Patches extracted from ECMWF IFS HRES analysis data on a 0.25° global grid. The dataset consists of 128×128 non-overlapping patches stored as float16 normalized tensors in Parquet files. It was created by meteolibre-dev and last updated on 2026-04-17.
Use Cases
- Training computer vision models for weather pattern recognition based on normalized patch tensors.
- Developing geospatial forecasting algorithms based on high-resolution global grid data.
- Benchmarking model performance on standardized meteorological data patches.
- Analyzing spatial patterns in weather model outputs based on tiled patch structure.
Strengths
- Data is normalized to float16 format, which can reduce storage and memory requirements.
- Patches are consistently sized at 128×128 pixels, providing a standardized input format.
- Source data is from ECMWF IFS HRES analysis, a recognized high-resolution weather model.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
- Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
- Source
- ECMWF IFS HRES analysis
- Collection Method
- Patches extracted and tiled from the original model output.
- Freshness
- Last updated 2026-04-17 18:14:03; freshness should be verified.
- Geography
- Global coverage on a 0.25° grid.