Building Inventories Technical Implementation

The Consequence Modeling Solution is designed to natively ingest the National Structures Inventory (NSI) and the Milliman Market Basket datasets out of the box, while also providing a pathway for users to integrate their own custom inventory data. Table 1 identifies the supported data sources, versions, expected input formats, and applicable consequence-modeling pathways.

Table 1. Supported Inventory Data Sources

Data Source Version Input File Type Consequence Modeling
NSI 2022 Public Version GeoPackage Inland, Coastal
NSI 2022 FEMA-Enhanced Version File Geodatabase Inland only
Milliman Market Baskets 2021 Uniform, Uncorrelated File Geodatabase Inland, Coastal
User-defined Inventories User-defined User-defined Inland, Coastal

For additional details on inventory requirements and methodology, refer to the Inventory Methodology Documentation.

Inventory Ingestion and Processing Approach

Regardless of which inventory type is provided, several required fields must be present to support loss calculations. However, the Consequence Solution is designed to minimize the amount of data preprocessing required from the user by implementing the following strategies:

  • Auto-detection of required fields based on common naming conventions
  • Auto-population of missing fields using documented default values
  • User-provided overrides for NSI and Milliman field names
  • Comprehensive documentation for each inventory source, including requirements for user-defined datasets
  • Schema validation to ensure completeness and correctness prior to modeling

Base Buildings Class

TODO: Define the Base Buildings Schema used for consequence analysis in both inland and coastal applications. All target_fields referenced in subsequent sections should be defined here.

National Structures Inventory (NSI)

NSI Example

The National Structures Inventory (NSI), developed by the U.S. Army Corps of Engineers (USACE), is a nationwide database of structures across the 50 U.S. states. At present, the NSI does not include coverage for U.S. territories. The publicly available NSI dataset provided many of the key fields used for consequence analysis; however, it is important to note that building and contents values are reported as depreciated values rather than full replacement costs. Full technical documentation for the NSI is available on the USACE NSI Technical Documentation page.

USACE also maintains a restricted version of the NSI accessible to federal users, which contains additional attributes derived from parcel data and other proprietary sources.

To support national hazard risk assessments, FEMA has developed an enhanced internal version of the NSI that extends coverage to the District of Columbia, Puerto Rico, the U.S. Virgin Islands, and Pacific Territories. This FEMA-enhanced dataset applies full replacement values consistent with the Hazus 7.0 Inventory Technical Manual and resolves several known data quality issues identified in the public NSI.

NSI 2022 Public Version

The 2022 Public NSI contains all attributes required to support both inland and coastal consequence modeling. Table 2 lists the key fields used in the Consequence Modeling Solution, including data types, assumptions, rules, and default values assigned when data is missing.

Table 2. NSI 2022 Public Data Attributes for Analysis

Analysis Attribute Data Type NSI Field Name Consequence Modeling Assumption Rule Default if Missing
Geometry Point Shape Inland, Coastal Point geometry Must be valid point geometry Error (required)
Unique ID Object ID fid Inland, Coastal Provided Must be present and unique Error (required)
Occupancy Type String occtype Inland, Coastal Provided; must map to Hazus types Must map to Hazus types RES1
Building Value Numeric val_struct Inland, Coastal Depreciated USD value Must be present Error (required)
Content Value Numeric val_cont Inland, Coastal Depreciated USD value Validate > 0; assign default if missing Default % by occupancy (see Inventory Methodology Table 3)
Number of Stories Numeric num_story Inland, Coastal Provided Validate > 0; assign default 1
Area / Square Footage Numeric sqft Inland Square footage Validate > 0; assign default Hazus defaults (Inventory Methodology Table 2)
General Building Type String bldgtype Inland Provided (M, W, H, S) Assign default if missing W (Wood)
Foundation Type String found_type Inland, Coastal Provided (C, B, S, P, I, F, W) Assign default if missing Slab
Foundation Height Numeric found_ht Inland, Coastal Feet above ground Assign default Slab = 1 ft; Shallow = 3 ft; Pile = 8 ft; Basement = 2 ft
Ground Elevation Numeric Ground_elv Coastal Feet (NAVD88) Required for coastal modeling Error (required)

NSI 2022 Public – Inland Foundation Type Mapping

For the NSI Public dataset, the found_type field must be mapped to the standardized inland foundation types used by the Consequence Modeling engine.

