Key unique features of the CoSMoS flood model include:

1. 1. Beyond the ‘bathtub’ approach: Many flood mapping products use a simple or modified “bathtub” method, where water levels are raised uniformly across the landscape without accounting for coastal processes. In contrast, CoSMoS flood modeling includes the full dynamics of coastal water levels (e.g., tides, waves, and storm surge).
   2. Future-focused: Rather than relying on historical storm records, CoSMoS uses projected changes in atmospheric conditions such as wind, temperature, and rainfall from global climate models to project coastal storms under changing climatic conditions through the 21st century.
   3. Addresses compound flooding: Accounts for the combined effect of coastal and riverine flooding.
   4. A broad suite of planning scenarios: Projections of multiple storm scenarios are provided under a suite of sea-level rise scenarios. These options allow users to manage and meet their own planning horizons and specify degrees of risk tolerance.
   5. A variety of hazard information for more nuanced planning: In addition to the traditional flood extent, CoSMoS also provides minimum and maximum flood extent (to help bracket uncertainty), flood depth, water surface elevation, velocity hazard, and duration.
   6. Pacific Coast flood models incorporate geomorphic change: Rather than assuming static topography, CoSMoS will account for future geomorphic changes (e.g., erosion) in flood projections for areas on the outer Pacific Coast (products not yet available).

1 – While the suite of CoSMoS scenarios remains the same across all geographies, the underlying modeling continues to be improved with advances in science (e.g., updated global climate models), technology (e.g., computing capabilities), and by incorporating additional processes important to specific regions. For example, we include the influence of coral reefs in Hawaii and the Pacific Island Territories, hurricanes and pluvial flooding in the Southeastern U.S., and have made significant improvements in the handling of compound coastal and riverine flooding for Washington.

CoSMoS products are primarily used for long-range planning and coastal resilience activities. CoSMoS results have been used to inform: 

• • Vulnerability assessments and coastal adaptation planning 
  • Municipal plans (e.g., Shoreline Master Programs, Comprehensive Plans, Hazard Mitigation Plans) 
  • Infrastructure/capital improvement plans 
  • Natural resource management/restoration plans 
  • Public engagement and communication 
  • Emergency management 

It is important to note that CoSMoS was not developed for site-specific engineering projects (e.g., site design). CoSMoS may not resolve all the fine-scale hydrologic processes that influence site-specific flooding (e.g., storm-drain network, soil infiltration rates). CoSMoS was designed as a screening tool to assess community-scale vulnerabilities and to help identify and prioritize where more detailed engineering studies may be required.

CoSMoS modeling results can be accessed in two ways: (1) as downloadable geospatial data compatible with most GIS software (https://cmgds.marine.usgs.gov/data-releases/community/Cosmos_SalishSea/), and (2) as part of the USGS Hazard Exposure Reporting and Analytics (HERA) web tools (https://www.usgs.gov/apps/hera), which allow users to visualize most CoSMoS hazard products and understand how coastal hazards could impact land, people, and infrastructure at the community scale. The HERA tools let users choose a place of interest and see how community exposure to certain hazards changes across a range of scenarios. For example, HERA can provide estimates of the number of people, miles of roads, or number of critical facilities that may be exposed to flooding with 100cm of sea-level rise and a 20-year storm.

2 – The following flood products/scenarios are not currently visualized in the HERA tools (available only as data downloads): water surface elevation (relative to NAVD88), flood duration, and the 10-year and 50-year storm frequency scenarios.

Flooding products for Washington include seven sea-level rise scenarios from 0 to 9.8 ft above current mean sea level (see table below). These can be combined with six possible coastal storm frequency scenarios (explained further below). Combined, CoSMoS provides 42 possible combinations for visualizing a wide range of coastal hazard scenarios. To maintain consistency across states, only four of the storm scenarios are included in the HERA tool. The 10-year and 50-year products are only available as downloadable geospatial data at this time.

