Overview

Purpose

The National Structure Inventory (NSI) is a collection of point-based structure inventories used to support natural and technological hazard consequence analysis. The NSI provides consistent structure locations and attribution, including occupancy type, square footage, foundation characteristics, population estimates, and economic value estimates. Flood risk analysis is the primary use case, but the data structure also supports consequence analysis for other natural and man-made hazards.

This document describes the NSI data structure and the methods used to produce the 2026 NSI Base dataset. The 2026 Base dataset is a nationally consistent modeled inventory. It is intended to provide a practical starting point for consequence analysis where more detailed local inventories are unavailable, too costly to produce, or infeasible within study time constraints.

Background

The National Structure Inventory Base layer was created and is maintained by the U.S. Army Corps of Engineers (USACE). It was originally developed to simplify geospatial preprocessing for consequence modeling, but it has since been used in a variety of USACE, FEMA, and interagency applications.

The NSI includes point locations for structures and a standardized set of attributes required by hazard consequence models. These attributes are exposed through downloadable datasets and a RESTful API. The standardized schema allows analysts and software tools to work with inventories from different locations and sources using a common set of fields.

Limitations Statement

The NSI Base dataset should be understood as a modeled national exposure inventory rather than a structure-by-structure verification dataset. Many attributes are estimated from parcel data, building footprints, Census data, national datasets, regional distributions, and other modeled assumptions. These assignments are intended to preserve reasonable aggregate distributions, but they should not be interpreted as verified structure-level observations. As a result, individual structure records may contain errors in location, occupancy type, square footage, number of stories, foundation type, foundation height, population, or value.

The Base dataset is most appropriate for national and regional screening, initial project setup, and studies where higher-resolution local inventory data are not available. For project areas where consequences are sensitive to individual structures or where decisions depend on high-confidence local estimates, analysts should review and refine the NSI using local data, surveys, or other project-specific information. Additional detail on limitations and suggested adjustments is provided in the FAQ Document.

Accessing the NSI

The NSI is available in GeoPackage, GeoParquet, and GeoJSON formats. State-level GeoPackage and GeoParquet files can be downloaded from the NSI Download web page. GeoJSON files can be requested for individual counties or user-specified bounding boxes through the NSI API.

The NSI API uses a consistent schema across datasets. This consistency allows software tools to query, display, and analyze structure inventories without requiring custom field mappings for each geography or source inventory.

NSI Base Data

The Base dataset is generated from multiple national and local data sources, including parcel data, business data, building footprints, HIFLD facility datasets, Census data, and other federal datasets.

The Base dataset is not an exact representation of existing conditions at every structure. It is a modeled inventory that combines observed data where available with statistical assignment, regional assumptions, and rule-based processing where direct observations are unavailable. Its purpose is to provide a nationally consistent minimum standard for consequence analysis and a starting point that can be improved with local data or project-specific surveys.

Public Fields

The NSI API uses a standardized set of public fields to describe each structure record. These fields support common consequence modeling workflows, including structure damage estimation, life safety analysis, population exposure, and economic loss estimation.

Datasets added to the NSI must conform to the required schema. When values are not directly observed, analysts are responsible for providing reasonable estimates and documenting the assumptions used to populate those fields. At present, the publicly available NSI Base layer is maintained by the NSI development team. Future upload functionality is expected to allow additional inventories to be added using the same schema. The table below lists the public fields currently exposed in the NSI Base schema.

Private Fields

Some NSI fields are restricted to federal users and mission partners because they contain source-specific, sensitive, licensed, or otherwise restricted information. These fields are not generally required to run standard consequence models, but they can provide useful context for analysts refining the Base dataset or evaluating the assumptions behind individual records.

NSI Source Data

The 2026 NSI Base dataset was created by integrating multiple data sources that describe development type, structure location, structure characteristics, population, and economic value. Parcel data often provide the initial use classification for a location, such as single-family residential, multi-family residential, commercial, industrial, or public use. Business datasets, facility datasets, and school datasets are used to identify specific non-residential uses that may not be adequately represented in parcel data alone.

