Hydrologic Data

The inundation study includes data from 25 real-time tide gages with most having continuous records beginning in the late 1990s or early 2000s. Tidal water-surface elevation at each gage is averaged and recorded every six minutes, then transmitted hourly by a satellite transmitter inside the tide gage equipment shelter and made available on the internet through the USGS National Water Information System (NWIS; U.S. Geological Survey, 2022b). Elevation data from these tide gages are referenced to NAVD 88. Annual maximum tidal peaks, as well as the daily high-high and low-low tidal statistics are published at each gage in addition to the instantaneous data that are available on NWIS.

Hydrodynamic Model

The hydrodynamic model used for this study was a product of the FEMA Region II storm surge study that was initiated to produce updated FIRMs. The model is a loose coupling of the ADCIRC coastal circulation and storm surge model with the SWAN model. ADCIRC computes wind and wave driven storm surges within the region of interest and SWAN computes wind-driven wave fields (Toro and others, 2011). The combined models functioned simultaneously on an unstructured, triangular mesh grid with specific node points and is referred to as an ADCIRC-unSWAN model. Throughout the remainder of this report the combined ADCIRC and unSWAN model will be referred to simply as the ADCIRC model. The ADCIRC model simulation production runs use inputs consisting of meteorological wind and pressure files describing atmospheric conditions, model grid and static parameter files, command files for the program, and a tide “hot start” file used to begin the simulation (Risk Assessment, Mapping, and Planning Partners, 2014b). Once a simulation completes, there are several output time-series that are created. They include atmospheric pressure, wind stress or velocity, water elevation, and depth-averaged velocity for the simulation run at all nodes in the model grid from the coast and overland (Risk Assessment, Mapping, and Planning Partners, 2014b). The water elevation time-series was output on a 400-second timestep for the FEMA model used for this study and was the output used to generate the flood inundation maps described within this report. There are also two outputs generated at user designated specific recording stations for elevation times series and depth-averaged velocity time series.

To select meteorological storm characteristics for use as input for the model, the Joint-Probability Method (JPM) approach was used. In the application of the JPM, all possible combinations of storm attributes along with their probabilities are considered, followed by evaluation of the surge effects from those combinations, and calculation of probabilities associated with combinations that generate peak water-surface elevations greater than the value of interest at a location of interest (Toro and others, 2011). Five critical parameters including central pressure, radius of maximum winds, maximum wind speed, heading, and the Holland B (dimensionless parameter that controls the wind profile) are considered in the application of the JPM. Further reading on the JPM approach can be found in “Guidance for Flood Risk Analysis and Mapping: Statistical Simulation Methods” (Federal Emergency Management Agency, 2016). Results from the JPM approach were used to define and input storm impacts for the flood-inundation map scenarios, as described below.

Storm Selection

For the FEMA Region II Coastal Storm Surge Study, 159 synthetic tropical storm and 60 extratropical storm production runs were generated in ADCIRC. To generate informative flood-inundation maps that are useful to both emergency management personnel and the public, the USGS narrowed the storm scenarios down for display. Many of the simulated storms had landfall locations on the eastern end of Long Island, New York, and were not considered useful for this study (table 2). The remaining storms were ranked based on greatest potential risk for moderate to severe flooding for New Jersey coastal communities. Six storms were selected among the 25 gages where inundation maps were generated. A summary of the characteristics associated with the selected storms are listed in table 3.

Storms were evaluated by examining the maximum elevation for all nodes in the model grid, known as the MAXELE file in ADCIRC. The MAXELE represents the maximum modeled water-surface elevation across all simulations at each mesh node. These nodes and associated maximum elevations were exported from ADCIRC and displayed on a map of the study area to evaluate spatial distribution of the MAXELE nodes along the New Jersey coastline (fig. 5).

To represent scenarios with the highest risk of moderate to severe flooding, storms with the highest density coverage of MAXELE nodes were selected as the primary storm for each county. Maximum wind speed and landfall location (and, by association, wind direction) proved most relevant when evaluating the intensity of potential flooding in coastal communities and the associated MAXELE node coverages. The primary storm selected at each tide gage was named STORM A and represented a storm track that closely followed the mid-Atlantic coastline and usually made landfall near the tide gage or adjacent community. For tide gages in the northern counties (Monmouth, Middlesex, Essex, and Bergen), STORM A represented a grazing sideswipe. These were generally selected to produce simulated flood inundation scenarios that included dynamic storm surge characteristics of an overwater to overland storm track transition.

A secondary scenario called STORM B was selected to simulate flood inundation scenarios associated with a storm remaining further offshore and then making landfall in either Cape May or Atlantic counties. STORM B scenarios would be expected to produce different inundation effects based primarily on wind direction difference from the alternative storm track as well as from the final landfall approach and location. Two coastal flood inundation scenarios were presented for each tide gage to illustrate the variability in storm surge flooding from distinctions in storm characteristics and local geography. It was decided, presenting more than two flood inundation scenarios could add confusion and limit use and understanding of these early warning and mitigation tools. In addition, the study did not combine flood inundation scenarios from multiple storms to preserve the unique differences generated by tidal and meteorological model forcing parameters as an attempt to illustrate the non-uniform and dynamic nature of many coastal storm surge events.