Implementation Rules:

  • The NSI found_type field is always used as the source field.
  • Foundation types must be translated using Table 3.
  • If a code does not match a supported value:
  • Treat the foundation type as uncertain, and
  • Assign the default inland foundation type SLAB.
  • No parcel-derived refinements are available in the public NSI.

Table 3. Inland Foundation Type Mapping for NSI 2022 Public Version

found_type Description Assigned Inland Foundation Type
C Crawl SHAL
B Basement BASE
S Slab SLAB
P Pier SHAL
F Fill SLAB
W Solid Wall SHAL
I Pile PILE

NSI 2022 Public – Coastal Foundation Type Mapping

TODO: Define coastal foundation type mapping rules for the NSI Public dataset.

NSI 2022 FEMA-Enhanced Version

The NSI 2022 FEMA-Enhanced Version allows users to calculate losses using full replacement costs and expands structure coverage to include Washington, D.C., Puerto Rico, U.S. Virgin Islands. This enhanced version also includes parcel-derived fields that support refinement of the inland foundation type assignment.

However, the FEMA-enhanced dataset does not include ground-elevation information. As a result, only inland consequence modeling is supported out of the box. Users may preprocess the dataset and manually supply ground elevations if they wish to perform coastal modeling using the user-defined inventory ingestion process. Table 4 summarizes the attributes used in the analysis, including data types, assumptions, rules, and defaults applied when values are missing.

Table 4. NSI 2022 FEMA-Enhanced Data Attributes for Analysis

Analysis Attribute Data Type NSI Field Name Assumption Rule Default if Missing
Geometry Point Shape Point geometry Must be valid point geometry Error (required)
Unique ID Object ID OBJECTID Provided Must be present and unique Error (required)
Occupancy Type String OCCTYPE Provided; must map to Hazus occupancy types Must map to Hazus types RES1
Building Value Numeric Hazus_Building_Values Full replacement cost (USD) Must be present Error (required)
Content Value Numeric Hazus_Content_Values Full replacement cost (USD) Validate > 0; assign default if missing Default % by occupancy (Inventory Methodology Table 3)
Number of Stories Numeric NUM_STORY Provided Validate > 0; assign default 1
Area / Square Footage Numeric SQFT Square footage (sq ft) Validate > 0; assign default Hazus defaults (Inventory Methodology Table 2)
General Building Type String GENERALBUILDINGTYPE Provided (M, W, H, S) Assign default if missing W (Wood)
Foundation Type String FNDTYPE Provided (C, B, S, P, I, F, W) Assign default if missing Slab
Foundation Height Numeric FOUND_HT Feet above ground elevation Assign default if missing Slab = 1 ft; Shallow = 3 ft; Pile = 8 ft; Basement = 2 ft
Foundation Type (Parcel) String P_FNDTYPE Provided Used only for parcel-based refinement logic None
Basement Type (Parcel) String P_BSMNT Provided Used only for parcel-based refinement logic None

NSI 2022 FEMA-Enhanced — Inland Foundation Type Mapping

For the FEMA-Enhanced NSI, parcel-derived fields (P_FNDTYPE and P_BSMNT) are used to refine the inland foundation type. The following rules determine the assigned inland foundation type:

Refinement Rules:

  • If P_BSMNT contains any of the following codes — B, U, I, F, P, L, D, Y — set inland foundation type to BASE.

  • If P_FNDTYPE contains any of the following codes — S, P, L, F, E, T, W, A — inland foundation type must stay blank/null, even when P_BSMNT indicates a basement.

  • If P_BSMNT does not indicate a basement, inland foundation type comes directly from P_FNDTYPE using the mapping in Table 5.