Structure Identification and Aggregation

Initial structure coordinates are estimated from source data, such as parcel centroids, address geocodes, or facility point locations. The NSI Generator refines these initial locations by matching structures to building footprints within the same parcel where footprint data are available. Source locations outside an associated parcel may be moved to a more appropriate parcel based on distance and similarity of plausible use codes.

When a parcel contains multiple footprints, structures are assigned to the largest suitable footprints first. When multiple structure types occur within the same parcel, assignment priority is generally given to schools, then commercial structures, and then residential structures. If all suitable footprints have already been paired with structures, additional records may be stacked at the same footprint location. See the FAQ for additional information on “stacked” structures.

If the generated inventory contains fewer residential units than indicated by 2020 Census housing unit counts, additional residential units are added. These units are preferentially assigned to empty footprints and existing multi-family structures, but they may also be assigned to large footprints associated with non-residential use when needed to reconcile housing unit totals.

Major Attribute Fields

Occupancy Type

Source categories are mapped to NSI occupancy types, which are broadly consistent with FEMA HAZUS occupancy classifications and USACE consequence modeling needs. Residential occupancy types are further refined during processing based on estimated housing units, number of stories, and basement status. For example, a single-family residential structure is assigned a RES1 subtype that reflects stories and basement condition.

Population Estimates

The NSI 2026 population estimates are designed to approximate current day and night exposure at the structure level. Because 2020 Census block-level population counts may not fully represent current conditions, the NSI combines 2020 block data, tract-level aggregation, county-level population estimates, LEHD worker flows, school enrollment, and facility-specific population data to estimate 2025 population exposure.

Population Pools

Commercial worker population was derived from the U.S. Census Bureau’s Longitudinal Employer-Household Dynamics (LEHD) database. This database contains counts for the number of residents leaving a census block to work and the number of workers arriving in a census block. The NSI sums these block-level estimates to the tract. Departing workers are subtracted from the residential population; as are enrolled students. With few exceptions, LEHD data is in 2023 values.

The NSI estimates tract-level population growth in an iterative process until the total increased population for the county, relative to 2020, is depleted. Population is first added to tracts that had no housing units in 2020 but now have housing units in the newly generated inventory. Next, population is distributed to tracts whose number of housing units is greater in the NSI than it was in the 2020 census. Finally, population is randomly assigned to census tracts until the population growth is fully distributed.

Population is assigned from 20 separate pools, reflecting combinations of Day and Night, Over and Under 65 years of age, and for five population categories, including Workers and Residents.

Structure Level Assignment

Once tract-level population estimates are made, population is assigned to particular structures within each tract. Population is assigned from commercial population pools to commercial businesses weighted by number of employees and square footage and from residential population pools to residences weighted by number of housing units. The three remaining population pools: institutional population, nursing home population, and student population are used to adjust or assign population to specific subcategories of non-residential structures.  The assignment process also accounts for the relative likelihood of those over 65 years of age to work or stay at home. RES4 (Hotels, Motels, etc.) receive visitors for day and night based on their square footage.

Economic Valuation

The NSI provides estimated values for structures, contents, and vehicles. For structures, the NSI distinguishes between replacement-cost-new and depreciated replacement value. Replacement-cost-new represents the estimated cost to construct a comparable new structure. Depreciated replacement value adjusts that estimate to reflect the age, condition, deterioration, and remaining useful life of the existing structure.

USACE flood risk management studies estimate expected consequences to existing assets. Because existing structures are not new, their economic value is generally not equal to the cost of constructing a new structure. Depreciated replacement value is therefore used to estimate the value at risk in the without-project condition and to avoid overstating benefits as though every damaged structure would be replaced with a brand-new asset.

Full Replacement Cost

Replacement-cost-new is estimated using each structure’s occupancy type, square footage, and an occupancy-specific cost per square foot. All values are reported in August 2025 price levels. National cost assumptions are informed by survey data, parcel use types, and other source inputs used to identify typical construction forms associated with each occupancy type.

Cost per square foot generally decreases as structure size increases, reflecting assumed economies of scale. The estimated cost per square foot is multiplied by the structure’s estimated square footage to produce the replacement-cost-new estimate.