Tide Gage Flood Inundation Extents and Areas of Uncertainty

Given the complex nature of tidal hydrodynamics near the inlets and back bay systems of New Jersey, the distance from a tide gage where coastal flood inundation can be reliably predicted is variable and generally unknown in the absence of specific, fine-resolution modeling. Traditional riverine flood inundation mapping is less complex because typical streamflow in rivers and streams is generally controlled by the shape, slope, and channel roughness. Flood inundation mapping along river reaches also benefits from calibration to stream flows that have been measured by direct or indirect methods and used to calibrate the associated step-backwater or other type of hydraulic model. The flood inundation maps produced in this study required developing an understanding between the behavior of the modeled water-surface elevation at the existing USGS tide gage and other locations within the potential extents of mapping for each tide gage.

The flood-inundation extents for each tide gage define the boundary for displaying the flood-inundation maps. Several procedures were employed to generate and tune the extents. Preliminary map extents were created in GIS by clipping ADCIRC node points and New Jersey Parcels (New Jersey Office of Geographic Information Systems, 2018) to each county. Polygons were aggregated at a 200-meter (m) scale then manually edited to remove holes in the polygon whose minimum bounding rectangle had a long axis smaller than 250 feet (ft). The remaining polygons were dissolved into one polygon per tide gage as the initial attempt at delineating mapping extents. In this study, flood inundation extents define the boundary determined for displaying inundation maps for each tide gage that would provide the recorded or reference tidal water-surface elevation data. Initial flood inundation extents were manually evaluated to remove areas that were hydraulicly disconnected, or otherwise outside of tidally affected areas. At other locations the potential map extents were reduced or eliminated because of large distances from the tide gage or closest ADCIRC node points. It was unclear if the water-surface elevations within these areas would respond consistently or remain hydraulicly connected to the forces affecting the water-surface elevation at the tide gage, so it was determined the most appropriate action was to remove those areas from the mapped extents. Inland areas were manually trimmed at major roadways, where applicable, or again at areas that stretched far beyond available model node points. Defining the flooding extents based on nearby USGS tide gages, allows for measured tidal water-surface elevations at the tide gage to be compared to simulated inundation water-surfaces at the tide gage, and the corresponding simulated flooding extent, for forecasting and visualizing inundation in communities.

A warning was created for some tide gage map extents to alert users if the relation between the near tide gage hydrodynamics and an area within the mapped extents may be less accurate than others. In these cases, a flag for areas of increased uncertainty was developed and is displayed on maps in the FDST viewer. Where appropriate, the flags mark locations where there is a lack of node points, DEM data may appear unreliable, or existing physical features such as impoundments in marsh areas may add unknown uncertainty to the modeled output. In some circumstances limited areas of inundation within a station extent were noted to respond inconsistently between increasing levels of inundation, possibly owing to temporal differences in the modeled water surfaces as the simulated storm moves through the area. Examples of this would include a limited area of the inundation grids showing significant inundation at the 4-foot level, but then “receding” when the 5- and 5.5-foot level was overlaid, followed by once again being inundated in the 6-foot and higher inundation levels. These areas were also designated with the flag for increased uncertainty. Gages that utilized these flags for areas of increased uncertainty included Maurice River at Bivalve, Sluice Creek at South Dennis, Cape May Harbor at Cape May, Ludlum Thorofare at Sea Isle City, Great Egg Harbor Bay at Ocean City, Absecon Creek at Absecon, East Thorofare at Ship Bottom, Barnegat Bay at Route 37 Bridge near Bay Shore, Barnegat Bay at Mantoloking, Watson Creek at Manasquan, Shrewsbury River at Sea Bright, Raritan Bay at Keansburg, Newark Bay at Passaic Valley Sewerage Commission (PVSC) at Newark, and Hackensack River at Hackensack, N.J.

Topographic and Bathymetry Data

To ensure proper representation of overland flow, a topobathymetric digital elevation model (TBDEM) from the USGS Coastal National Elevation Database (CoNED) Applications Project was utilized to generate inundation layers and depth grids (Danielson and others, 2018). CoNED TBDEMs are constructed from multisource topographic and bathymetric data that are aligned vertically and horizontally to a common reference system (Danielson and others, 2018). This hydro-enforced TBDEM represented most bridges over navigable waterways as having their bridge decks removed, thus reflecting an elevation representative of the bathymetry of the associated waterway. These bridges are displayed as inundated at all levels of water-surface elevation. Some smaller bridges and culverts were not “removed” from the hydro-enforced TBDEM and will display the actual elevation of the bridge deck relative to the water surface until the bridge deck physically becomes inundated by the rising water-surface elevation.

