Comparison of Flood Inundation Modeling Frameworks within a Small Coastal Watershed during a Compound Flood Event

The flooding brought about by compound coastal flooding can be devastating. Before, during, and immediately 15 following these events, flood inundation maps, or Events Maps, can provide essential information to emergency management. However, there are a number of frameworks capable of estimating Event Maps during flood events. In this article, we evaluate three such Event Map frameworks in the context of Hurricane Harvey. Our analysis reveals that each of the three frameworks provide different inundation maps that differ in their level of accuracy. Each of the three Event Maps also produce different exposure and consequence estimates because of their physical differences. This investigation highlights the need for a 20 centralized means of vetting and adjudicating multiple Event Maps during compound flood events empowered by the ability to distribute Event Maps as geographic information system (GIS) services and coalesce Event Maps into a common operating picture. Furthermore, we provide evidence that the ability to produce multi-model estimates of Events Maps to create probabilistic Event Maps may provide a better product than the use of a lone Event Map.


Introduction
Each year, tropical storms devastate portions of the coastal United States. From 1980-2020, tropical storms accounted for $945.9 billion in damages with an average of $21.5 billion in damages per event (Fast Facts: Hurricane Costs, 2021). Tropical storms bring strong winds and heavy rainfall that are the primary drivers of compound flooding. Strong winds and high tide create storm surge, pushing coastal waters inland and inundating land that is typically dry. Inland, heavy rainfall leads to direct 35 runoff and saturation excess runoff from the land surface into inland waterbodies. The combination of inland runoff and storm surge creates compound coastal flooding. Recent studies highlight how the combination of inland drainage and coastal surge are important in properly estimating compound floods (Gori et al., 2020;Loveland et al., 2021).
In order to inform emergency managers and the public at-large, agencies such as the National Oceanic and Atmospheric Administration's (NOAA's) National Weather Service (NWS), the U. S. Army Corps of Engineers (USACE), the Federal 40 Emergency Management Agency (FEMA), and the U. S. Geological Survey (USGS) produce estimates of flood inundation for inland, coastal, and compound flood events. The Integrated Water Resources Science and Services (IWRSS) refers to such flood inundation maps as Event Maps (IWRSS, 2013;2015). Event Maps are help emergency managers communicate situational awareness, devise response plans, and inform decision makers (NWS, 2012;IWRSS, 2013;Maidment, 2017;Longenecker et al., 2020). However, data availability to create Event Maps can vary dramatically across the world and can 45 originate from a number of sources. The disparate origins of multiple Event Maps for an event can create unnecessary confusion and conflicted decision making for decision makers.
A number of frameworks and methodologies exist to create accurate Event Maps. For inland fluvial flooding, NOAA's National Water Center (NWC) co-developed and implemented the height above nearest drainage (HAND) inundation model that uses the Manning's equation to precompute inundation libraries to couple with hydrologic forecasts from the National 50 Water Model (NWM) (Liu et al., 2018;Viterbo et al., 2020). The HAND methodology requires a minimal amount of input data that are available over large geographic scales. Alternatively, USACE developed the AutoRoute model that functions in a similar manner to the NWC's HAND implementation, requiring minimal inputs, making it capable of producing flood inundation maps over continental-scale geographic extents (Follum 2013;Follum et al., 2016;Tavakoly et al., 2021). HAND and/or AutoRoute perform well as first order approximations of fluvial flooding (Afshari et al., 2018;55 Johnson et al., 2020). However, these low complexity models do possess less skill when compared to higher fidelity hydraulic models (Hocini et al., 2021). One of the more notable limitations of steady-state inland models such as HAND and AutoRoute is their limitations in coastal watersheds. HAND and AutoRoute are fluvial-only flood models and their Event Maps do