For most occupancy types, the NSI applies a national cost curve adjusted for local construction wages. Single-family residential structures are additionally adjusted based on the income of the structure’s Census tract relative to the income of its Census division. This follows the traditional HAZUS assumption that housing quality and construction cost vary with income. Higher-income tracts are assumed to have more expensive finishes and materials, while lower-income tracts are assumed to have less expensive features.

The below figure illustrates the RES1 replacement-cost-new adjustment for a 2,000 square foot, one-story home with no basement. The structure cost is interpolated from the tract income ratio, with lower-income tracts receiving a greater share of the economy-cost assumption and higher-income tracts receiving a greater share of the custom or luxury-cost assumptions. The percentile line is included for context and shows the approximate share of tracts at or below each income ratio. Most tracts fall near an income ratio of 1.0 and the cost ranges from $125 to $335 per square foot before county wage adjustment.

The national averages are further adjusted by the county wage costs as estimated by the Bureau of Labor Statistics. In counties where construction wages are higher than average prices are increased by that ratio, while counties with lower wages have reduced construction costs. No adjustments are made for differences in material costs or differences in regional construction practices.

Depreciated Structure Value

Depreciation is the adjustment from replacement-cost-new to the remaining value of an existing structure. It accounts for deterioration that occurred before the flood and for variation in remaining useful life. In practical terms, depreciation prevents a 50-year-old structure from being valued the same as a newly constructed structure with the same footprint, materials, and occupancy type.

For USACE flood risk assessments, depreciation is not merely an accounting convention. It is a benefit-estimation concept. The objective is to estimate the expected economic loss from flooding to the existing structure, not the cost of constructing a new structure. Depreciated replacement value therefore represents the value at risk in the without-project condition.

The NSI estimates depreciation based on both the structure's age and the likely level of maintenance and reinvestment within the surrounding area. Age-based depreciation is consistent with common appraisal and replacement-cost approaches, such as those used by commercial cost-estimating methodologies. However, the NSI is intended to represent the value of the existing surviving building stock, which reflects decades of maintenance, renovation, and component replacement. As a result, depreciation is adjusted using Census-derived indicators including tract income, renter occupancy, and age demographics, which serve as proxies for the level of reinvestment likely to occur within a tract.

Rather than applying a single depreciation curve nationwide, the NSI uses a family of depreciation curves representing lower-, typical-, and higher-maintenance conditions. Structures in tracts with characteristics associated with lower reinvestment follow a steeper depreciation path, while structures in tracts associated with higher reinvestment follow a flatter path. Intermediate tracts are assigned values through interpolation between these curves. This approach allows the NSI to capture broad differences in the condition of the housing stock while remaining practical for national-scale application.

The below figure illustrates how the depreciation model produces different remaining-value estimates for structures of similar age depending on the tract-level maintenance assumption. The "Poorly" and "Well" curves represent lower- and upper-bound maintenance conditions, respectively, while the "Average" curve represents a typical level of reinvestment. Remaining value declines with age under all assumptions, but the rate of decline varies based on the estimated maintenance characteristics of the tract.

Content Value

Content values are estimated by multiplying depreciated structure value by an occupancy-specific content-to-structure value ratio. These ratios are shown in the occupancy type table. Applying the ratio to depreciated structure value implicitly assumes a similar degree of depreciation for contents.

The content-to-structure value ratios are based on averages from studies that include expert elicitation, survey data, and claims data. Ratios tend to be higher for occupancy types with specialized equipment or relatively low structure costs. For RES1 structures, the NSI uses a ratio of 0.5. This differs from the 1.0 ratio assumed in USACE EGM 04-01. Users applying EGM damage functions calibrated to structure value should use the EGM ratio. For most other applications, the NSI ratio is intended to better approximate typical content value.

Vehicle Value

The NSI estimates the likely depreciated replacement value of vehicles at residential and non-residential structures. Values are varied based on vehicle ownership rates in a tract, tract income, and the ratio of high clearance (pickup trucks, SUVs, etc.) to low clearance vehicles (sedans, etc.) in a state and their respective costs. Tract-level vehicle ownership estimates and state level vehicle clearance estimates are provided in the NSI as separate fields. The economic value estimates assume vehicles are located at home 80% of the time while non-residential structures implicitly assume each worker’s vehicle is located at the structure 20% of the time.

