCA) CCSM4 data at approximately 6 km spatial resolution over the US (Pierce et al. 2014). LOCA is one of the statistical downscaling methods designed to generate climatological data with higher spatial and temporal resolution, which transforms the large-scale GCM with lower special and monthly temporal data to a 6 km × 6 km (1/16th of a degree in latitude and longitude). Pierce's (Pierce et al. 2016) study on the LOCA application in California gives a good representation of the extreme events and the spatial coherence compared to other statistical downscaling methods.

For sites outside the contiguous 48 states: Historical data used are the ½ degree global dataset provided by the ISI-MIP project at the Max Planck Institute for Meteorology (Hempel et al. 2013). Climate projections used data from HadGEM2-ES dataset, also provided by the ISI-MIP project.

For sites within the 48 contiguous states, a 30-year climatological baseline was established from DAYMET gridded historical climate data. The baseline is the average daily mean for the 30-year gridded time series, which provides an average daily weather pattern for daily temperatures (°C), maximum daily temperature (°C), minimum daily temperature (°C), and daily precipitation amount (mm).

The 30-year average climate variables were then used to develop decadal time series of daily climate values for the decades around 2030 and 2050. For each projected climate scenario, RCP 4.5 and RCP 8.5, a daily anomaly was computed for the selected model scenario (projected year – 30-year average daily base year for each variable of interest) over the ten-year period, 2026–2035 for 2030 and 2046–2055 for 2050. These provide daily climate anomaly records representing the decades centered at 2030 and 2050. Historical and projected LOCA data were used for this step.

To create a ten-year period of daily climate values for each decade of interest for each site, the daily anomaly time series was applied to the DAYMET historical 30-year average daily data. This procedure was done for each variable of interest (t_min, t_max, and precip) and for each future scenario (RCP 4.5 and RCP 8.5), to produce a ten-year daily time series for each combination. Spatial resolution of the new ten-year data is approximately 1 km.

For sites outside the 48 contiguous states, historical data were extracted for the 30-year period (1975–2004) directly from the ISI-MIP HadGEM2-ES model dataset. Projected data for two decadal periods (2026–2035 and 2046–2055) were also extracted directly from this same dataset. The values extracted are daily temperatures (°C), maximum daily temperature (°C), minimum daily temperature (°C), and daily precipitation amount (mm).

R statistical packages were created and used to generate the climate summary. The DaymetLOCA package generates the site-specific projected climate data. The ClimatePrimers package extracts and provides daily summary data for temperature (min and max) and precipitation for the specified 30-year historical period and the projected two decade periods. These daily values are then averaged across a specified location such as site or basin boundary. Thses data are then used by hydrology applications to determine consecutive 3-day rainfall events that is used to calculate the flood risk for the target years under RCP 4.5 and 8.5.

A design storm is a hypothetical storm used to design infrastructure, evaluate flood hazards, and/or inform land use planning and resource management. Working with projected climate data for flood modeling has some limitations. In order to acquire precipitation data for a target year in the future, climate data are projected for ten years (±5 years of target year) and averaged to reduce errors and improve accuracy of projected data for the target year. Projected daily precipitation data from 2026 to 2035 and 2046 to 2055 were used to estimate design storms for emission scenarios in 2030 and 2050, respectively. Initially, each ten-year dataset was averaged, however, averaging daily precipitation data across the ten-year period dampens the various storms since the occurrence of storms within each year varies temporally. As a result, algorithms were developed to screen the raw data and identify the biggest storm in each year. The identified annual storm events were then averaged (one storm per year for ten years), based on a threshold of eliminating any years with non-storm events, by omitting values below the 50th percentage. The same approach was applied to historical weather data to acquire a baseline storm that could be compared across the projected scenarios. Daily precipitation data from 1996 to 2005 were used to estimate baseline design storms for the year 2000.