  • If P_FNDTYPE is null, use the NSI-provided FNDTYPE field and map it using the same rules as the NSI 2022 Public Version (found_type). FNDTYPE in FEMA-Enhanced is equivalent to found_type in the Public NSI.

Table 5. NSI 2022 FEMA-Enhanced Parcel Foundation Type Mapping

P_FNDTYPE Description Assigned Inland Foundation Type
P Piers SHAL
A Concrete SLAB
W Wood SHAL
C Crossed Walls SLAB
S Slab SLAB
K Concrete Block SHAL
O Other NULL
G Stone SHAL
D Masonry SHAL
L Piling PILE
R Retaining Wall SHAL
T Footing SLAB
B Crawl / Raised SHAL
F Mud Sill SLAB
E Earth SLAB
Z Placeholder If P_BSMNT indicates basement → BASE, otherwise Slab

Milliman Market Baskets Data

Milliman Uniform Book Example The Milliman Market Baskets were developed by Milliman, Inc. to support FEMA’s Risk Rating 2.0 initiative. Milliman created Market Baskets for all U.S. states and territories to provide a representative sample of single-family homes (RES1) used in the development of rating factors. Market Basket locations were derived primarily from CoreLogic ParcelPoint data, supplemented with U.S. Census and National Hydrography Dataset (NHD) information, and refined through extensive quality control to ensure accuracy and realistic spatial distribution.

Three categories of data, referred to as “books”, were created from the Market Baskets:

  1. Uniform Book – Each property assigned identical structural and coverage characteristics, allowing geographic variability to be analyzed independently.
  2. Uncorrelated Market Basket – Contains randomized property and policy characteristics not correlated with geography; foundation type and first-floor height remain linked to prevent implausible combinations.
  3. Correlated Market Basket (Inforce Dataset) – Joined with FEMA policy data (GFE access required); attributes are correlated to reflect realistic joint distributions and align with observed parcel and NFIP exposure data.

The Consequences Solution is designed and tested using the Uniform and Uncorrelated Market Basket datasets. Table 6 describes the analysis attributes used across these books. Although the schema is consistent, the method of attribute imputation varies by dataset. For example, the Uncorrelated Market Basket randomizes property and coverage characteristics, while the Correlated Market Basket applies state-specific distributions to more closely represent actual conditions.

The Milliman Market Basket datasets support both coastal and inland loss calculations, as they include the necessary structural, coverage, and geographic attributes for each modeling environment. Their use, however, is limited to single-family residential (RES1) structures, as the datasets were specifically developed for rating factor development under FEMA’s NFIP framework. For more details, refer to the FEMA (2022).

Table 6. Milliman Data Attributes for Analysis

Analysis Attribute Data Type Milliman Field Name Assumption Rule Default if Missing
Geometry Point Shape Point geometry Must be valid point geometry Error (required)
Unique ID Object ID Location Provided Must be present and unique Error (required)
Occupancy Type String Not provided Assume RES1 Assign default RES1
Building Value Numeric BLDG_VALUE Full replacement cost (USD) Must be present Error (required)
Content Value Numeric CNT_VALUE Full replacement cost (USD) Validate > 0; assign default Default % by occupancy (Inventory Methodology Table 3)
Number of Stories Numeric NUM_STORIE Provided Validate > 0; assign default 1
Area / Square Footage Numeric Not provided Assume 1,800 sq ft Assign default Hazus RES1 default = 1,800 sq ft
General Building Type String Not provided Assume Wood Assign default W (Wood)
Foundation Type Numeric foundation Milliman codes: 2, 4, 6, 7, 8, 9 Should be populated; assign default Slab
Foundation Height Numeric FIRST_FLOO First-floor height above ground (ft) Should be populated; assign default Slab 1 ft; Shallow 3 ft; Pile 8 ft; Basement 2 ft
Ground Elevation Numeric DEMft DEM ground elevation (ft, NAVD88) Must be present Error (required)
Basement Type Numeric BasementFi 0=None, 1=Unfinished, 2=Finished Provided Default = None
Content Insurance Deductible Numeric CNT_DED Provided Not used in loss calculations None
Content Insurance Limit Numeric CNT_LIM Provided Not used in loss calculations None
Building Insurance Deductible Numeric BLDG_DED Provided Not used in loss calculations None
Building Insurance Limit Numeric BLDG_LIMIT Provided Not used in loss calculations None