Structure Characteristics

The NSI includes several structure attribute fields that influence consequence estimates by affecting damage functions, stability criteria, exposure estimates, or valuation. When possible, these fields are populated from structure-specific source data. Where structure-level data are unavailable, values are estimated using regional distributions, source-specific assumptions, or rule-based assignment.

Square Footage and Number of Stories

Methods for estimating square footage and number of stories differ between single-family residential structures and other occupancy types. Single-family residential structures often have more reliable parcel data, including reported square footage and number of stories. In addition, there is typically only one primary structure on a single-family parcel, reducing ambiguity about which structure the parcel attributes describe.

When square footage is missing for single-family residential structures, it is estimated from the building footprint where available. If no footprint is available, square footage is randomly assigned from a distribution that varies by year built and Census region. When number of stories is unavailable, it is estimated from building height where height data are available. Structures over 5 meters are assumed to have at least two stories, with each additional 3.1 meters adding another story. If neither height nor footprint data are available, number of stories is randomly assigned from a distribution that varies by year built and Census region.

For RES3 and non-residential structures, parcel-reported square footage and stories are generally less reliable. These parcels often contain multiple buildings, incomplete structure-level attribution, or values that cannot be confidently assigned to a specific footprint. For these structures, square footage and number of stories are primarily estimated from footprint area, building height, housing unit demand, employee-based space demand, and occupancy-specific rules.

Construction Type

Construction type is mapped from parcel data where available. Where construction material is not available, it may be inferred from exterior wall type. When neither structure-specific construction material nor exterior wall information is available, construction type is assigned from HAZUS-based regional distributions. These assignments support consequence modeling applications that require broad building material categories, such as wood, masonry, manufactured housing, concrete, or steel.

Foundation Type and Height

Foundation type is mapped from parcel data where available. When parcel foundation data are unavailable, foundation type is assigned using regional and hazard-setting assumptions based on Energy Information Administration data, American Housing Survey data, HAZUS data, and USACE survey information.

For riverine settings, American Housing Survey data provide foundation type estimates by Census division, while Energy Information Administration data provide basement status by Census division. These regional base rates are refined to state-level assumptions using USACE national survey data and frost-depth assumptions. For coastal settings, HAZUS building scheme information is used to assign foundation types such as slab, pier, crawl space, basement, and pile. Assignment probabilities vary by flood zone and structure age relative to flood mapping.

Where foundation information is partially available, such as when only basement status is populated, county-level rates are adjusted so that unknown records preserve reasonable local proportions across foundation categories.

Foundation heights are assigned by foundation type. For example, RES1 slab foundations are assigned a default height of 0.75 feet and basement foundations are assigned a default height of 2 feet. These assumptions were informed by a 2021 USACE survey and generally align with median survey values.

Supplemental Information

Version Notes

Dataset version: NSI 2026 Base

Price level: August 2025

Population level: April 2025

Primary coordinate system: WGS84

Primary intended use: national and regional consequence screening, feasibility-level analysis, and project setup where better local inventory data are unavailable

Recommended QA: local review of structure locations, occupancy types, foundation heights, values, and population assumptions for high-consequence areas

Acronyms

Resources and References

Draft Resources and References

This is a paste-ready draft for your three-part reference section, organized the way you have been building the NSI 2026 technical documentation and updated with the higher-confidence items I could verify in public sources. I sorted entries by lead author or organization and flagged the strongest “likely used” research items in the methods section. 

Guidance and Standards

U.S. Army Corps of Engineers. Engineer Manual 1110-2-1619: Risk-Based Analysis for Flood Damage Reduction Studies. USACE.

U.S. Army Corps of Engineers. Engineer Regulation 1105-2-100: Planning Guidance Notebook. USACE

U.S. Army Corps of Engineers. Engineer Regulation 1105-2-101: Risk Analysis for Flood Damage Reduction Studies. USACE.

