Datasets

Goals and principles for datasets used to compare spatial verification methods.

Bridging The Gap datasets

Comparison datasets for meta-verification

This collection supports meta-verification: comparing spatial verification methods for deterministic and ensemble forecasts. It aims to make existing datasets easier to access, propose new datasets, and help users interpret results and understand the role of each dataset or experiment.

Schematic of the comparison datasets organization
Organization of the comparison datasets.

Principles

  • Surface variables: temperature, precipitation, and wind.
  • Both deterministic and ensemble forecasts.

  • Reuse existing datasets whenever possible.
  • Avoid external challenges related to observation or analysis quality and common grids.
  • Seek temporal and spatial overlap between real-world datasets.
  • Keep data size manageable.

Original dataset work

2026

  1. arXiv
    A dataset of one-dimensional idealized probabilistic fields
    2026

Selected datasets

Ensemble

Type Dataset Description
Synthetic 1D idealized probabilistic fields 1D idealized probabilistic fields
Real-world LAM-EPS Regional NWP
Real-world TIGGE Global

Deterministic

Type Dataset Description
Synthetic New geometric test cases Idealized cases
Synthetic Idealized test cases —
Perturbed Fake Test Cases Modified precipitation observations
Real-world WP-MIP Global; NWP vs AIWP vs hybrid