Multi-model forecast ingestion
Stormlines brings together GEFS, AIGEFS, ECMWF ENS, AIFS ENS, and GEPS as new runs arrive from three forecast providers.
Stormlines / An independent product
Read every path.
Know what to expect.
Stormlines turns 185 ensemble forecasts into storm tracks, timing windows, model comparisons, and direct answers about what could happen next.
Explore Stormlines- Forecast coverage
- 185 tracks
Five global ensembles - Design & development
- Alan Shiflett
Product strategy, UX & engineering - Scope
- Forecast ingestion, storm tracking, interactive maps & editorial explanation
The product problem
A storm does not
have one path.
A single forecast line looks precise. An ensemble tells a more honest story: repeated model runs spread across a range of possible paths, speeds, and strengths.
Stormlines keeps that disagreement visible, then helps a reader answer practical questions about direction, arrival, distance, rain, and confidence. Story mode leads with the answer. Nerd mode opens the underlying tracks, filters, and model readouts.
Inside the forecast
Start with the answer.
Open the evidence.
The same forecast data supports a quick read, a full model comparison, and a world view.
01 / The readable forecast
Put timing and uncertainty beside the map.
The primary view combines a likely path, uncertainty cone, arrival windows, chance near a place, and plain-language answers. The map stays connected to the questions people are trying to answer.
02 / The world view
Follow the weather beyond one storm.
The globe places tracked storms beside the next week of rain and snow. A time control lets readers move through the forecast, while a place search turns the world view into a local one.
What I built
The system behind
the story.
Storm detection and synthesis
The product groups ensemble members into tracked storms, follows their movement, and summarizes the range without hiding model disagreement.
Layered map interaction
Readers can move from the most likely path to every model track, filter by ensemble, inspect probability shading, and play the forecast through time.
Two depths of explanation
Story mode keeps the forecast readable. Nerd mode exposes the models, filters, measurements, and technical detail for readers who want them.
Forecast verification
The product compares earlier forecasts with what happened, so model performance can be examined instead of taken on faith.
Related capability
Data platforms & custom softwareHave complex data people need to understand?
We can turn a dense model, feed, or operating system into a product people know how to use.