Since the design storms are based on just ten years of climate projections data, it does not present a sufficient dataset to identify extreme precipitation events (e.g., 100-year or 500-year storms). Instead, they represent much smaller events that could occur more frequently. Projection methods did not allow for determination of design storm probability. Hence, this novel approach of identifying frequent (high-recurrence interval) storms is used as an indicator to compare across the various time frames and climate scenarios. Three-day storm events were used as design storms for flood modeling because rainfall occurring over consecutive days can cause soil saturation, overland flow, and compounding runoff.

Worldwide, 104 regions were studied and assessed to identify the need for flood risk assessments at the individual locations. The following criteria were used to disqualify sites from being subjected to flood risk assessment:

• If the location is geographically separated from watersheds and does not have any streams in or around the location

• If the site is located at a high elevation causing any flooding to run off and discharge outside the region

• If the region does not have any documented historical flooding issues, as well as climate projections indicate no significant increase in precipitation for the future target years

• If the location has sophisticated stormwater management systems already in place to manage flooding

• If the location does not have the availability of crucial spatial data needed for flood modeling.

Based on these criteria, the list of assessed locations was reduced to 43 sites that qualified for full flood risk assessments.

For the locations selected for the flood risk assessment, watershed boundaries were delineated. Based on the region of interest to conduct the flood modeling, the associated area of land upstream that contributes to drainage is identified as the watershed area to be delineated.

For most locations within the United States, USGS's online StreamStats application was used for delineating the watershed. Based on the selected point of interest, the application generates the watershed boundary that contributes to any drainage from the area upstream of the selected point. The shapefile of the boundary can then be downloaded from the StreamStats application. If StreamStats does not have watershed data for any particular location within the US, then Hydrologic Unit Code (HUC) shapefiles were accessed from the Natural Resources Conservation Service (NRCS) database. The NRCS database has the entire United States delineated into watersheds at different resolutions (more than four different sizes of watersheds), and the associated shapefiles can be found online and downloaded from the NRCS/USDA data gateway. In order to choose the appropriate HUC boundary, all HUC sizes shapefiles are first downloaded for the region of interest. These shapefiles are then brought into ArcGIS to investigate how the different sizes of HUC boundaries are related. Using the digital elevation model (DEM), aerial imagery and/or a topographic map, analyses were conducted to identify the appropriate watershed area by deleting HUC boundaries that are hydrologically separated from the area contributing to drainage from the point of interest. For sites outside of the USA, where HUC boundaries are not available, watersheds were delineated using the ArcHydro tools package in ArcGIS using the available DEM and the point of interest as input. This tool uses a point shapefile (point of interest) and the DEM of the area to delineate the contributing runoff area upstream of the selected point. Alternatively, the Spatial Analyst toolbox called ‘Watershed’ in ArcGIS can also be used to delineate the watershed using the DEM, point of interest, and flow direction raster.

Floodplain area mapping includes hydrologic and hydraulic regional analyses to model and project the inundation due to rainfall events. The U.S. Army Corps of Engineers' (USACE) Hydrologic Engineering Center (HEC) Hydrologic Modeling System (HMS) software was used for hydrology modeling to simulate runoff and estimate discharge over the contributing watershed following design storms. HEC–River Analysis System (RAS) 2D software was used for hydraulic modeling to evaluate potential stream channel overflow at each of the locations. ESRI ArcGIS tools, such as ArcHydro, were used for preprocessing geospatial data used in hydrologic and hydraulic modeling.

Figure 1 shows the process workflow involved in flood modeling.

This section provides details on the various steps involved in creating and simulating the hydrology and hydraulic models.

Various national and international open source GIS data repositories were accessed in addition to the independent organization spatial geodatabase, to collect the required geospatial data needed to conduct flood modeling:

• Elevation data: Independent organization spatial geodatabase, USDA, USGS, NOAA, ArcOnline and other state/county/city data repositories

• Land cover data: NLCD, MC2 (a dynamic global vegetation model that simulates vegetation type in response to climate change projections under various emission scenarios (Bachelet et al. 2015) and other state/county/city data repositories

• Soils data: USGS Web Soil Survey and other state/county/city data repositories

• Watershed boundaries: USGS HUC boundaries, USGS StreamStats and ArcHydro Tools

• Stream network: NHD, independent organization spatial geodatabase

• Common infrastructure picture (CIP) data: Independent organization spatial geodatabase containing location information on buildings, roads, streams, etc.