Sea Level Rise Scenarios

Coastal Storm Frequency Scenarios

CoSMoS flood hazard data are provided for a wide range of sea-level rise scenarios and they are intentionally not tied to a specific point in time. This allows users to consult the latest sea-level rise projections for their local geography and consider their planning/project time horizon and level of risk tolerance. In many cases, users have opted to assess vulnerability to multiple sea-level rise and storm scenarios.

There are two key sets of sea-level change projections available for use in Washington State: (1) highly-localized and probabilistic projections published in 2018 by Miller et al., and (2) projections from the 2022 Federal Interagency Technical Report (Sweet et al. 2022).  How to Choose: A primer for selecting sea-level rise projections for Washington State (Raymond et al. 2020) is available on the Washington Coastal Hazards Resilience Network website: https://wacoastalnetwork.com/research-and-tools/slr-resources/. This document provides guidance for understanding and using highly-localized and probabilistic sea-level rise  projections published in 2018 and available online (Miller et al., 2018). A similar guide, Application Guide for the 2022 Sea Level Rise Technical Report (Collini et al. 2022) is available to walk users through applications of a set of sea level projections published in 2022 (Sweet et al. 2022). These projections are presented as a set of curves that users select between, and are available for individual tide gauge locations in Washington State. These projections are also available online. 

Sea level projections can be connected to the sea level scenarios in CoSMoS as follows:  a user may choose sea level scenarios in CoSMoS that align with  sea-level projections at various time horizons (e.g., 2050, 2080, 2100), or that represent different sea-level rise likelihoods  (low, intermediate, high) for a given time horizon. Consider the sea-level rise projections from Miller et al. (2018) below (Figure 1).  A user might consider the 1.6 ft CoSMoS SLR scenario and understand that it could occur as early ~2080 under the high emissions trajectory (RCP 8.5, blue line), or as late as ~2090 under the lower emissions trajectory (RCP 4.5, red line).

Figure 1. An example showing how one may reference sea-level rise projection curves from Miller et al. (2018) to determine which CoSMoS scenarios to choose. The RCP 4.5 and 8.5 projections account for different global emissions pathways and ocean and atmospheric warming. Black dashed lines indicate four CoSMoS SLR scenarios, and how they might correspond to different pathways through time. Circles indicate where CoSMoS scenarios intersect with the median projections.

In 2025, the Washington Department of Ecology began amending rules for the Shoreline Management Act that will specify how local governments must address the impact of sea-level rise and increased storm severity in Shoreline Master Programs. As part of this process, the Department of Ecology will be providing additional guidance on selection of scenarios. Visit this webpage to stay informed on the status of the rulemaking process: https://ecology.wa.gov/regulations-permits/laws-rules-rulemaking/rulemaking/wac-173-26-27-shoreline-management-act.

In CoSMoS, storm return frequencies describe the likelihood of a certain magnitude of flooding (i.e., water level) in a specific location caused by a wide range of storm conditions. They do not reflect the frequency of a certain type of storm (e.g., an atmospheric river), or the frequency of certain types of causal factors (e.g., high winds or heavy rainfall). For example:

• • A 20-year storm means there is a 1-in-20 chance (5%) that flood waters reach this water level in any given year 
  • A 100-year storm means there is a 1-in-100 chance (1%) that flood waters reach this water level in any given year

These are long-term statistical averages: a 20-year storm could occur in back-to-back years, or multiple times in one decade, followed by a long, quiet period of no storms. Also, because these statistics are calculated for every model grid cell, return period water level elevations vary spatially. The 100-year water level elevation in downtown Seattle is likely different than the 100-year water level elevation in Tacoma, or in the South Park neighborhood along the Lower Duwamish.

It is important to note that the methodology for determining CoSMoS storm return frequencies differs from FEMA’s calculation of “100-year flood” zones. FEMA maps are based on historical flooding conditions. CoSMoS, in contrast, projects future flooding under a changing climate by simulating hundreds to thousands of storms using global climate model projections of wind, atmospheric pressure, and rainfall. See “How is CoSMoS modeling different from FEMA” for further discussion. 