U.S. Army Corps of Engineers. Engineer Regulation 1105-2-103: Policy for Conducting Civil Works Planning Studies. USACE.

Source Data Documentation

Federal Emergency Management Agency. USA Structures. FEMA.
URL: https://gis-fema.hub.arcgis.com/pages/usa-structures 

IPUMS USA. IPUMS USA. IPUMS, University of Minnesota.

Microsoft. USBuildingFootprints. GitHub repository.

National Center for Education Statistics. Common Core of Data. NCES.

U.S. Bureau of Labor Statistics. Occupational Employment and Wage Statistics (OEWS). BLS.

U.S. Census Bureau. American Community Survey (ACS). Census Bureau

U.S. Census Bureau. American Housing Survey (AHS). Census Bureau.

U.S. Census Bureau. Characteristics of New Housing. Census Bureau.

U.S. Census Bureau. Community Resilience Estimates (CRE). Census Bureau.
U.S. Census Bureau. Longitudinal Employer-Household Dynamics (LEHD). Census Bureau

U.S. Department of Transportation, Federal Highway Administration. Highway Statistics Series. FHWA

U.S. Energy Information Administration. Residential Energy Consumption Survey (RECS), 2020 Housing Characteristics Tables. EIA.
U.S. Geological Survey. 3D Elevation Program. USGS.
 

Methods and Supporting Research

Eriksen, Michael D., and Anthony W. Orlando. Returns to Scale in Residential Construction: The Marginal Impact of Building Height. SSRN, 2021.

Federal Emergency Management Agency. Hazus.

Glaeser, Edward L., Joseph Gyourko, and Raven Saks. “Why is Manhattan So Expensive? Regulation and the Rise in House Prices.” Journal of Law and Economics 48, no. 2 (2005): 331–369. 

Harding, John P., Stuart S. Rosenthal, and C. F. Sirmans. “Depreciation of Housing Capital, Maintenance, and House Price Inflation: Estimates from a Repeat Sales Model.” Journal of Urban Economics 61, no. 2 (2007): 193–217. T

International Association of Assessing Officers. Standard on Mass Appraisal of Real Property. Revised 2025. International Association of Assessing Officers.

International Code Council. Building Valuation Data. International Code Council.

Kim, Duckgil, Yonsoo Kim, Daewon Jang, and Chunwoo Baek. “A Study on the Estimation of CSVR for Flood Damage Analysis.” Journal of the Korean Society of Hazard Mitigation 19, no. 6 (2019): 303–311.

Knight, John R., Thomas Miceli, and C. F. Sirmans. “Repair Expenses, Selling Contracts, and House Prices.” Journal of Real Estate Research 20, no. 3 (2000): 323–336. 

Lynch, Eric. Cost of Constructing a Home-2024. National Association of Home Builders, 20 January 2025.

U.S. Army Corps of Engineers, Fort Worth District. Dallas Floodway Project Appendix E: Flood Risk Management Analysis. USACE

U.S. Army Corps of Engineers, Institute for Water Resources. National Economic Development Procedures Manual: Urban Flood Damage, Volume I. U.S. Army Corps of Engineers, Institute for Water Resources.

U.S. Army Corps of Engineers, Memphis District. Memphis Metropolitan Stormwater–North DeSoto Feasibility Study, Desoto County, Mississippi, Appendix L: Economics. USACE

U.S. Army Corps of Engineers / Interagency Performance Evaluation Task Force. Volume VII: The Consequences. IPET.

U.S. Army Corps of Engineers, Jacksonville District. Appendix C: Florida Keys Economics Appendix. USACE

Mission, Vision, and Goals

Mission
Provide access to consistent, nationally available, point-based structure inventories with attribution that supports consequence analysis for natural and man-made hazards.

Vision
Support federal, state, local, tribal, territorial, academic, and private-sector partners through a shared structure inventory framework that improves hazard consequence analysis, disaster response, and long-term risk planning.

Goals

1. Provide broad access to nationally consistent structure inventory data.
2. Improve the quality, transparency, and documentation of structure inventory data.
3. Support disaster response through better exposure and consequence information.
4. Support planning for future disasters by improving the consistency and usability of hazard consequence inputs.