FEMA Disaster & Public Assistance Intelligence

Apify Actor for querying official FEMA OpenFEMA datasets. It preserves source records, adds normalized fields, supports bounded OData filtering/pagination, and emits transparent run summaries.

Data sources

The Actor uses FEMA's public OpenFEMA API over HTTPS:

• DisasterDeclarationsSummaries
• PublicAssistanceFundedProjectsDetails
• PublicAssistanceGrantAwardActivities

No login, API token, browser automation, cookies, or proxy is required.

A missing record means that the selected dataset and filters returned no matching row. It does not prove that no FEMA assistance exists. Dataset coverage and field availability vary by program and time.

Minimal input

{
  "datasets": ["disaster_declarations"],
  "states": ["CA"],
  "maxResults": 100,
  "pageSize": 50,
  "outputMode": "normalized"
}

Supported datasets

• disaster_declarations: disaster declarations and designated-area records.
• public_assistance: funded public-assistance project details.
• grant_award_activities: public-assistance grant award activity records.

Filtering

Structured filters include states, counties/designated areas, disaster numbers, incident and declaration types, applicant IDs, date bounds, modification timestamp, and federal-obligation amount bounds where supported by the selected dataset.

Advanced users may provide dataset-specific OData clauses through advancedFilters. The Actor rejects empty or obviously unsafe clauses before any network request. $select and $orderby are also supported.

All requests are bounded by maxResults, maxPages, pageSize, and maxRuntimeSeconds. Retryable HTTP responses such as 429 and 5xx are retried with bounded backoff. Permanent 4xx errors are reported rather than retried indefinitely.

Output modes

• normalized: normalized fields plus source metadata and, by default, raw_record.
• raw: original FEMA records plus the final run_summary.
• flat: normalized flat records.
• summary: aggregate statistics plus the final run_summary.

Normalized records include record_type, source_dataset, source_api_version, source_url, retrieved_at, source_last_updated, and deterministic raw_record_hash.

Original FEMA fields are not discarded when includeRawFields is enabled. Monetary meanings remain separate; for example, federal_share_obligated, total_obligated, and total_project_amount are never collapsed into a generic amount.

Row granularity and relationships

rowGranularity accepts:

• designated_area: retain declaration rows at designated-area level.
• disaster: aggregate declaration areas into disaster views.
• both: retain both declaration rows and related public-assistance project rows.

Relationships are only created using verified disaster-number and state keys. Unmatched public-assistance rows remain in output with an explicit relationship status.

Run summary

The Actor emits one record_type=run_summary record when includeSummary is enabled. It distinguishes:

• success
• empty
• partial
• failed
• truncated

Counts distinguish source rows, unique records, unique disasters, designated areas, projects, applicants, and separate monetary totals.

Limitations

• The initial release is stateless. Historical delta detection and prior-run change comparison are not enabled.
• FEMA dataset records have different coverage, update timing, and field semantics.
• A successful HTTP response alone does not guarantee matches; the response shape and dataset record list are validated.
• AI-generated summaries are not enabled and are not official FEMA conclusions.

Local verification

python3 -m pytest -q
python3 -m compileall -q src tests
python3 -m src

For an Apify run, the Actor reads input through the Apify SDK and pushes each output record to the default Dataset.