• Environmental data: Independent organization spatial geodatabase.

Typically, data collected from open source databases need additional processing before they can be used for modeling. Varying spatial resolution, extent, quality of data, and attributes as well as varying data formats need to be processed before being compiled for use.

A hydrologic model is a representation of how a particular watershed reacts due to rainfall runoff. The hydrologic model generates a hydrograph by taking account of the physical watershed characteristics and precipitation data. A hydrograph illustrates flow (discharge) at a specific location in the watershed in cubic feet per second over the specified period of time.

Analyzing the hydrologic network of any watershed is the first step in identifying how the bodies of water, stream network, and watershed basins are interconnected. The spatial location where the final hydrograph is needed determines the total watershed basin to be evaluated. This watershed basin is the boundary for which climate projection data are generated for various scenarios.

1. Hyetograph development: Using the selected design storms (‘Storm selection for flood modeling’ section) from climate projections data, hyetographs are developed for various scenarios using the NOAA Atlas 14 tool. A hyetograph is a representation of a storm's precipitation intensity over time. Since the temporal resolution of the projected 3-day storm is in daily summaries (24-hour intervals), NOAA Atlas 14 is used to create hyetographs of three 24-hour rainfall distributions with 5-minute intervals over the 3 days. The late-peaking storm distribution was selected for all locations.

Of note, these distributions are for design storms, not necessarily actual storms. A real storm may not follow the same pattern. However, this is the most widely adopted approach to simulate probabilistic pattern for a future storm for modeling purposes.

The final hyetograph is an incremental depth of rainfall every 5 minutes for 3 days.

• 2.

  Arc Hydro Tools: ESRI's ArcHydro tool was used to calculate the following parameters:

• • •

    Area using watershed data

  • •

    Longest flow path using watershed and stream network data

  • •

    Slope using elevation and watershed data

  • •

    Curve number (CN) using soils, land cover and watershed data.

Within the developed hyetographs, the rainfall intensity progressively increases until it reaches a maximum and then gradually decreases within each 24-hour interval. Lag time defines where this maximum precipitation occurs during the 3-day hyetograph and how fast that maximum is reached. Using the above listed parameters, lag time is calculated using the Soil Conservation Service (SCS) method.

• 3.

  Unit hydrograph development: The SCS method (Natural Resource Conservation Service's ‘National Engineering Handbook, Chapters 15 and 16) is used for developing the unit hydrograph.

• 4.

  Calculating infiltration rates and imperviousness: The initial and constant method is used to calculate the infiltration losses using soils data. The infiltration rates account for precipitation losses in a watershed. In general, only a fraction of precipitation ends up contributing to actual runoff, as some of it is lost in infiltration and storage. Based on the NRCS National Engineering Handbook (Chapter 7), infiltration rates are calculated using soils data and the corresponding depth to the water table. For an area with the water table above or below 100 cm, the NRCS provides a range of infiltration rates for each type of soil group. After identifying the infiltration rates for each portion of the land in the sub-basin, a constant infiltration rate is calculated using a weighted area analysis.

The percent of impervious land is calculated using land cover data. After identifying the imperviousness for each portion of the land in the sub-basin, the total percent imperviousness is calculated using a weighted area analysis. For this analysis, the soil type and impervious areas are assumed to not change over the years 2030 and 2050, but additional modeling could be completed in the future to account for planned development.

• 5.

  HEC-HMS model: Using the developed hyetographs, unit hydrographs, watershed characteristics (area, lag time, etc.), infiltration rates and percent imperviousness, a HEC-HMS model is set up. A hydrologic network is built including sub-basins, river reaches, reservoirs, etc. within the model. Junctions are included where tributaries from sub-basins join a river or where two or more sub-basins drain. River routing is conducted based on available data using either the kinematic wave or lag routing methods. The model is then computed for the various climate scenarios.