Storm scenarios in CoSMoS are therefore composites of many possible storm events. They incorporate the combined effects of:

• • Sea-level anomalies (seasonal variations and El Niño–related processes)
  • Storm surge (increased water levels from winds and low pressure)
  • Wave set-up (increased water levels due to breaking waves)
  • River discharge (floodwaters pushing back at the coast, increasing local water levels)

By simulating these many drivers together, CoSMoS produces a more realistic picture of potential future flooding under different return intervals. However, it is important to understand that a “100-year storm” in CoSMoS is not a prediction of one particular storm, but rather a statistical representation of many possible storms that could cause these extreme water levels. Tidal cycles are also included in the modeling, and results show that extreme water level events in King and Pierce counties typically coincide with high tides.

In contrast, the “average daily condition” scenario was defined as approximately Mean Higher High Water (MHHW) and includes only tidal forcing and average stream discharge conditions (i.e., no surge or wave set-up) to represent a typical “calm” day.

Detailed methodology for CoSMoS flood modeling in King and Pierce counties is provided in three journal articles covering the following: (1) regional Salish Sea wave and hydrodynamic models (Parker et al., in prep), (2) regional streamflow model (Buitink et al., 2025), and (3) local overland flood models for both Pierce and King counties (Nederhoff et al., 2025).

The overall approach followed the USGS CoSMoS (Barnard et al. 2014) framework, and in particular the methodology employed most recently in the U.S. Southeast Atlantic (Barnard et al., 2024; Nederhoff et al., 2024a). The overall modeling approach followed a 3-tiered structure (Figure 2), with global scale data providing atmospheric and oceanographic boundary conditions as well as information on the influence of climate change. Future projections were based on the Shared Socioeconomic Pathway 5-8.5 (SSP5-8.5, similar to the RCP8.5 scenario). SSP5-8.5 represents a “business as usual” scenario in which carbon emissions are not effectively reduced until past 2050. Regional-scale wave, hydrodynamic, and hydrologic models take the downscaled global forcings and calculate water elevation, waves, and streamflow through the Salish Sea region, at the nearshore for oceanographic processes and at ~20 m (NAVD88) topographic contour for streamflow. A local-scale model then takes the nearshore water levels and wave heights, combined with streamflow and a high-resolution topo-bathymetric elevation model, to simulate overland flood propagation. Models were extensively calibrated and validated using several approaches, including comparison to historical data and to the FEMA 100-year flood hazard maps.

Figure 2. Conceptual framework for data moving from global to local scale modeling. Blue boxes represent models while orange boxes represent data.  Arrows delineate passing of data between models.

3 – Shared Socioeconomic Pathway (SSP) scenarios were used in the most recent set of climate model experiments, known as the Sixth Phase of the Coupled Model Intercomparison Project (CMIP6), which formed the basis for the Sixth Assessment Report (AR6) of the Intergovernmental Panel on Climate Change (IPCC). They have replaced the Representative Concentration Pathways (RCPs) used for CMIP5 and AR5.

FEMA Flood Insurance Rate Maps delineate the current 100-year water level extent for the purposes of the National Flood Insurance Program based on historical conditions. Due to the diversity of FEMA studies and methodologies across the U.S., specific details about individual models are not always available (https://msc.fema.gov/portal/home). Typically, extreme boundary conditions determined from historical data are routed through 1- or 2-dimensional hydraulic models. The purpose, underlying models, and input data (e.g., Digital Elevation Model, ocean and atmospheric forcing conditions) are different than the CoSMoS flood modeling. In addition to methodological differences, FEMA maps do not provide projections for sea-level rise or other recurrence period floods. Further details can be found in Nederhoff et al. (2025), which compares the flood extents from a 100-year CoSMoS flood model run under current climate conditions to the latest FEMA 100-year flood maps for King and Pierce counties.

