The 2025 EPS Excellent Paper Award is granted to the paper by Nishizuka et al. (2021).

Nishizuka et al. (2021) developed an operational solar flare prediction model based on deep neural networks, named Deep Flare Net (DeFN). DeFN provides probabilistic forecasts of solar flares occurring within the next 24 hours, including both binary classifications and predictions of the maximum flare class expected during the forecast window. This paper is distinguished by its originality and long-term impact in two key respects. First, it represents one of the earliest applications of deep learning techniques to operational problems in space science, at a time when such methods were still rarely adopted in this field. By introducing modern neural network architectures into solar flare forecasting, the study played a pioneering role and helped open a new research direction that has since become widely influential. Second, the DeFN model was not developed merely as a proof of concept but was designed and implemented as a fully operational system. The model has been running routinely since January 2019, providing continuous flare forecasts at https://defn.nict.go.jp/. This sustained operational use clearly demonstrates the practical value, reliability, and maturity of the proposed approach, going well beyond typical academic studies. For these reasons—its methodological originality, strong quantitative performance, and demonstrated long-term operational contribution to space weather forecasting—the paper was selected for the 2025 EPS Excellent Paper Award.