HEC-HMS generates the final hydrograph for each scenario with flow rate/discharge at the specified location over 72 hours with 5-minute intervals.

A hydraulic model is used to simulate the floodplain area at any location using the discharge hydrographs from hydrology modeling, elevation data, land cover data, environmental data, and CIP data.

1. Building the 2D mesh: A 2D mesh is a digital representation of the physical terrain of the model extent. The extent of the model is selected based on the site boundary and adjacent areas of interest for flood modeling and set up within the appropriate spatial projected coordinate system (state plane coordinates for most of the US sites). The elevation data layer is imported into the HEC-RAS 2D model to acquire the digital terrain and water surface elevations. If the channel bathymetry data are not captured within the elevation data, channel areas are mapped and elevation with the channel is dropped to account for channel depth/capacity. Based on the spatial resolution, elevation data are further manipulated to account for hydraulic structures like culverts, bridges, dams, etc. Stream and road network data are imported into the model and assigned as breaklines. Breaklines help in stabilizing the model by refining the cell sizes within the 2D mesh. CIP data are imported into the model to assign buildings and structures as obstructions within the 2D mesh area. Land cover data are imported into the model and Manning's n roughness coefficients are assigned to each land cover classification. These roughness coefficients define the resistance for the terrain in the 2D flow area and have a large impact on the model results. The NLCD dataset is used for baseline land cover data and the MC2 dataset is used for projected landcover for the years 2030 and 2050. The USACE recommended Manning's n coefficient values were used within the models.

2. 2D boundary conditions: Once the 2D mesh is created, the boundary conditions are established at upstream (inflow) and downstream (outflow) ends of the channel. The inflow boundary condition is used to load the hydrologic information using the flow hydrograph. Since the flood modeling is conducted based on a projected 3-day storm, the inflow boundary conditions will be set to unsteady flow data. The outflow boundary condition is used to define the outflow discharge information in the form of water surface elevation (typically set as ‘normal’ depth).

3. Computational options: The simulations are computed using full momentum equations, for higher accuracy, compared to diffusion wave equations. A computational time interval of 6 seconds is used to acquire accurate and stable results.

4. Flood maps displaying the extent of inundation are generated for the target years of 2030 and 2050 under the RCP scenarios of 4.5 and 8.5.

Based on the projections from floodplain modeling, impacts on built infrastructure and assets are analyzed to identify various vulnerabilities. The key vulnerable infrastructure assets are identified, listed, and assessed based on the severity of the vulnerability and confidence in the assessment.

The CIP data contain spatial data layers for site assets like real property (buildings, structures, housing, and storage) and transportation (roads, vehicle parking, bridges). These CIP data are overlaid on the projected inundation area to identify vulnerable assets. Assets within the projected inundation area are identified to be highly vulnerable and assets within 38 feet of the inundation boundary are identified to be moderately vulnerable. The National Standard for Spatial Data Accuracy (NSSDA) is typically used to report the radius of a circle of uncertainty such that the true or theoretical location of a point falls within that circle 95% of the time. Based on the horizontal positional accuracy of the base map, the region of uncertainty is defined to be 38 feet, the same value that Federal Emergency Management Agency (FEMA) uses.

A high confidence designation indicates the CIP data fully provided the information needed on the infrastructure assets and a satisfactory vulnerability assessment was conducted. Medium confidence indicates limited data availability and not all entities have enough information for assessment. Low confidence indicates that questionable entries/data were provided for certain assets and it needs to be confirmed by local authorities.

An extensive and comprehensive adaptation library (with over 200 alternatives) has been developed and mapped to the identified vulnerabilities, to determine applicable adaptation projects for consideration. Based on the vulnerabilities projected due to flooding, a set of adaptation strategies is proposed for consideration. Suggested adaptation projects are rated by their difficulty to implement and their relative efficacy. Implementation is ranked from 3 to 1, with 3 being the easiest to implement and 1 being the most difficult to implement. Efficacy is ranked from 1 to 3, 1 being the least effective and 3 being the most effective. The ecological impacts related to adopting each of the projects is stated to be positive if no negative impacts are expected. If these projects are expected to have negative ecological impacts, they are rated 1 (being as low negative impacts) through 3 (being high negative impacts) along with a corresponding literature reference/link.

