Impact of Adjustments in Surface–Atmosphere Coupling for Model Forecasts in Stable Conditions

• Published in: Weather and forecasting Vol. 40; no. 8; pp. 1371 - 1381
• Main Authors: Kähnert, Marvin, Sodemann, Harald, Remes, Teresa M., Homleid, Mariken
• Format: Journal Article
• Language: English
• Published: Boston American Meteorological Society 01.08.2025
• Subjects: Atmosphere; Atmospheric boundary layer; Atmospheric models; Boundary layers; Cooling; Coupling; Eddy flux; Exchange coefficients; Forecast accuracy; Heat; Heat sinks; Humidity; Inversions; Lower atmosphere; Numerical weather forecasting; Performance evaluation; Prediction models; Richardson number; Stable boundary layer; Stratification; Surface boundary layer; Surface fluxes; Surface layers; Temperature; Tuning; Vertical profiles; Weather; Weather forecasting; Arctic region; Finland
• ISSN: 0882-8156; 1520-0434
• DOI: 10.1175/WAF-D-24-0163.1

The surface–atmosphere coupling is a crucial factor in representing the stable boundary layer in numerical weather prediction models. Different aspects of the surface–atmosphere coupling are, therefore, often subject to model tuning. Here, we investigate the effect of such adjustments to the operational weather prediction models AROME-Arctic and the MetCoOp Ensemble Prediction System (MEPS). Currently, an upper limit on the Richardson number is imposed on the turbulent exchange coefficients in the surface layer. In AROME-Arctic, stable stratification is prohibited from influencing the surface fluxes. In MEPS, a limited range of stability is allowed. We focus on the impacts of these model settings on the vertical structure and evolution of the stable boundary layer by analyzing inversion strengths in the surface layer and in the lower atmosphere. In AROME-Arctic, the surface always stays coupled to the atmosphere, acting as a continuous heat sink that sustains atmospheric inversions. Conversely, in MEPS, the surface layer decouples from the atmosphere, resulting in pronounced surface-layer inversions and a reduction in atmospheric inversion strength, reminiscent of the well-known run-away cooling problem. At the same time, MEPS shows colder surface and 2-m temperatures, reducing the warm bias compared to observations from Sodankylä, superficially providing justification for the tuning step. We argue that evaluating the performance of 2-m temperature alone is insufficient for assessing model performance in the stable boundary layer without a validation of the vertical structure in the atmosphere. Additionally, we emphasize that transparent communication and detailed documentation of model tuning are essential for precise model interpretation and lasting progress in improving complex forecasting systems.