# Storm History

Storm History lets users find earlier storm evidence, reopen a briefing, and understand what information remains available. A supervisor can revisit a Missouri report; a canvasser can return to an assigned area; a researcher can select a past window for comparison. This is a product capability with a shared testing role, not a separate archive built only for developers.

The repository has bounded MRMS event and revision reads plus retained benchmark windows. A complete searchable customer history, durable report archive, and physical-storm catalog remain proposed. See the [History and Replay contract](/stormtrust/product/platform-and-data/intelligence/history-replay-contract/) for implementation evidence and gaps.

## Find the right history

Begin with place and time, then narrow by evidence type, hazard, coverage, or report where supported. Results should show the geographic scope, observation window, latest retained revision, source completeness, and available actions. Keep the selected query when returning from a result.

An entry must say whether it represents a national feed, a regional analysis window, an established physical storm, or a report. A county intersecting a warning is not automatically a county with observed hail. A regional window can contain several storms, and an empty search result does not prove no storm occurred there.

## Open evidence in context

Selecting an entry can open [Storm Map](/stormtrust/product/platform-and-data/intelligence/storm-map/), [Storm Replay](/stormtrust/product/platform-and-data/intelligence/storm-replay/), or an authorized [Storm Report](/stormtrust/product/platform-and-data/intelligence/storm-report/). Preserve the analysis scope, source and report revisions, time window, and geographic selection across those views.

Show a report's original published revision and any superseding revision as separate choices. Later corrections should help users understand change without rewriting what a recipient previously received. Public weather evidence does not make a tenant's reports, assignments, or field notes public.

## Coverage is part of the result

History needs an explicit coverage and retention promise. Distinguish evidence not collected, partially collected, still processing, retained, expired, restricted, and unavailable after a failed retrieval. A catalog entry may remain when its raw data or rendered frames no longer exist.

Current [operational and benchmark retention](/stormtrust/product/platform-and-data/live-revisions-and-historical-evaluation/#operational-and-benchmark-retention) is not an unlimited customer archive. A saved link alone does not preserve its inputs. Define which report evidence, geometries, rule versions, and frames are retained before promising a period of historical access or reproducibility.

## Start with the report pilot

The first product slice can list retained Missouri pilot windows and report revisions, disclose their available evidence, and reopen one for review or replay. It does not require nationwide historical search, every county's parcels, or an automated physical-storm tracker.

Use those same records and read contracts for regression benchmarks. Fixture labels, permissions, retention policy, and execution authority remain explicit; shared machinery does not turn a test result into a production observation.