Of the 43 national and international locations that were studied as a part of this research, results for a representative location are presented for this case study. The size of the contributing catchment area for this location was approximately 370 square kilometers.

Table 1 displays the comparison of baseline (reference year 2000) and projected precipitation totals for high-recurrence interval rainfall events at the location. Rainfall events are projected for the years 2030 and 2050 assuming RCP values of 4.5 (moderate emissions scenario) and 8.5 (high emissions scenario).

Hyetographs were generated (Figure 2) for the identified storms (Table 2) using NOAA Atlas 14. NOAA Atlas 14 is the precipitation-frequency atlas of the US and includes associated confidence limits along with additional information such as temporal distributions and seasonality.

Hydrographs (Figure 3) providing discharge flow for the location were generated as a result of hydrology modeling.

Storm hydrographs (Figure 3), land cover data, environmental data, and elevation data were input into a hydraulic model to estimate inundation from stream channel overflow. Figure 4 displays the projected changes in floodplain areas for the location, between the baseline and future time frames under two different RCP scenarios.

Figure 4 displays the projected changes in floodplain area due to rainfall events for each target year and climate scenario. The percentage inundated area provides the significance of changes and projected level of impacts. Based on Figure 4, flooding is projected to inundate up to 23.3% of the area within the 2030 and 2050 time frames under worst case climate change conditions.

The precipitation mostly decreases compared to baseline in the projected climate scenarios. Even with a decrease in projected precipitation (Table 1), there is a projected increase in inundations between the scenarios, as shown in Figure 4. Inundation projections were influenced by two variable inputs: (1) variation in total precipitation between design storms and (2) variation between the daily distributions of precipitation over the 3-day period. Even though the total precipitation for the 3-day storm is decreasing, more precipitation in one day can cause more run off due to saturation conditions. Additionally, variability in the projected changes in land cover can impact the projected inundation. Projected changes in land cover, even upstream of the area of interest, have a huge impact on the flood plain area analysis and the corresponding results, since it impacts the run-off rates (hydrologic modeling) and the roughness co-efficient (hydraulic modeling). For example, for one of the sites, the landcover is projected to change from sub-tropical shrubland to sub-tropical grassland under the RCP 4.5 scenario and to temperate cool mixed woodland under the RCP 8.5 scenario. Within the hydrologic model, projected land cover type intersected with soils and depth to water table dictates the friction, infiltration rate, and run off rate.

Based on the projected floodplain area inundation, the potential impacts on built infrastructure and assets were assessed to identify various vulnerabilities. The CIP data are overlaid on the projected inundation area due to flooding to identify vulnerable assets. Assets within the projected inundation area are identified to be highly vulnerable and assets within 38 feet from the inundation boundary are identified to be moderately vulnerable.

Tables 2 and 3 lists the various assets identified to be vulnerable due to flooding with the characteristics as shown in the footnotes to Table 2.

Based on the identified vulnerabilities, Table 4 displays an example summary of adaptation strategies.

A short description of the technology details for the adaptation strategies is provided for the decision-makers, along with an infographic for better understanding.

This research overcomes several existing limitations in the world of civil planning within the realm of changing climate. While experts are currently focused on isolated issues perfecting the science of climate projections, flood modeling or regional planning individually, this research provides a comprehensive framework and methodology for utilizing climate projections for flood modeling in an effective way to promoted informed planning for decision-makers. In addition, the project outcomes establish protocols for translating changing climate into actual flood risk and associated vulnerabilities and also provides potential adaptation strategies when planning for mitigation and/or resilience.

The authors would like to thank the following researchers for their technical contributions supporting this work: Dr Liz Caldwell, Melinda Clarke, Dr Dennis Ojima, Robert Flynn, Tom Hilinski, Lucas Chop, Dr Christopher Thornton, Meredith Miller, Xury Deputy, Ryan Davis, and Vatsal Gupta.