Data Output and Adjustment

Hydrodynamic data from the FEMA Region II Coastal Storm Surge Study were provided to the USGS in the form of ADCIRC model output files which were accessed and queried through the Surface-water Modeling System (SMS). This study accepted the model calibration and verification performed by FEMA Region II and its contractors and utilized the final output for map development. Model simulations for the Region II study were run for 72 hours with storm forcing increased or “ramped” during the first 12 hours. This project primarily focused on generating maps of flood inundation resulting from various selected storm scenarios, so data output from the model ramp-up period, when initial forcings were not directly impacting the New Jersey coastline, were excluded from processing for this study. The simulation data used to generate inundation maps were extracted for all New Jersey node points approximately 24 hours before landfall through the end of the simulation for an average of 48 hours. Of the large sample of node points in the FEMA Region II study ADCIRC and unSWAN mesh file, only those that fell within a buffered polygon shapefile of New Jersey were extracted. Simulation data were output from SMS as text and formatted for input into ArcGIS and Python geoprocessing scripts.

Model simulation data were output in meters, vertically referenced to the mean sea level (MSL) tidal datum, so a transformation to the USGS standard of NAVD 88 in feet was required. Where possible, VDatum was used to vertically transform the data to NAVD 88 (fig. 6). For some gages, there were areas where transformation data for a reference tidal datum did not exist, mainly farther away from the coast and in tidal back bays or marshy areas. In these instances, a conversion factor was generated based on all points in the tide gage station extent where transformation was possible and used to approximate a transformation at points outside the area of VDatum. The conversion factor was calculated using equation 1, where the average conversion factor, c_avg, computed for the points within the map extent of each tide gage where VDatum was applicable, was added to the MSL water-surface elevation.

For gages where use of c_avg was necessary, the value of c_avg was compared to the nearest NOAA tide gage with an existing MSL to NAVD 88 transformation to ensure the calculated value was reasonable. An example is shown in figure 7, using gage 01396060 Arthur Kill at Perth Amboy, N.J., and a single timestep in storm NJB0107, where only 43 percent of node points in the mapping extent with non-null values could be transformed by VDatum and yielded a c_avg of 0.092 ft (standard deviation 0.04 ft). NOAA gage 8531545 Keyport, Raritan Bay N.J., approximately 5 miles southeast, has a published vertical transformation value of 0.10 ft (National Oceanic and Atmospheric Administration, 2022b).

Development of Flood-Inundation Maps

Once the hydrodynamic model output was converted to NAVD 88, time step, node latitude and longitude, and water-surface elevation were converted to individual point coverages. For each point coverage, a subset of points was selected based on the relevant tide gage station extent and a water-surface elevation raster dataset was interpolated with a 3-meter cell size using Esri’s Topo to Raster tool (Esri, 2021). The tool was run with the datatype parameter set to “Spot” optimizing the method for point data input instead of contour data input. This selection alters the default values for the roughness penalty and tolerance parameters, but otherwise the tool was run with default parameter values. After generation, each interpolated water-surface raster dataset was sampled at the grid point nearest the tide gage locations to generate a simulated hydrograph for the selected storm at the tide gage in feet NAVD 88 (fig. 8).

Timesteps in the hydrograph were matched with a list of target tide gage readings in half-foot increments ranging from 1.5 ft to the 1-percent annual exceedance probability elevation. The hydrographs were also aligned with the NWS flood stages at the tide gages. Only timesteps preceding the maximum water-surface elevation value were evaluated and labeled as follows:

Organizing the timesteps into these three categories allowed for more efficient processing of the data. Matches labeled “Exist” were ready for generating depth grids and needed no additional processing, but those labeled “Inter” and “Bathtub” required additional processing to generate the missing water-surface datasets.

For matches labeled “Inter,” a new water-surface raster dataset was created using Esri’s Cell Statistics tool. Each cell value in the new raster was the maximum between cell values in the water-surface raster datasets with tide gage readings just below and just above the target value. The newly generated raster dataset represented a “worst case” of flood inundation based on the available simulation data. Some storm scenarios showed a rapid rise in water-surface elevation exceeding 0.5 ft per timestep at a tide gage which resulted in the creation of duplicate water-surface raster datasets. Additionally, duplicates arose from NWS flood stages being close to the 0.5 ft target, for example 4.4 ft versus 4.5 ft. Duplicates were removed from further processing.

For matches labeled “Bathtub,” a new water-surface raster dataset was created using Esri’s Raster Calculator. The difference between the target value and the maximum recorded value in the hydrograph was added to each cell in the water-surface raster dataset corresponding to the maximum value. This raster represented a “hybrid bathtub” surface that artificially raised the maximum storm-forced surface to the target tide gage reading. The term “bathtub” is often used when generating a water-surface elevation that represents a single elevation and is essentially flat like water in a bathtub. In this study the term “hybrid bathtub” was selected to describe the method of offsetting each node in the water-surface elevation raster by a fixed value. The resulting water surface is not flat like a typical bathtub surface. It matched the shape and contour of the modeled timestep surface but with a fixed offset (bathtub value) added to each raster value. “Bathtub” layers were generated at whole foot increments only.

