cGRHyMoLAP is a new continuous rainfall–runoff model that describes catchment storage and river flow within a compact two-state dynamical system, without intermediate calculations. Tested across 623 British catchments, it preserves the behaviour of GRHyMoLAP while improving runoff-generation timing and reducing computational cost. The results show that continuous modelling provides a simple, efficient and interpretable basis for representing rainfall–runoff dynamics.

We studied how ice processes along the Vistula River in Central Europe have changed over time, using river records, satellite images and weather data. Ice has become less common as winters have warmed, but the changes differ along the river. Dams, pollution and other human activities also influence river ice processes. Understanding future changes therefore requires considering both climate and local human impacts.

Estonia's rapakivi intrusions are largely concealed beneath younger sedimentary cover. Using gravity and magnetic data, we reconstructed the three-dimensional architecture of Märjamaa–Kloostri, Naissaare and Neeme. The models reveal contrasting internal geometries and persistent deeper magnetic anomalies, despite a shared regional tectonic setting. No continuous connection between the intrusions is resolved, indicating that local emplacement conditions likely influenced their development.

Tracking glacier change requires frequent and detailed measurements of their surface. We use machine learning to combine complementary satellite observations and produce glacier elevations every three months at 100 m resolution across the Canadian Arctic. The resulting elevations are substantially more accurate than the original satellite measurements and reveal detailed changes over time, including the evolution of glacier surges.