Milliman — Inland Foundation Type Mapping

For both the Uniform Book and Uncorrelated Market Basket datasets, the foundation field contains numeric codes that must be mapped to inland foundation types used by the Consequences Solution. These mappings support correct depth–damage function assignment.

Table 7. Milliman Inland Foundation Type Mapping

foundation Description Assigned Inland Foundation Type
2 Basement BASE
4 Crawlspace SHAL
6 Pier SHAL
7 Fill / Wall SLAB
8 Slab SLAB
9 Pile PILE

Milliman — Coastal Foundation Type Mapping

To be developed.

Milliman Data Schema:

{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "$id": "https://example.com/schemas/milliman_schema.json",
  "title": "Milliman Buildings Schema",
  "version": "0.1.0",
  "description": "Typical schema for Milliman building data (Uniform, Uncorrelated Market Basket)",
    "default fields": {
        "location": {
            "type": "string",
            "description": "Location ID of building which serves as the unique identifier",
            "default target field": "id",
            "required": true,
            "nullable": false
        },
        "BLDG_VALUE": {
            "type": "number",
            "description": "Building replacement cost ($USD)",
            "default target field": "building_cost",
            "required": true,
            "nullable": false
        },
        "CNT_VALUE": {
            "type": "number",
            "description": "Contents replacement cost ($USD)",
            "default target field": "contents_cost",
            "required": true,
            "nullable": false
        },
        "NUM_STORIE": {
            "type": "number",
            "description": "Number of stories in the building",
            "default target field": "number_stories",
            "required": true,
            "nullable": false
        },
        "foundation": {
            "type": "number",
            "description": "Foundation type code (2 = basement; 4 = crawlspace; 6 = pier; 7 = fill or wall; 8 = slab; 9 = pile)",
            "default target field": "foundation_type",
            "enum": [2, 4, 6, 7, 8, 9],
            "required": true,
            "nullable": false
        },
        "BasementFi": {
            "type": "number",
            "description": "Basement finish type code (0 = no basement; 1 = unfinished basement; 2 = finished basement)",
            "default target field": "basement_finish_type",
            "enum": [0, 1, 2],
            "required": true,
            "nullable": false
        },
        "FIRST_FLOO": {
            "type": "number",
            "description": "First floor height (feet above ground) ",
            "default target field": "first_floor_height",
            "required": true,
            "nullable": false
        },
        "DEMft": {
            "type": "number",
            "description": "Digital elevation model (DEM) elevation (feet NAVD88)",
            "default target field": "dem_height",
            "required": true,
            "nullable": false
        },
        "BLDG_DED": {
            "type": "number",
            "description": "Building insurance deductible  ($USD)",
            "default target field": null,
            "required": false,
            "nullable": true
        },
        "BLDG_LIMIT": {
            "type": "number",
            "description": "Building insurance limit ($USD)",
            "default target field": null,
            "required": false,
            "nullable": true
        },
        "CNT_DED": {
            "type": "number",
            "description": "Contents insurance deductible ($USD)",
            "default target field": null,
            "required": false,
            "nullable": true
        },
        "CNT_LIMIT": {
            "type": "number",
            "description": "Contents insurance limit ($USD)",
            "default target field": null,
            "required": false,
            "nullable": true
        }
    }
}

However, users may provide a crosswalk (a.k.a. override) from the target fields to the user's non-standard fields.

Example of User-Provided Override for Milliman Field Names: (keys=target_field, values=user's field)

{
    "building_cost": "my_custom_building_cost_field"
}

User Defined Data

TODO: Insert definitions a user-defined buildings dataset. Seems this should reflect the base class.