All the raster datasets, regardless of generation method, were converted into depth grids by subtracting the TBDEM in feet NAVD 88 from the water-surface elevation raster datasets using Esri’s Minus tool. A reclassified raster dataset was created for each depth grid where water depths less than 0.1 ft were changed to 0 and removed, and values greater than 0.1 ft were changed to 1 and converted to a polygon for review and quality assurance.

Quality Assurance

Manual quality assurance of each inundated area at all tide gages was required to assess hydrologic connectivity to ocean or back bays and evaluate the topographic complexities present in coastal hydrologic systems. Inundated areas that were detached from the coastal waters were examined to identify connections with the ocean and tidal tributaries, such as over dunes, or through culverts under roadways. If the connections were deemed reasonable through review of the aerial orthoimagery, the disconnected areas of inundation were retained. Some of these were very easy to identify as the orthoimagery displayed open marsh type land cover that reflected frequent tidal inundation. Other areas were less clear, as was the case for several impoundments found within the station extents in the Monmouth County tide gages. For example, Deal Lake controls the level of the impoundment using a hydraulic flume structure that drains to the ocean. Specific elevations for the hydraulic structure were not obtained. Within the inundation layers, it was decided to retain the lake as “inundated” once the generated polygon displayed it as inundated despite it being slightly disconnected from the main inundation grids, as the possibility of inundation from the ocean through the flume structure exists. Disconnected polygons that are adjacent to another inundation map extent were compared with the adjacent inundation layers of similar gage heights to determine possible connectivity. If inundation was identified at a similar gage height on the adjacent map extent, the apparent “disconnected” polygon was retained. All disconnected areas of the inundation polygons determined to be erroneously delineated were removed.

After manual review and editing of the flood-inundation polygons, they were arranged in a map library designated by gage number and storm scenario identifier and run through a final processing routine to enhance the display of the final layers. The routine includes two geoprocessing steps using GIS python tools. First, each polygon is run through a tool with the default parameters set to fill small gaps inside larger polygons. Most of the gaps are artifacts from converting the raster dataset into a polygon. Second, the arranged polygons are run sequentially through a tool that calculates overlap between the inundation polygons. Taking a “lower” and a “higher” input, the tool creates an overlap polygon only where inundation from the lower tide gage reading extends beyond that of the higher tide gage reading. If an overlap exists, it is merged with the higher gage reading polygon, which ultimately becomes the next lower gage reading in the sequence. In this manner, the final polygons only show an expanding inundated area as gage readings increase and favors the “worst case” interpretation of the simulated storm data. These final polygons are then used to extract final depth grids from their respective progenitor raster datasets, with overlap polygon area extracting from the next lower raster dataset, and any “NODATA” or negative values replaced with 0.1 ft.

Evaluating Model Accuracy and Applicability

Four of the historical extratropical storm simulations modeled for FEMA's Coastal Storm Surge Study included time periods when modeled data overlapped with the period-of-record at the USGS tide gages. These extratropical simulations included nor'easters from October 25, 2005, April 16, 2007, May 12, 2008, and November 13, 2009. The latest simulation from November 13, 2009, was selected to compare against historical recorded tide data at several tide gages throughout the study area. An example in figure 9 shows modeled versus historical tide elevation data at USGS station 01411390 Cape May Harbor at Cape May, New Jersey. The modeled data showed generally consistent patterns with historical data throughout the tidal signature leading up to the event with peaks and valleys aligned (fig. 9). Some differences would be expected as the nature of the model mesh resolution results in some deficiency in capturing complexities within harbors and narrow channels (Risk Assessment, Mapping, and Planning Partners, 2014c). Specific differences between modeled and historical amplitudes were not calculated as part of this comparison. The simulated peak for the event matched the historical peak with a reasonable difference of approximately 1 ft. Differences less than 2 ft generally indicated good agreement within a study area (Risk Assessment, Mapping, and Planning Partners, 2014c).

Generated inundation extents were overlaid with FEMA 1-percent annual flood hazard and 0.2 percent annual flood hazard zones in figure 10 to ensure the modeled inundation layers were not grossly underpredicting or overpredicting inundation extent within communities. To accomplish this, the finalized flood inundation polygon at a gage elevation closely representing the published Base Flood Elevation (BFE) at the gage location was selected for comparison. An example of this process using USGS tide gage 01411390 Cape May Harbor at Cape May, N.J., is provided below.

