Examining Entropic Unbalanced Optimal Transport and Sinkhorn Divergences for Spatial Forecast Verification

View Full Paper

Ask AI

Bookmark

Authors

Discussion

Analysis shows Sinkhorn divergence effectively verifies precipitation forecasts, highlighting its potential in spatial verification methods.

Sinkhorn divergence provides a reliable method for verifying precipitation forecasts, reducing phase error issues.
Testing demonstrated that this method aligns well with expert assessments, confirming its robustness in practice.
Utilizing entropic regularisation allows for efficient approximation of transport costs, facilitating the mapping of distributions.
The approach enhances understanding of forecast errors, offering novel visualizations and robust summary scores for evaluations.

Francis et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark

Analysis shows Sinkhorn divergence effectively verifies precipitation forecasts, highlighting its potential in spatial verification methods.