not inherently contain the pluvial or coastal components of flooding. Further, coastal watersheds tend to have minimal topographic relief where one-dimensional (1D) models, such as HAND and AutoRoute, traditionally struggle to produce accurate flood 60 inundation maps. Low topographic relief tends to create backwater effects that AutoRoute cannot physically account for (Follum et al., 2016;. Further, where topographic relief is low HAND can be sensitive to errors in the underlying terrain https://doi.org/10.5194/nhess-2022-27 Preprint. Discussion started: 16 February 2022 c Author(s) 2022. CC BY 4.0 License. (Johnson et al., 2020). Thus, steady-state hydraulic models, such as HAND and AutoRoute, tend to have limited effectiveness in providing Event Maps during compound coastal floods in coastal watersheds.
For coastal flooding, NOAA's National Hurricane Center (NHC) produces Event Maps that estimate coastal flooding from 65 storm surge using the the Sea, Lake, and Overland Surges from Hurricanes (SLOSH) model (Jelesnianski et al., 1984;Experimental Potential Storm Surge Flooding Map, 2022 In response to the limitations of existing fluvial and coastal flood mapping frameworks, Wing et al. (2019) use the Fathom-US large-scale hydraulic modeling framework (Wing et al., 2017) to perform Event Map estimation for Hurricane Harvey.
The Wing et al. (2017) framework can account for coastal, fluvial, and pluvial flooding. Wing et al. (2019) compare the Fathom-US flood inundation results to the NWC HAND methodology. Wing et al. (2019) find that the Fathom-US framework is more accurate than the NWC HAND methodology for the Hurricane Harvey simulations due to better representation of the 75 complex physics that occur during compound coastal floods.
Beyond the large-scale modeling frameworks such as the NWC HAND or Fathom-US, there are local-scale compound flood models in data rich environments that can have higher spatiotemporal resolution and are capable of producing Event Maps that combine coastal, fluvial, and pluvial flooding. For example, the USACE Models, Mapping, and Consequences (MMC) Production Center will work with local USACE districts and divisions to create and distribute Event Maps during flood events 80 using existing Corps Water Management System (CWMS) model frameworks or develop new model frameworks on-the-fly (Winders et al., 2018). The simulation times of these frameworks can be a hindrance in their ability to produce a timely Event Map. However, these models can provide a benchmark for what is achievable with increased model fidelity and resolution. Further, we may be able to more effectively utilize these high fidelity simulations for Event Maps through surrogate modeling techniques (Bass and Bedient, 2018;Zahura et al., 2020;Contreras et al., 2020;Kyprioti et al., 2021), similar to how the NWC-85 HAND and Fathom-US utilize a precomputed riverine hydraulics in those implementations Wing et al., 2019).
This paper seeks to investigate if different modeling frameworks produce substantially different Event Maps during compound coastal flood events. We evaluate and quantify the differences by using a Hurricane Harvey case study where a recently developed local scale framework exists and compare this to the AutoRoute and Fathom-US frameworks. Hurricane Harvey is 90 a now infamous compound flood event brought about by a combination of wet antecedent conditions, heavy inland rainfall, and sustained high water levels at the coast (Valle-Levinson et al., 2020). Our comparison of the three frameworks centers on the physical differences in each Event Map and if those differences lead to differences in estimated exposure and consequences.
To our knowledge, this is the first evaluation of Event Maps produced with different flood map frameworks that seeks to evaluate differences in the Event Maps by examining both the physical differences in the Event Maps and the estimated 95 exposure and consequences from those Event Maps.

Methodology
To perform our comparison, we applied a recently developed unsteady hydrologic and hydraulic modeling in the Clear Creek watershed, south of Houston, Texas. Figure 1 demonstrates the location of the Clear Creek watershed that covers an area is roughly 698.91 km 2 . The region has a history of repeated flooding, including flooding during Hurricane Harvey, and is subject 100 to rapid development and urbanization (Brody et al., 2018).