Referencing the most recent FIRM panel 34009C0284F for the City of Cape May, the gage location is compared to adjacent BFEs. The BFE for the actual gage location (on a pier) places it within a VE zone, which sets the BFE to 11 feet NAVD 88. FEMA V and VE Zones are areas along coasts within the FEMA-designated Special Flood Hazard Area subject to additional hazards associated with storm-induced waves (Federal Emergency Management Agency, 2021a). Although the USGS tide gage is within the VE Zone, the tidal water-surface elevations recorded by the gage use a combination of physical means, such as baffling disks in the gage well, as well as averaging algorithms applied to the raw unbaffled recorded data to filter the effects of wave action so the resulting data are more representative of still-water elevations. In addition, the model output used to generate inundation layers for this project does not include waves modeled on top of the still-water tidal elevations. In general, comparison to V and VE Zones was not considered for this project, and thus, the adjacent AE Zone or coastal AE Zone elevation was selected instead of the VE Zone BFE where applicable for comparison to the modeled inundation layers. The AE Zone is the base floodplain within the Special Flood Hazard Area where BFEs are available. In the described example for Cape May, an elevation of 8 feet NAVD 88 was used from the adjacent AE Zone. Using this information, the 8-ft inundation layer for the gage at Cape May was selected for comparison to the FEMA flood hazard zone extents.

Comparison of the FEMA zones versus the modeled inundation layers was a qualitative process focused on ensuring general consistency and identifying anomalies. Anomalies could present as regions within a station extent that show inundation extending well beyond the 0.2-percent flood hazard zone, or areas where inundation does not fill in a large portion of the 1-percent flood hazard zone. Anomalies such as these would suggest enhanced uncertainty within the modeled layers requiring delineation on the finalized map libraries and associated online mapper. Generally, the modeled inundation layers were consistent with the boundaries defined by the FEMA 1- and 0.2-percent annual flood hazard layers.

It should be noted that not all areas of the coast reflect updated, digitized flood hazard maps. Direct comparison in these areas is more challenging and, therefore, was not performed. Examples of communities without updated digitized flood hazard maps include Ventnor City and Atlantic City, N.J. More information on FEMA effective study dates can be found at the FEMA Flood Map Service Center (Federal Emergency Management Agency, 2021b).

Evaluating Community Impacts at Moderate Flood Levels

Moderate flooding, or what has been referred to as “moderate flood level” or “moderate flood stage” has generally been defined at the local governmental level and influenced by officials in each community to help identify when widespread flooding of roadways and structures begins, transportation routes become impassable, and risk to life and property starts to become a major concern. Understanding how communities are impacted by inundation from coastal flood waters associated with moderate flooding is important information necessary for local officials tasked with protection and resiliency. In the following sections the variability of impacts resulting from simulated flood levels are described and compared.

Understanding Community Size and Density

An analysis was done using simulated inundation maps referenced to water-surface elevations at the selected USGS tide gages used in this study. The simulated coastal flooding layer closest in elevation to the selected NWS flood stages (moderate or major) was evaluated to estimate the community impacts, where available. It is difficult to estimate or predict the presence or movement of people in each community so the total number of buildings (referred to in FEMA Hazus analysis as total building stock) was used as a surrogate for the general percent of people exposed to inundation. To quantify the degree of community impact, the analysis tried to identify buildings with a high likelihood of inundation and major roads and bridges that potentially experienced disruption of access due to inundation from coastal flood waters. At moderate flood stage, disruption of access to major roads and bridges was considered a high impact because it was assumed to disrupt connectivity between barrier islands and the mainland. This infrastructure is critical for moving people and assets out of potentially dangerous coastal flood areas. If access to these transportation routes is compromised at the moderate flood stage, it is assumed evacuation during major flooding events would be even more difficult and dangerous. The total area and depth of inundation for communities near USGS tide gages were also evaluated as characteristics to help quantify the community impact. Comparing the change in magnitude of total buildings and major roads potentially inundated at moderate to major flooding provides important planning, mitigation, and response information to help community, state, and federal managers increase resiliency and reduce the loss of life and property.

The community impact analysis used the extent and depth of inundation estimated from two selected synthetic storms, described earlier in this report, for the mapped extents associated with each USGS tide gage and compared them with building and major road information to evaluate flooding impacts. The area of mapped extents for each tide gage is referred to as station extents and are generally not consistent in size because of the natural irregular shape, geography, and land use properties of the New Jersey coastline. The use of percentages in categories like “total building stock inundated” are presented to describe the degree of impact because the use of totals or totals by category would be more difficult to compare. For example, figure 11 shows the total number of buildings for each station extent and the number of buildings greater than 2,000 square feet in the same extent to illustrate the variability in inundation between station extents. An initial comparison of total area, percent of area classified as urban, total number of buildings (building stock), and building size greater than 2,000 square feet, was done to develop a baseline understanding of community variability before evaluating area inundated. The total area and area classified as urban for each station extent are shown in figure 12.