The minimum and maximum flood extents show the potential range of flood extent that may be experienced for a given scenario. The flooding simulations account for physical processes that contribute to future coastal flooding, producing outputs useful for adaptation planning.  However, as with all simulations, it is not a perfect representation of present or future conditions. To account for potential uncertainty in boundary conditions, two additional simulations were performed with “high” and “low” parameter estimates as follows (see Nederhoff et al. 2025 for detailed rationale):

• • (Increase/Decrease) riverine discharge by 20%
  • (Increase/Decrease) coastal water levels by 50 cm 
  • (Decrease/Increase) riverbank height by 50 cm

By using the minimum and maximum flood extent information, a user can consider the potential for less or more extreme flood exposure given the range of accuracy of the model and all its supporting components (including the Digital Elevation Model).

The Digital Elevation Models (DEMs) used for CoSMoS combine the best available topographic and bathymetric data to create seamless, high-resolution models of coastal elevation. Elevation data for the entirety of the coastal regions of King and Pierce counties were derived from the USGS Coastal National Elevation Database (CoNED) topographic model of Puget Sound (Tyler et al., 2020; https://topotools.cr.usgs.gov/topobathy_viewer/). The CoNED dataset provides a seamless topobathymetric DEM at a 1-meter resolution, constructed from the latest high-resolution datasets, including light detection and ranging (LiDAR) topography, multibeam and single-beam bathymetry, and other sources. Datasets were merged into a continuous surface to ensure spatial consistency and accuracy. Because the DEM contains information from multiple sources collected at different times, more recent human-made changes (e.g., residential or commercial development) may not be captured in certain areas.

The CoNED DEMs were used to characterize nearshore zones, beaches, rivers and levees as accurately as possible, as these details are critical for capturing the hydrodynamic processes that govern coastal and riverine flooding.  As source datasets were not as accurate within river channels, river bathymetry was modified to improve the representation of channel morphology and flow dynamics within the local-scale flood model simulations (see Nederhoff et al., 2025 for details). River bathymetry remains a significant source of uncertainty in all flood hazard modeling, and CoSMoS is no exception.

The DEM is a “bare earth” representation of the earth’s surface, meaning that surface objects are removed. Removed objects include buildings, other temporary structures, and vegetation. Features such as earthen levees and breakwaters are preserved. In addition, subgrid features were added to the overland flood model to enforce flow blocking topography from things like railroad berms, riverbanks, or levees that were not well represented in the DEM or unable to be resolved due to model resolution. While significant effort was put into capturing many of these important fine-scale features, not all of them could be identified. All subgrid features and changes to the DEM are documented as part of the Super-Fast INundation of CoastS (SFINCS) model input files included in the USGS data release (see “How can the CoSMoS products be accessed?”) and can be referenced to better understand what specific fine-scale features are included in the modeling.

Vertical land motion (long-term uplift or subsidence) was not explicitly included in the modeling because data at the necessary scale and resolution were unavailable at the time. Estimates of VLM for coastal Washington vary spatially, ranging from 4 mm/year of uplift to -1 mm/year of subsidence (Newton et al. 2021). These changes influence relative sea-level changes in a given location; however, VLM has already been accounted for in the relative sea-level rise projections of Miller et al. (2018).  Thus, CoSMoS scenarios remain robust (see “How do I choose which scenarios to use for planning?”), especially when paired with the min/max flood hazard zones that account for model uncertainty (see “What do the minimum and maximum hazard zones represent”).

CoSMoS outputs do not account for storm water drainage infrastructure (e.g., storm drains) because regional scale data on drainage infrastructure do not exist and were not feasible to integrate into the current modeling architecture at a county level.  Flood hazard maps may, therefore, overestimate flooding in areas where water could drain via subsurface features and should therefore be considered as a “worst case” should drainage become blocked. It also may underestimate flooding into low-lying areas that are protected from overland flooding by surrounding topographic features (e.g., a berm), but that may actually be hydrologically connected via a storm drain or culvert. These low-lying disconnected areas are identified as polygons in a product called “flood prone low-lying” and should be considered for the most robust assessment of potential flood exposure risk.