Modeling Framework Configurations 105
We performed our analysis by creating maximum inundation extent Event Maps produced by three frameworks: the HEC-River Analysis System (HEC-RAS) framework, the AutoRoute framework, and Fathom-US framework. Figure 2 illustrates the inputs for each modeling framework. The proceeding section describes these frameworks in detail. We utilized only observed meteorological and coastal data to ensure that limitations in forecast skill are not present.    (Gesch et al., 2002(Gesch et al., , 2010 for the study area. The 2016 collection of the National Land Cover Dataset (NLCD, Yang et al., 2018) and literature-derived 135 roughness coefficients as described in Follum et al. (2017Follum et al. ( , 2020 provide estimates of surface roughness. Because the chosen DEM does not contain bathymetry, we implement the simple bathymetric estimation methodology within AutoRoute (Follum et al., 2020)

Evaluation Methods
We perform two layers of analysis in our assessment to ascertain key differences between each of the three Event Maps. We In Equation 1, is the number of HWMs that are within the flooded extent of each Event Map and N designates the number of HWMs.
Following the methodology outlined by Wing et al. (2021) we assess the estimated WSE from each framework by estimating 160 error and bias.
In Equation 2 and Equation 3, WSEmod designates the WSE at the inundated pixel nearest to each HWM location modeled by 165 each Event Map framework, and WSEobs designates the WSE observed at each HWM location.
The second analysis provides a comparison of exposure and consequence estimates from each Event Map. To perform our exposure and consequence analysis, we utilize the Go-consequences model and the National Structural Inventory (NSI) (USACE, 2021a;2021b;2021c). The NSI is a point based structural inventory describing structures throughout the United States. The NSI supports the assessment of consequences of structures resulting from natural and man-made disasters (USACE, 170 2021c). Go-consequences uses the NSI to compute building damage and population exposure from flooding. Go-consequences uses a water depth estimate at NSI point locations, and uses the default depth-damage functions used within the HEC-Flood Impact Analysis (HEC-FIA) software and assigned by the USACE Economic Guidance Memorandum 04-01 (USACE, 2003).
In this instance, our flood damage assessment does not adjust damages to account for brackish water damage (USACE, 2021b).
To visualize the resulting point damage and exposure estimates, we used the point damage locations and their associated dollar 175 damage and building population counts to construct kernel density maps in ArcGIS version 10.8 (Kernel Density, 2022). The kernel density plots can provide a 'hot-spot' analysis to compare to collected Federal Emergency Management Agency (FEMA) flood insurance claim locations (Arctur, 2021). Although total monetary damage to buildings and their contents is difficult to observe following a flood event. Flood insurance claims represent a fraction of the overall damage and may represent the only spatially explicit observations of monetary flood damage. Shao et al. (2017) summarize that Galveston and 180 Harris County, TX, have rates of flood insurance purchase that fall between 26-50%. Our study domain falls between Galveston and Harris County. Using these flood insurance purchase rates along with the total FEMA flood insurance claims from Harvey constitutes an approximate upper and lower bound of total monetary damage to buildings and contents.

Simulation Comparison
We first compare the results from the HEC-RAS, AutoRoute, and Fathom-US frameworks to observed HWMs by estimating 190 locational accuracy. HWMs designate locations where floodwater reaches a given location and leaves behind evidence of floodwater presence in the form of mud lines, seed lines, etc. (Koenig et al., 2016). USGS quantifies the uncertainty of the HWM WSE measurements they collect. In our study domain, USGS considers 53% of HWMs in the study area of poor quality, 34% of fair quality, and 13% of good quality. What these qualitative descriptors translate into quantitatively is an average of ± 9 centimeters of uncertainty in the study domain HWM WSEs. 195 Each models Event Map should contain each HWM. Table 1 is an assessment of locational accuracy for each model under the assumption that the maximum inundation extent should contain the HWM locations. Interestingly, we can see that the Fathom-US model is more accurate at capturing HWM locations within the inundation extent than both the AutoRoute and the HEC-RAS model. This result contradicts our assumption that the HEC-RAS model will be more accurate given the level terrain resolution and calibration/validation performed upon the model. 200 However, expressing model skill in terms of locational accuracy has limited viability, given that the model could inundate the 210 entire study area and achieve 100% accuracy. Indeed, comparing the results from HEC-RAS, AutoRoute, and Fathom-US maximum water surface elevations (WSEs) to observed HWM WSEs reveals a different outcome. Figure 4 illustrates scatterplots comparing each simulation's maximum WSE to HWM WSE observations and Table 2 summarizes the error and bias of each framework. The orange line in all Figure 4 plots is the desired 1:1 relationship between observation and model results and the hashed line is the line of best fit from a least squares regression analysis. In Figure 4 and Table 2, we see that 215 the HEC-RAS framework produces more precise and accurate WSE estimates than both the AutoRoute and Fathom-US frameworks with points tightly packed along the dashed regression line that align well with a 1:1 line and lower error value.
The HEC-RAS frameworks biases toward underestimation with bias of -0.28 MASL. The Fathom-US framework tends to overestimate WSE, with the regression line falling to the right of the 1:1 line and a positive bias of 0.60 MASL. The AutoRoute framework has a less consistent tendency than the HEC-RAS and Fathom-US frameworks. With only 23% of HWM locations 220 falling with the inundated area, AutoRoute appears to underestimate inundated area. However, the AutoRoute Event Map biases towards significant overestimation with large over predictions illustrated in Figure 4. Injecting each USGS HWM's WSE measurement uncertainty into our error analysis, we find that USGS measurement uncertainty in the HWM WSEs translates into an average of about ±1 cm difference in errors reported in Table 2. Overall, the performance of the HEC-RAS and Fathom-US frameworks is better that the AutoRoute framework. We certainly expected AutoRoute to underperform in 225 this scenario given the relatively simple numerical scheme and a lack of both pluvial and coastal flooding.

