Abstract The Integrated Forecast System (IFS) of the European Centre for Medium‐Range Weather Forecasts (ECMWF) includes both a hydrostatic and a non‐hydrostatic option. The hydrostatic option is used in all current operational forecast configurations, while the non‐hydrostatic option is used for research purposes. Testing of the non‐hydrostatic option at kilometre‐scale resolutions has revealed stability problems that so far have hindered a detailed in‐depth evaluation and comparison with its hydrostatic counterpart. Recently, significant efforts have been made to enhance the stability of the IFS non‐hydrostatic dynamical core, enabling stable and accurate non‐hydrostatic kilometre‐scale simulations. In this article, we describe and analyze these improvements, demonstrating their impact through tests on several case studies at grid spacings as fine as 1.4 km. The forecast skill of the revised non‐hydrostatic model is evaluated and compared with that of the hydrostatic model. Our results show that, while non‐hydrostatic effects are captured as the horizontal grid resolution approaches 1 km, their impact on forecast skill at resolutions of 3–5 km is relatively small in short‐ and medium‐range weather forecasts and mostly confined in the lower stratosphere over the Tibetan Plateau during midwinter.
Vivoda et al. (Sun,) studied this question.