# Storm Impact

Product design

This page describes the product model and requirements. Examples illustrate the model, not a guaranteed API or completed feature. Check the [application contracts](/stormtrust/apps/) and [dataset inventory](/stormtrust/product/platform-and-data/datasets-and-source-contracts/) for implemented scope. Use [operations guides](/stormtrust/engineering/operations/) for runnable procedures.

Storm Impact helps users decide where to investigate residential roofs after a storm. The intended outcome is to prioritize roofs most likely to need replacement, using storm evidence, property context, and a validated outcome model when those inputs exist.

## Geographic drill-down

The first [Storm Report](/stormtrust/product/platform-and-data/intelligence/storm-report/) uses **Missouri → counties → reviewed canvassing areas**. It prioritizes weather exposure before parcel matching. A work area can be a sourced neighborhood, a supervisor-reviewed territory, or a clearly labeled grid area. Its boundary, name, version, and county intersection must be traceable.

The later property workflow is **states → counties → county portions of ZIP/ZCTA areas → parcels**. Neighborhoods can remain an optional field grouping; they do not add a mandatory step between a ZIP/ZCTA portion and its parcels. Each list and map retain the same evidence window and ranking snapshot.

A ZIP code is a delivery identifier; a ZCTA is a Census statistical area. The [ranking contract](/stormtrust/product/platform-and-data/intelligence/impact-ranking/) defines intersections and membership. Do not assume ZIP areas nest in counties.

## Prioritize with the evidence available

The first report uses **weather-exposure priority**: qualifying radar exposure within accepted county and reviewed work-area geometry. It needs geographic analysis and quality checks, but no parcel or roof counts. The [report contract](/stormtrust/product/platform-and-data/intelligence/report-contract/#weather-exposure-ranking) owns the proposed area measure and its approval requirements. A supervisor can add attributed residential or access context without disguising it as a model result.

Later **parcel-exposure priority** uses accepted residential parcel data to compare exposed properties. **Replacement priority** requires roof identity, permitted outcome labels, calibration, validation, and a versioned model. Neither weather area nor exposed parcel counts are probabilities of needing a new roof. Missing parcel coverage does not prevent an area briefing; missing work-area geometry does prevent a neighborhood ranking.

## Explain a rank

Every ranked result should identify the ranking basis, affected modeled units, source coverage, observation time, model or rule version, and supporting evidence. Display unknown and unevaluated areas alongside ranked results so a missing county dataset is not mistaken for low impact.

In the report pilot, selecting a county opens its reviewed work areas and starting markers. In the later property workflow, clicking a state narrows to its counties; clicking a county narrows to its ZIP/ZCTA portions; clicking a portion opens its ranked parcels. A parcel dossier then explains the source attributes, roof evidence if available, and next permitted investigation action. Contact eligibility remains a separate decision.

## Property records

A property record may aggregate normalized address, geocode, parcel, structure geometry, building and roof attributes, licensed ownership data, occupancy, commercial metadata, tenant relationships, and weather exposure history.

## Source hierarchy

Property information may come from licensed providers, government geography, permitted public assessor sources, tenant imports, user corrections, and StormTrust derivations. Source precedence is explicit.

## Property overrides

Tenant corrections do not overwrite vendor provenance:

```txt
provider value: 1998
tenant override: 2003
effective value: 2003
```

The interface identifies an effective value that contains an override.

## Roof intelligence

Potential roof attributes include type, material, estimated age, replacement date, pitch, area, facets, condition, warranty status, and installer.

Not every attribute exists for every property. Unknown is a valid value.

## Residential and commercial models

Residential and commercial prioritization use different economic context. Commercial workflows may emphasize square footage, facility type, ownership entity, roof type, portfolio membership, and building use.

## Property exposure

A proposed `ImpactAssessment` links a property to an event revision. These fictional values are not a calibrated model result or an implemented API:

```json
{
	"propertyId": "prop_123",
	"eventId": "ST-2026-08-16-KC-001",
	"eventRevision": 4,
	"meshMaxInches": 1.72,
	"poshMax": 91,
	"exposure": "high",
	"confidence": 0.88,
	"classification": ["radar-derived", "report-observed"]
}
```

## Damage likelihood

A future damage-likelihood model may use estimated hail, confidence, roof material and age, storm duration, repeated exposure, slope, and environmental context.

The result remains an estimate until a qualified physical inspection confirms damage.

## Opportunity score

Opportunity is a business concept. Potential inputs include impact likelihood, property and roof relevance, contactability, service-area fit, customer relationship, recency, and operational capacity.

The proposed scoring algorithm must be versioned and explainable. Business opportunity must not replace the exposure or replacement ranking basis.

## Eligibility

Eligibility is separate from opportunity. A high-value property may remain ineligible for a contact channel because of consent, suppression, tenant policy, jurisdiction, provider license, customer relationship, or channel rules.

## Property dossier

A dossier combines the property and roof profile, event history, current impact, evidence timeline, confidence, photos, reports, customer history, and action history.