Causes of Framework Differences and Uncertainty
In general, we see the HEC-RAS, AutoRoute, and Fathom-US frameworks generate different Event Maps and that each is an that the AutoRoute framework. Here we explore the major drivers of differences and uncertainty amongst the estimated Event Maps.
One of the major differentiations of the AutoRoute framework from the HEC-RAS and Fathom-US frameworks is the missing coastal component of the Event Map. AutoRoute has proven capable in a variety of inland scenarios (Follum et al., 2017;255 2020) and when compared to higher resolution, inland models (Afshari et al., 2018). However, in this instance, it appears that the simplified physics in our AutoRoute simulation do not accommodate the complex physical interactions that occur during this compound coastal flood. In our case study, the AutoRoute Event Map under-predicts WSE but is also prone to large outliers of over-estimation in WSE estimation. As we expected, the HEC-RAS and Fathom-US frameworks outperform the AutoRoute framework in terms of error. Creek HEC-RAS model only considers the total runoff coming into the Clear Creek domain from Armand Bayou. When developing an Event Map with different frameworks, the user must understand the assumptions made by the modeler. In this instance, the application of runoff from Armand Bayou enters the HEC-RAS framework. However, because the runoff is not applied in a distributed manner throughout the watershed, an under representation of modeled inundation occurs upstream of Armand Bayou's pour point, effectively removing pluvial and fluvial flooding from the region. 280 The Fathom-US framework is the only framework we consider that explicitly accounts for pluvial flooding from precipitation.
The Fathom-US framework does this by performing a rain-on-grid simulation that relates local soils and land use to infiltration capacity and drainage design standards (Sampson et al., 2013;Wing et al., 2019). The AutoRoute framework is exclusively