Dam operations are an important control on riverine and compound flooding, but they are challenging to model because they are complex and likely to change in the future as various factors (environmental, political, etc.) affect decision making. The CoSMoS modeling approach applied simple logic rules to dam management (see Buitink et al., 2025 for more details). As a result, the streamflow inputs are a significant source of uncertainty, especially for extreme events. Streamflow uncertainty is included in the minimum/maximum flood hazard products (see “What do the minimum and maximum flood hazard zones represent?”), but these do not capture emergency dam operations with no historical analog.

Washington Sea Grant and partners are in the process of updating the Puget Sound Parcel-Scale Sea Level Rise Vulnerability Assessment to improve the geospatial analysis conducted in Phase 1 (Miller et al., 2023; Coastal Geological Services et al., 2022). The project calculates and maps a sea-level rise vulnerability index for every parcel in Puget Sound that accounts for potential impacts to both the built and natural environment. The updated (Phase 2) index will include exposure to flooding, erosion, and rising groundwater. 

The main differences are:

• • CoSMoS provides hazard mapping information that can be used for exposure analysis and vulnerability assessments (hazards only, no impacts)
  • HERA provides exposure analysis at the community- and county-scale for a selection of specific assets (e.g., how many people, roads, hospitals are in the hazard zone)
  • The Parcel-Scale Vulnerability Assessment provides an integrated metric for every individual parcel in Puget Sound that accounts for exposure risk from multiple coastal hazards (flooding, erosion, rising groundwater) and potential impacts to both the built and natural environment.

The Phase 2 Parcel-Scale assessment required data to be publicly available with Puget Sound-wide coverage in order to be included in the update. Because of timing of CoSMoS data availability, only the CoSMoS groundwater hazard information is being incorporated into the Phase 2 project. As additional CoSMoS data become available, they could be incorporated into future iterations of the Parcel-Scale Assessment tool. More information on the Parcel-Scale Vulnerability Assessment is available here: https://wacoastalnetwork.com/puget-sound-parcel-scale-sea-level-rise-vulnerability-assessment/

The Pierce and King County efforts fit within a larger effort to expand CoSMoS results for the greater Salish Sea region. The regional-scale flood modeling infrastructure was originally developed and published in 2023, including the regional wave model (Crosby et al., 2023) and regional hydrodynamic model (Grossman et al., 2023). The first local-scale overland flood model was developed for Whatcom County (Nederhoff et al., 2024b). 

While the Whatcom County results remain reliable and high quality for that geography, science and modeling continue to advance. Thus, we have updated flood modeling for King and Pierce counties to better capture climate and compound (i.e., combined coastal and riverine) flood dynamics, as well as to improve computational efficiency. Key updates included:

• • Updating global scale climate forcing data to the latest generation ensemble of high-resolution global climate models, as well as implementation of a method that limits biases from global climate models and improves computational efficiency.
  • Adding dynamic streamflow modeling to more accurately resolve compound flooding.
  • Shifting from event-based to response-based characterization of hazard return periods. This improves event characterization by ensuring that the co-occurrence, sequencing, and persistence of coastal and riverine flood drivers are more accurately captured across space and time. It also improves statistical representation of extremes.
  • Significantly improving computational efficiency by not resolving infragravity waves, which contribute very little to total water levels in the King/Pierce region of Puget Sound.
  • Improvements that allowed the local overland flood model to be run over much larger domains, thereby producing a more accurate and seamless water level surface.

Similar flood products will be rolled out for the remaining Puget Sound counties, with a regionally complete dataset anticipated by the end of 2027. A timeline of products by geography is regularly updated on the Coastal Hazards Resilience Network website (https://wacoastalnetwork.com/resources/tools/cosmos/). 

At this time, USGS is focusing on completing flood modeling for all of Washington state. While there are no plans to update existing flood models, updates could be considered if a significant change occurs, specific end-user needs arise, and sufficient funding and staff capacity are available.