Ensemble Forecast CRPS Values for 1-28 Day Lead Times

Robert Moss published ensemble forecast log-transformed CRPS values on figshare in April 2026. The dataset contains mean CRPS values aggregated for lead times of 1-7 days, 8-14 days, 15-21 days, and 22-28 days across multiple jurisdictions. The data indicates forecast performance decreased as lead time increased.

Use Cases

• Benchmarking probabilistic forecast models based on CRPS values across different lead times.
• Analyzing the relationship between forecast error and lead time based on the described performance degradation.
• Comparing forecast performance across different jurisdictions as indicated by the per-jurisdiction and aggregate values.

Strengths

• Data is structured with clear temporal aggregation into four weekly lead-time windows.
• Includes a cross-jurisdiction aggregate row for overall performance assessment.
• Released under a permissive CC-BY-4.0 license.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.

Provenance

Source

Robert Moss via figshare.

Collection Method

Likely derived from ensemble forecast model outputs.

Time Range

Lead times cover 1 to 28 days.

Freshness

Last updated 2026-04-22 18:00:19; freshness should be verified.

Geography

Multiple jurisdictions are referenced, but specific locations are not named.