The station extents for the USGS Mantoloking (01408168) and Seaside Heights (01408748) tide gages have the largest number of building structures per mapped area in the study with more than 20,000 structures per station extents. These areas include communities such as Mantoloking, Toms River, Lavallette, and Seaside Heights. Although these station extents have the largest number of structures, neither of these areas are in the top three largest areas in the study. Additional analysis of building structure data shows neither area associated with the Mantoloking and Bay Shore tide gages have the highest density of building structures in the study, but collectively they all rank in the top half of all station extents. Understanding the variability of building density among station extents may provide additional information to better assess impacts to people and property when compared to simply looking at total area or building count in a station extent. Figure 13 shows the distribution of building density among mapped station extents. As indicated earlier, the Mantoloking and Bay Shore station extents had the highest total building counts, but when analyzed for building density Mantoloking had the second highest building density and Bay Shore was not in the top 5 highest densities. Mantoloking station extent also was not in the top 5 largest areas or top 5 highest percentages of urbanized areas. Further investigation into the community characteristics highlighted, for example, that evaluating total building count and building density for mapped station extents helps show a potentially greater community exposure and impact if inundated by one of the storm scenarios when compared to some areas that may have a larger area of inundation.

Understanding Moderate Flooding on Buildings and Major Roads

The extents and depth of moderate flooding for the community impact analysis evaluated the impacts from scenarios STORM A and B described in this study. Understanding the count and density of homes in the mapped station extents associated with each tide gage provided better information to help evaluate the potential community impacts from coastal flooding than simply relying on the size of the area. Adding information about first flood elevations and estimated depth of flooding at each structure and along major roads allowed for a more advanced understanding of the potential severity of impacts to the residence of the community and their ability to evacuate or reach safe shelter.

Major roads were defined as roads in classes 1 through 6 in the “route_subtype” field within the New Jersey Department of Transportation (NJDOT) Roadway Network geospatial data (New Jersey Department of Transportation, 2021). These roads typically included interstates down to major county roads. The NJDEP building footprint dataset (New Jersey Department of Environmental Protection, 2019) was used to define first-floor elevations, lowest adjacent grade (LAG), and highest adjacent grade, and if the structure type was classified as critical (CS). A building’s structure was considered inundated at moderate flooding if the water-surface elevation was computed to be greater than the LAG elevation using the selected inundation layer for moderate flood stage. The water-surface elevation at each structure was determined by selecting the cell within the structure footprint with the maximum elevation.

At bridge crossings, a slightly modified approach was needed. At some bridges along major roads or over causeways, inlets, and back bay water bodies, the bridge structure can be arched or have a higher deck and low steel elevation near the center of the span when compared to the elevation at the abutments. In other situations, the bridge deck may remain above flood levels regardless of an arch or curvature to the bridge deck, but the approaching roadway becomes inundated cutting off safe access to the bridge. This potentially more complex situation was solved by combining the approaching roadway and the bridge into a single feature so they could be analyzed together. If either the bridge becomes inundated or any part of the approaching access roadway, the bridge is determined to be inundated and added to the list of impacted structures. As previously mentioned, the hydro-enforced TBDEM used for this study had many bridge decks already removed from the elevation model. For the analysis portion of this study these bridges were represented by an approach feature, as otherwise, they would always calculate as “inundated.”

Evaluating the impacts of moderate flooding required selecting coastal flood inundation layers closest to the NWS moderate flood level elevation, converted from mean lower low water (MLLW) to NAVD 88, and intersecting them with the buildings and road datasets to calculate what structures were identified as inundated. As described earlier, relying solely on total counts of structures inundated makes comparisons between communities or mapped station extents more difficult. The impact in some communities is more straightforward because they are established at a generally lower elevation, so many of the total structures are inundated at moderate flooding. In these areas, elevation, early warning, and evacuation are typically part of the strategy. In other areas, a smaller number of total structures may be inundated. However, if strategic areas and major access roads are inundated, a larger percentage of the community can be impacted. Without that vital infrastructure, the community’s ability to evacuate or access a bridge to the mainland will likely be cut off, even if their building structure or home was not inundated. One example of this is along the northern end of the Mantoloking station extents. Although the total percentage of the station extents inundated is not the highest among the mapped tide gage station extents, it does have the highest total number of buildings. As a tidal storm surge reaches moderate flooding, many buildings from Mantoloking North to Point Pleasant Beach may be cut off from accessing Route 35, Ocean Avenue, Ocean Road (Route 88), and other roads that can be used by those that would be otherwise trapped between back bay inundation and the barrier island dunes. This example is illustrated in figure 14. To better understand the magnitude of potential damage, this study looked beyond the initial analysis of building structures simply being inundated during moderate flooding by computing the depth of flooding at each structure and dividing the results into two categories: (1) the percent of structures inundated with 1 foot or more of flood water above the first-floor elevation; and (2) those inundated with 3 ft or more of flood water above the first-floor elevation at moderate flood stage. The percent of structures identified as inundated with greater than 1 ft and greater than 3 ft of flood waters during the moderate flooding scenarios for each tide gage are illustrated in figure 15.

