Probabilistic measures afford fair comparisons of AIWP and NWP model output

TL;DR

The paper introduces the potential continuous ranked probability score (PC), a new measure for fair comparison of AI and NWP weather prediction models, demonstrating its effectiveness on WeatherBench 2 data.

Contribution

It proposes a novel, invariant measure (PC) for comparing deterministic model outputs after statistical postprocessing, enabling fair and meaningful model evaluations.

Findings

GraphCast outperforms ECMWF HRES in PC measure.

PC aligns well with operational ensemble CRPS.

The method enables fair comparisons without predefined loss functions.

Abstract

We introduce a new measure for fair and meaningful comparisons of single-valued output from artificial intelligence based weather prediction (AIWP) and numerical weather prediction (NWP) models, called potential continuous ranked probability score (PC). In a nutshell, we subject the deterministic backbone of physics-based and data-driven models post hoc to the same statistical postprocessing technique, namely, isotonic distributional regression (IDR). Then we find PC as the mean continuous ranked probability score (CRPS) of the postprocessed probabilistic forecasts. The nonnegative PC measure quantifies potential predictive performance and is invariant under strictly increasing transformations of the model output. PC attains its most desirable value of zero if, and only if, the weather outcome Y is a fixed, non-decreasing function of the model output X. The PC measure is recorded in the unit of the outcome, has an upper bound of one half times the mean absolute difference between outcomes, and serves as a proxy for the mean CRPS of real-time, operational probabilistic products. When applied to WeatherBench 2 data, our approach demonstrates that the data-driven GraphCast model outperforms the leading, physics-based European Centre for Medium Range Weather Forecasts (ECMWF) high-resolution (HRES) model. Furthermore, the PC measure for the HRES model aligns exceptionally well with the mean CRPS of the operational ECMWF ensemble. Across application domains, our approach affords comparisons of single-valued forecasts in settings where the pre-specification of a loss function -- which is the usual, and principally superior, procedure in forecast contests, administrative, and benchmarks settings -- places competitors on unequal footings.