Implications of Model Differences
As expected, each of the three modeling frameworks we consider estimate different Event Maps. The differences in Event Maps translate into different estimates of consequences and exposure. Table 3  where either AutoRoute and not HEC-RAS or HEC-RAS and not AutoRoute predict flooding, average depths are 3.8 m and 1.1 m, respectively. Thus, the AutoRoute framework appears to estimate higher total damage values while exposing fewer buildings than those produced by the HEC-RAS framework because of a bias towards greater water depth estimates at those locations. These results indicate that the differences in each Event Map produce different estimates of both exposure and consequences. We use the locations of the buildings impacted, the damage to those building, and the number of people within those buildings from each Event Map go-consequences analysis to construct a kernel density map ( Figure 6) where we see a spatial pattern that matches the tabular values in Table 3. The HEC-RAS framework estimates that the highest density of impact will be in the western and southern portions of the study domain. As stated before, the HEC-RAS framework omits distributed internal 325 boundary conditions in Armand Bayou watershed in the northeast portion of the study area, due to the modeling assumptions.
The AutoRoute framework estimates that the highest density of impacts will occur in pockets throughout the study domain.
The Fathom-US framework mimics the spatial pattern of the HEC-RAS framework but broadens estimates of impact throughout the entire study domain and in particular in the northeast section that the HEC-RAS framework omits. Overall, the densities portrayed in Figure 6 match well will the magnitudes of consequences and exposure portrayed in Table 3  The Event Maps produced by each framework are different in terms of their spatial and physical composition and each estimates different consequences and exposures to the floodwaters. We may assume that the Event Map produced by HEC-RAS is the most accurate given the better fit between observed and simulated WSE (Figure 4 and Table 2). However, the 340 HEC-RAS framework is not without error, has a lower locational accuracy than the Fathom-US framework (Table 1), and does not intend to represent flood inundation in the northeast section of the study region (Armand Bayou). Furthermore, as we compare FEMA flood insurance claim locations from Hurricane Harvey (Arctur, 2021) to each Event Map, we find evidence that the HEC-RAS framework Event Map is indeed excluding flooding in the northeast portion of the study area. Figure 7 compares the location of FEMA insurance claims for structures in the AOI and the estimate of buildings per square kilometer 345 from Figure 6.  When we calculate the proportion of FEMA insurance claims falling within each Event Map's flood inundation extent (Table  355 4), we see that none of the frameworks captures all FEMA claims and the pattern echoes the quantitative pattern of HWM data in Table 1. However, if we sum all FEMA claims that fall within at least one of the Event Maps we capture a slightly greater portion of FEMA claims. Thus, our case study may further illustrate the importance of a multi-model, probabilistic approach to Event Map creation that can convey uncertainty in the chosen Event Map framework (e.g., HEC-RAS, AutoRoute, Fathom-US). Deterministic Event Maps, considering only one source of inundation mapping, do not convey to the decision maker the 360 cascading uncertainty in the methods, tools, and data that create the Event Map (Merwade et al., 2008). NOAA's National Hurricane Center (NHC) has worked extensively with stakeholders to develop products that assist with decision-making. Thus, an Event Map produced for a given compound coastal flood should likely follow a similar convention.

Model Proportion of FEMA flood claims within Event Map
Flooded Area

All Frameworks Combined 86%
In 2018 dollars, residents of the study area made roughly $898 million in total FEMA flood insurance claims in the wake of Hurricane Harvey. Thus, using the assumption that between 26-50% of residents in our study domain possess flood insurance, 370 and dividing the total flood insurance claims by these proportions, we estimate between $1.2-3.6 billion of total flood damage to structures and contents resulted because of Harvey in our study domain. These values bounds roughly correspond with the $0.7-3.3 billion estimated by each Event Map. Thus, we surmise that our total damage estimates we produce roughly align with what occurred in reality during Hurricane Harvey.

