dataset · Global

aardvark-weather

Hugging Face dataset av555/aardvark-weather

Claim this listing

Represent aardvark-weather? Send a claim for review so we can discuss listing updates. Claims do not automatically grant ownership or verification.

Data offered docs-verified: AMSU-A (gridded satellite data) · AMSU-B/MHS (gridded satellite data) · HIRS (gridded satellite data) · ASCAT (gridded satellite data) · GRIDSAT (gridded satellite data) · IASI-PCs (gridded satellite data) · HadISD (station observations) · ICOADS (ship observations) · IGRA (upper atmosphere observations) · Sample timeslice for model demo runs · ML-ready training dataset 2007-2019
Coverage docs-verified: 2007-2019
Access docs-verified: Public Hugging Face dataset released under CC BY-NC-ND 4.0 (non-commercial, no-derivatives); no auth listed in ENTRY

Use cases llm-inferred: Non-commercial academic benchmarking of ML weather models against historical station and satellite observations · Running the Aardvark Weather demo on the sample timeslice to test model ingestion · Research training of observation-based weather models on 2007-2019 inputs (non-commercial only)
Audience llm-inferred: meteorologists · climate scientists · ML researchers

Constraints docs-verified: Prototype datasets are not available in real time · Non-commercial no-derivatives licence, inherited from source datasets; commercial consultancy use is not permitted · Complex patterns of missingness; all missing values are NaN · ERA5 training targets and climatology are not included and must be downloaded separately from WeatherBench · Model weights are in a separate repository (github.com/annavaughan/aardvark-weather-public) · Hugging Face dataset viewer is not available · Training data covers 2007-2019 only; extension to 2025 is in progress and not yet released

Verified 2026-10-15 via HF API: exists; 560 downloads/30d, 18 likes.

Metadata observed 14/10/2026 · research sweep: manual-2026-10-15 (Hugging Face datasets)