A Data-Driven Model of the Geomagnetic Storm Recovery Phase

The recovery phase of geomagnetic storms poses a significant challenge for accurate space weather forecasting. Traditional models based on a constant decay rate fail to capture its prolonged and variable duration, leading to uncertainties in predicting impacts on technological systems. Here, we use the high-resolution SuperMAG 𝑆⁢𝑀⁢𝑅 index and a stochastic Langevin approach to reveal the intrinsic multi-timescale nature of the recovery phase. We demonstrate that the decay timescale is not constant but is a function of the storm's instantaneous intensity, a feature that cannot be captured by standard empirical models. From this dependence, we derive a novel evolution equation for the ring current that accurately describes the recovery as a scale-free power-law decay. This data-driven model provides a significant improvement in reproducing the observed recovery profiles over previous formulations and represents a crucial step toward developing more physics-based and reliable forecasts for the conclusion of geomagnetic storms, with direct benefits for operational space weather services.