Product

Decision-support software built for review.

Explore historical hurricane replay today, or discuss software for a problem in your own data: records that need bringing together, an analysis that needs to be reproducible, or results people need to inspect.

Explore the Observatory Scope a project or evaluation

A working example

Follow a historical storm from record to replay

1. Start with the records

Katrina’s historical sequence contains recorded positions, maximum sustained wind, and pressure at successive times.

2. Step through the replay

The Observatory displays the track and observations alongside the Temporal V2 diagnostic describing patterns in intensity history.

3. Inspect the result

Move through the timeline, compare the observed values and displayed state, and copy a link to a particular replay step.

This example demonstrates historical review. Temporal V2 does not directly measure physical storm structure or establish that the same method works in another application.

PTEM Observatory is descriptive and observational. It does not issue forecasts, warnings, operational guidance, or causal claims. See Research for the methods and limitations.

Inspect the Katrina replay

Software capabilities

From source records to inspectable results

Preserve the source

Retain source identity, availability timing, and input versions when bringing records into a data pipeline.

Inspect the assessment

Keep original data separate from the assessment and displayed result, with missing or conflicting information visible.

Replay the result

Retain input and software versions so analytical results can be reproduced and checked.

Build review interfaces

Develop browser applications and APIs for inspecting data and assessments.

For internal R&D, PTEM Labs developed Docker-based processing workers, AWS/Terraform configurations, and checks for estimated costs and permitted work before a run. Engineering capabilities include repeatable build configurations and automated schema, regression, and integration tests. The capabilities used in a project depend on its requirements.

Start a discussion

Scope an evaluation around your question

Tell us what you want to understand or improve, which data sources you have, who would review the results, and any access or deployment constraints. A description of the data is enough to start the conversation.

Together, we can define the inputs, a comparison approach, and what would count as a useful result. Depending on the agreed scope, an evaluation may produce analytical outputs, comparative findings, documentation, and read-only review of results using preserved inputs.

Evaluation is scoped individually. Materials and review access are agreed before work begins; customer-specific integration, deployment, and cybersecurity readiness require separate qualification. An evaluation does not by itself establish operational readiness.

Observatory evaluation access does not include proprietary source code or authorize forecasting, warnings, catastrophe-model replacement, rate-setting, or automated operational decisions. See Validation for how evidence is assessed.

Discuss an evaluation