The evaluation of total building counts, percentage inundated, and depth of inundation for structures at moderate flood stage provides data to begin evaluating the combined impacts of coastal flooding on communities within the mapped station extents. Analyzing the compound effect of inundation to building structures, disrupted access to major roads, and damage or impact escalation from forecast error presents a comprehensive outlook of overall coastal community flood impacts. The next section of this report provides more information about this compound analysis.

Understanding Community Risk Variability from Moderate Coastal Flooding

The coastal communities covered by these new coastal flood inundation maps extend from Hackensack, Perth Amboy, and Keansburg along the northern coast to Atlantic City, Stone Harbor, and Cape May at the southern end of the New Jersey coastline. Moderate flooding has generally been defined as when widespread flooding of roadways and structures begins, transportation routes become impassable, and risk to life and property starts to become a major concern. Utilizing both coastal storm scenarios used in this study, the range of building structures impacted at moderate flood stage for each station extent ranged from more than 14,000 structures to less than 100. Figure 16 depicts some of the differences visualized with structure inundation at a moderate flood stage of 3 ft between the gages at Mantoloking and Bayshore. A structure was considered inundated if one cell from the moderate flood stage layer intersected the building footprint with an elevation above the LAG for that structure. The average percentage of structures estimated to be inundated at moderate flood stage was slightly more than 5 percent when using an average of the two storm scenarios. Several communities were estimated as having more than 20 to nearly 50 percent of the buildings inundated using the same averaged criteria while two communities stand out with estimates of more than 80 percent of the buildings inundated using the same two storm average criteria (fig. 15). The percent of these structures estimated to be inundated by 1 ft or more above the first-floor elevation ranged from nearly zero to almost 20 percent in the communities of Sea Isle City and Ship Bottom.

Along with building structures, the percentage of major roads and bridges in the station extents for each tide gage were evaluated. At moderate flood stage the mean percentage of major roads affected by flooding was almost 7 percent. A few station extents indicated more than 15 percent of major roadways affected by flooding and one community exceeded 30 percent of major roadways affected. When the flood stage was increased to NWS major flood stage, the mean percentage of major roadways affected by station extents increased to slightly more than 11 percent (fig. 17A). The combined impact from flooding on building structures and major roads presents a comprehensive dataset to help evaluate the impact on coastal communities from coastal storms.

It is important to recognize evaluating impacts at moderate flood stage should include an understanding of potential escalation of impacts if the forecast or peak storm surge exceed moderate flood stage. This concept allows for anticipating uncertainty related to a storm forecast with the potential of disproportionate increases in damages beyond the threshold of a flood category like moderate flooding. For example, if the storm surge scenarios estimate a rapid expansion of damages or impacts beyond the limits, or edge, of moderate flooding it could be suggested that the range of moderate flooding is too broad. Emergency managers, responders, and the greater communities in general may benefit from a slightly reduced range to allow for a more proportional increase in damages. Analyzing the change in potential impacts between moderate and major flooding to evaluate if the edge of the thresholds allows for a somewhat gradual increase in damages as flood levels transition from moderate to major flooding was defined as an “edge analysis” for this study. The difference in elevation between the NWS moderate and major flood stages at most tide gages used in this study was generally about 1 ft. This was considered an acceptable increase for an edge analysis for these coastal areas due to the inherent uncertainty of hurricane and tropical storm force winds and associated storm surge. The mean increase in percent of building structures inundated per station extents was almost 12 percent with a few station extents showing increases of more than 20 to 30 percent (fig. 17B). The median increase was almost 8 percent reflecting that only a handful of communities within the station extents of the Keansburg, Barnegat Light, Ocean City, and Stone Harbor tide gages change substantially between moderate and major flooding. Notably, the communities of Sea Isle City and Ship Bottom demonstrate amongst the lowest percent increase, likely owing to the significant building inundation that is already occurring at moderate flooding. The mean increase in percent of buildings with 1 ft or more of inundation above the first-floor elevation was only 1 percent which may indicate as coastal flooding increased, the inundation spread outwards.

Many factors combine to influence general perception of coastal flood risk for different communities. Major hurricanes or coastal storms sometimes provide a perception of risk for several years, but it may diminish over time. The community understanding of coastal flood risk can also be influenced by several factors related to the most recent major storm track, distance of wind field from the community, and overall size and strength of the storm. To help strengthen resiliency and improve early warning impact mitigation plans, the combined vulnerability of building structures and transportation routes (major roads and bridges) to inundation from the simulated moderate flood scenarios was analyzed. Gaining a better understanding of the variability of flood impacts among coastal communities of New Jersey during moderate flooding (when it is perceived that significant impacts begin to occur) could provide valuable information about which communities may be at greater risk relative to the average of all communities in the study. Furthermore, use of a few common characteristics to experiment with methods to assign a mathematically computed score or moderate flood risk value for each mapped area was assumed. Using an equation to evaluate the variability of moderate flood risk helps reduce the subjectiveness of identifying one area as more vulnerable than another, and it provides a common means of calculating an estimated community risk.

