Custom weather data

A dataset shaped around your work.

Model output for the place, fields and time span your team needs. Built to a spec we agree with you, and priced per project.

Have a clear spec, or just a question? Both are a good place to start.

Who it is for

Start with the decision the data serves.

AI and ML

Training and evaluation datasets

Define a consistent archive, levels, variables, training rights and a layout your pipeline can read.

Scope your dataset

Energy and renewables

Wind, solar and siting

Specify heights, radiation fields and hindcast years for resource studies and power-curve work.

Scope your dataset

Insurance and risk

Events and long archives

Scope wind, rain and severe-weather fields around your event set, footprint and validation needs. Hail guidance needs an availability review.

Scope your dataset

Agencies and research

A spec you can reproduce

Bring the domain, model settings and source-data requirements for your service or study.

Scope your dataset

Broadcast and media

A feed for your audience

Set the market, forecast horizon, run cycles, fields and rights for your graphics or weather service.

Scope your dataset

Not sure yet?

Start with the question

Tell us what you are building. You can ask for help with the technical scope.

Scope your dataset

What you get

Project-specific output. An agreed spec.

We scope, generate, package and deliver model outputs and derived datasets. The quote sets out which fields, formats, documentation and checks are included.

The data

  • Region, domain size and grid spacing
  • Variables, units and vertical levels
  • Archive dates or forecast horizon and cycles
  • Output interval, forcing and ensemble needs

The handover

  • Requested NetCDF, Zarr or GRIB2 layout
  • Chunking, storage size and file inventory
  • Your cloud, a delivery bucket or an API
  • Rights for training, use and redistribution

Format and delivery support are confirmed in the scope.

The evidence

  • Requested data card: coverage, units, gaps and limits
  • Model version, physics and namelists
  • Initial and boundary-data provenance
  • Verification plan and sample acceptance checks
How to judge forecast skill

Bring a WRF or AceCAST 5 spec, or ask us to help define one. Regional MPAS with custom meshes is coming soon; you can request early access. Model options, source data and any verification work need review for your project.

Public example · April 23, 2026

MITRE and The Weather Company's Weather 1K

MITRE and The Weather Company, 23 April 2026: Weather 1K licensed for GRAF R&D; announcement names AceCAST among modelling tools used to develop it. Described as CONUS 1 km, ten-minute, about 7 PB, sea level to 62,000 ft. Not a TempoQuest resale of Weather 1K.

Read the announcement · Our article

How it works

Agree the details before the scale.

  1. Scope

    Share the use case and what you know. Agree the spec, rights, quote and a technical point of contact.

  2. Sample

    Request a small sample. Agree its scope and checks before planning a large order.

  3. Full delivery

    Once terms and sample checks are agreed, set the delivery plan and handover requirements.

Priced on request. Domain size, history, cadence, verification and delivery shape the quote. Samples and delivery are subject to agreed scope and terms.

We review every request.