How to Improve Event Map Creation Techniques 375
Efforts are ongoing to coordinate Event Map creation at the federal level. The three frameworks discussed in our study are not the only techniques available to create Event Maps during flood events. As previously mentioned, the NWC produces HANDderived Event Maps using the NWM (Viterbo et al., 2020) Thus, there is a need to reconcile and adjudicate multiple Event Maps to ensure consistency in decision-making efforts during flood events. In response to this need, the IWRSS consortium has set about operational plans for coordinating Event Map 385 production through the integrated Flood Inundation Mapping (iFIM) effort (Gutenson, 2020). The iFIM group confers before, during, and after major flood events in order to promote awareness of the various Event Map creation efforts. The iFIM effort is in its infancy, gathering together to understand the where and when of Event Map production. However, this is a necessary first step in building cohesion in developing appropriate Event Maps. In our current context, the iFIM group would have been https://doi.org/10.5194/nhess-2022-27 Preprint. Discussion started: 16 February 2022 c Author(s) 2022. CC BY 4.0 License.
aware that the HEC-RAS framework should not be representative of the northeast section of the study domain and that the 390 AutoRoute framework generally performs poorly in low gradient coastal watersheds. This adjudication process would have likely led to the iFIM group promoting the Fathom-US framework for use in the northeast section of the study region and the HEC-RAS framework in the rest of the study area as the most appropriate Event Map.
To empower the iFIM group, additional steps to enable interoperability and sharing of maps across multiple levels and divisions of government will also be necessary. From a practical perspective, this means developing data services to share amongst the 395 different agencies. NOAA's NWS, USACE, and USGS all provide access to Event Maps through geographic information systems (GIS) services. The next step will be the engagement of other Federal entities and those that fall outside of the Federal agencies. Simply exposing these Event Maps as GIS services and allowing the iFIM group to import them within a common operating picture will empower the Event Map adjudication and promotion process.
The iFIM intends to promote the most appropriate Event Map for a given flood event and location. However, as we have seen 400 with this case study of Clear Creek during Hurricane Harvey, a deterministic Event Map can be problematic for compound coastal flooding given that all chosen modeling frameworks produce an imperfect assessment of reality. As Table 4 displays, our combination of all three Event Maps encompasses a greater proportion of FEMA flood claims than one location alone.
Thus, we have some initial evidence to suggest that the delivery of a multi-model Event Map should be the preferred methodology to Event Map delivery. 405 However, a chosen Event Map framework highlights only one aspect of the uncertainty within Event Map creation. This assessment has not considered the uncertainty associated with the use of numerical weather prediction (NWP) models. Even with gains in NWP forecast skill, the use of ensemble prediction remains key to understanding the uncertainty when predicting chaotic weather systems. Ensemble prediction entails the perturbation of initial conditions and model numerical schemes to create a range of possible meteorological conditions (Palmer, 2017). Thus, the delivery of an ensemble, multi-model 410 probabilistic Event Map should be the preferred methodology to deliver an Event Map in order to convey uncertainty to decision makers. Figure 8 represents a hypothetical inland centric, forecast system where n number of NWP hydrometeorlogical ensemble forecasts force a tide and storm surge model and the hydrometeorlogical and tides and surge ensembles force each Event Map framework. The result of such a system would be a multi-model ensemble based probabilistic Event Map, similar to that proposed by Zarzar et al. (2018). In general, expansion of the full expression of knowledge 415 uncertainties, extending beyond model selection and NWP forcing into areas such as coefficient determination for hydraulic structures, will generally benefit the portrayal of flood risk in Event Maps.  Two of the large-scale frameworks (Fathom-US and AutoRoute) we employ here have the potential to generate timely probabilistic Event Maps using hydrometeorological ensemble forecasts (Wing et al., 2019). However, to take advantage of hydrometeorological ensemble forecasts within local-scale frameworks, such as our HEC-RAS example, we must effectively reduce model runtime. The setup and runtime of these local frameworks may affect the timeliness of Event Map creation, 425 which is crucial during emergency operations (Follum et al., 2017;Longenecker et al., 2020;Gutenson et al., 2021). However, local scale frameworks offer high fidelity, high-resolution products that can improve a probabilistic Event Map. One means to reduce model runtime for local scale models is to develop and train surrogate models that can dramatically reduce the computational runtime of high fidelity, high-resolution modeling while delivering similar results (Zahura et al., 2020;Contreras et al., 2020;Kyprioti et al., 2021). In fact, Bass and Bedient (2018) have already developed such a surrogate 430 modeling approach to create an Event Map within our study area that loosely couples inland and coastal models, forcing both with a full range of potential tropical cyclone characteristics. However, accurately training surrogate models for compound hazards is not trivial, given the need to expose the surrogate technique to numerous pre-existing simulations that account for the multitude of physical interactions, initial conditions, etc. that expand beyond tropical cyclone forcing.
Inevitably, improvements in numerical schemes and input data should provide improvements in Event Map creation. In their 435 review of the literature, Santiago-Collazo et al. (2019) determine that 96% of the literature they analyse presents compound coastal flood inundation modeling strategies employ one way coupling. By one-way coupling, we mean were outputs from one model (e.g., inland) are fed into another model (e.g., coastal) by way of internal or external boundary conditions and no https://doi.org/10.5194/nhess-2022-27 Preprint. Discussion started: 16 February 2022 c Author(s) 2022. CC BY 4.0 License. feedback occurs between the coupled models. The HEC-RAS and Fathom-US frameworks discussed here are examples of one-way coupling strategies as the models insert coastal surge into both frameworks via downstream head boundary conditions. 440 Santiago-Collazo et al. (2019) advocate for the use of more robust coupling strategies to account for the complex interaction between inland runoff and storm surge; such as loosely-coupled, tightly-coupled, or fully-coupled modeling strategies. Further, Event Map improvement will undoubtedly occur as improvements to the widespread availability of critical input datasets occur. For instance, the USGS collection of improved DEM data is steadily decreasing vertical and relative DEM errors (Gesch et al., 2014). 445