The first method tested utilizes a few characteristics for building structures and transportation including percent of major roads inundated and number of major bridge crossings with disrupted access due to flooding. The resulting value was defined as the Coastal Community Moderate Flood Risk (CCMFR) score. Communities scoring a higher CCMFR score would be considered more vulnerable to coastal flood inundation. The primary categories tested were total buildings in the moderate flood area, percent of building structures with any level of inundation, building structures with greater than 1 ft of flooding above first-floor, building structures with greater than 3 ft of flooding above first-floor, percent of major roads inundated, and percentage of major bridges inundated or with inundated approaches rendering the bridge unusable. Using these categories, a weighting factor (defined as the ratio of the percent of community impact to the average of all communities) was applied to each category and the resulting value was computed for each mapped extents of moderate flooding.

The second method is similar to the first but uses two sets of fixed weighting factors for each characteristic. The selection of which weighting factors are used is done with a distribution factor (df). The z-score, also referred to as the standard score in statistics, was computed for total building structures inundated at moderate flooding and used to define df. The z-score is the measure of how many standard deviations the value is above or below the standard deviation of the dataset. For these equations, df was computed for moderate flooding using the percent of total building structures inundated in the local station extents to the average percentage of buildings inundated for all mapped station extents in the study. The value of df was computed for each storm scenario at all station extents and the two-storm average was used in the application of CCMFR_weighted equation. Similar to the base CCMFR score, communities scoring a higher CCMFR_weighted score would be considered more vulnerable to coastal flood inundation. Once computed, a positive value of df greater than 1 is used to apply more weight to percent of total structures inundated and percent with 1 ft or more of inundation and less on those with greater than 3 ft and total roads inundated. As noted earlier, building structures exposed to inundation are being used as a surrogate for people being directly impacted by flooding. It was assumed a community with a percentage of buildings, and therefore people, inundated beyond one standard deviation from the average should have a higher weight applied to total volume of exposure and slightly less on depth greater than 3 ft at the inundated structures. The value of df is used to select between the following equations providing the ability to adjust weighting for communities with total percent of buildings inundated positively skewed by more than one standard deviation at moderate flooding.

If the distribution factor (df) is greater than 1,

If df is less than 1,

The resulting experimental methods produced similar results partially because they apply similar importance to building structure inundation. The CCMFR scores ranged from very low (nearly zero) to a high of just over 50. Although a ratio and weighting method were both tested and additional weighting was applied for communities with higher-than-average building inundation, the results were generally similar. A threshold of twice the average for all communities was identified as an initial cutoff point used to select mapping extents that may be at higher risk for impacts from moderate flooding. Overall, five station extents exceeded this threshold although one of the five was just over the cutoff limit. Considering the four station extents that clearly exceeded the threshold, two of the four were identified by both equations. The station extents for the tide gages at East Thorofare at Ship Bottom (USGS station no. 01409146) and Beach Thorofare at Margate (USGS station no. 01411330) exceeded twice the average score for both experimental equations. The resulting CCMFR scores for each tide gage mapped area are shown in figure 18. This analysis tried to focus on evaluating the variability of flood impacts among areas within the mapped extents for each tide gage to help identify outliers or areas where increased flood risk education or mitigation efforts would have the highest cost to benefit ratio. Reducing subjective determinations of compounded coastal storm surge flood risk is a result of using measurable characteristics and developing mathematical equations. Future studies could improve this analysis by reviewing documented damages, and applying more characteristics, demographics, better categorization of transportation routes, and critical facilities to test and validate usefulness of the CCMFR value.

Uncertainties and Limitations Regarding Use of Flood-Inundation Maps

Although the flood-inundation maps represent the boundaries of inundated areas with a distinct line, some uncertainty is associated with these maps. The flood boundaries shown were estimated based on interpolating measured water stage and simulated water-surface elevation derived from the output of a hydrodynamic model developed by FEMA Region II. Unique and somewhat unpredictable meteorological factors such as rapid changes in wind speed and direction associated with coastal storms and hurricanes can produce actual water-surface elevations and inundation boundaries that deviate from the inundation scenarios presented in this study. Additional uncertainties and limitations previously described, such as hydraulic connectivity and distance to the tide gage within an inundation map extent, may result in deviations from actual water-surface elevations during a coastal storm event. If this series of flood-inundation maps will be used in conjunction with NWS coastal forecasts, the user needs to be aware of additional uncertainties that may be inherent or factored into NWS forecast procedures (for more information, see http://water.weather.gov/ahps/pcpn_and_river_forecasting.pdf).