Conclusions
In this manuscript, we compare three different Event Map creation frameworks for a small coastal watershed, Clear Creek, near Houston, Texas during Hurricane Harvey. These frameworks are the HEC-RAS framework, the AutoRoute framework, and the Fathom-US framework.
We estimate the maximum flood inundation raster from each framework, considering this our Event Map (IWRSS, 2013). We 450 then compare each framework's Event Map to USGS HWMs in two ways. First, we assess whether the Event Map contains each HWM within the estimated flood extent. Second, we compare observed WSE from the USGS HWM to estimated WSE in the Event Map. Our analysis indicates that Event Map accuracy can vary based upon either of these assessments. The Fathom-US framework contains the most HWMs but also tends to overestimate WSE. The HEC-RAS framework contains less HWMs but also tends to have relatively more accurate WSE. The AutoRoute framework is the least accurate of the three, 455 appears to underestimate flood extent, and highlights how simplified flood inundation mapping methods are not ideal for representing compound coastal flooding. Our analysis illustrates that no one Event Map is infallible and is subject to the uncertainties present in the model's numerical scheme, the model inputs (e.g., terrain), and the model's configuration.
We the estimate the exposure and consequences of each Event Map using the NSI and go-consequences. We find quantitative and spatial differences in the exposure and consequences produced by each Event Map. The differences we find between each 460 Event Map further illustrate why a singular, deterministic Event Map is not preferable. We compare our exposure and consequence estimates to the locations of FEMA flood claims and use FEMA damage claims totals to estimate a total damage. Visually ( Figure 7) and numerically, the comparison of simulated exposure and consequence estimates compare favorably to our approximate observations. The results lend credence to our ability to utilize accurate Event Maps, the NSI, and goconsequences and produce a relatively accurate exposure and consequence assessment for a flood event. Thus, the combination 465 may be a useful tool set for evaluating the impacts of floods before, during, and after they happen.
Our study highlights the need to rectify and adjudicate the various Event Maps created during flood events. In response to this need, IWRSS formed the iFIM to perform interagency comparison and consolidation of Event Maps. GIS web services empower the iFIM and adding additional Event Maps to the iFIM common operating picture will improve the Event Map selection and discovery process.
Large-scale Event Map creation techniques, such as AutoRoute and Fathom-US may be capable of operating in real-time during flood events. To develop Event Maps properly for compound floods and beyond, future research should focus on means to reduce runtime in local-scale models that offer high-fidelity numerical schemes and high-resolution input data. Surrogate modeling may offer such an approach but the difficulties in training a multivariate surrogate model are not trivial. Decreased runtimes may offer the ability to instantiate multiple model simulations while not compromising model fidelity. This would 475 make possible probabilistic Event Maps for compound coastal floods that capitalize on the fidelity and resolution of localscale models.

Author contribution
JLG and AAT conceptualized the study. JLG designed the study and conducted the experiments. JLG drafted the manuscript.
MSI and OEJW assisted with data collection and preparation. WPL and COH assisted with model set up and implementation. 480 MSI, OEJW, MDW, and TCM provided comments and feedback on the manuscript draft.

Data availability
The presented HWM data are accessible on the USGS Flood Event Viewer (Flood Event Viewer, 2021).