As atmospheric models move to higher resolution and resolve smaller scales, the maximum modeled wind speed also tends to increase. Wave models tuned to coarser wind fields tend to overestimate the wave growth under strong winds. A recently developed semiempirical parameterization of the Charnock parameter, which controls the roughness length over surface waves, substantially reduces the aerodynamic drag of waves in high winds (above a threshold of 30 m s⁻¹). Here, we apply the formulation in a recent version of the wave model WAM (Cycle 4.7), which uses a modified version of the physics parameterizations by Ardhuin et al. (2010, https://doi.org/10.1175/2010jpo4324.1) as well as subgrid obstructions for better performance around complex topography. The new Charnock formulation is tested with wind forcing from NORA3, a recently completed nonhydrostatic atmospheric downscaling of the global reanalysis ERA5 for the North Sea, the Norwegian Sea and the Barents Sea. Such high‐resolution atmospheric model integrations tend to have stronger (and more realistic) upper‐percentile winds than what is typically found in coarser atmospheric models. A 2‐year comparison (2011–2012) of a control run against the run with the modified Charnock parameter shows a dramatic reduction of the wave height bias in high‐wind cases. The added computational cost of the new physics and the reduction of the Charnock parameter compared to the earlier WAM physics is modest (14%). A longer (1998–2020) hindcast integration with the new Charnock parameter is found to compare well against in situ and altimeter wave measurements both for intermediate and high sea states.

A significant wave height of 7 m has been measured five times by the northern Baltic Proper wave buoy in the Baltic Sea, exceeding 8 m twice (2004 and 2017). We classified these storms into two groups by duration and wave steepness. Interestingly, the two highest events exhibited opposite properties, with the 2017 event being the longest storm on record. This storm is also the first where the harshest wave conditions were modelled to occur in the western part of the Baltic Proper. The metrics quantifying the storm's duration and steepness might aid in issuing warnings for extreme wave conditions.

We analyse changes in meteotsunami occurrence over the past century (1922–2014) in the Gulf of Finland, Baltic Sea. A major challenge for studying these short-lived and local events is the limited temporal and spatial resolution of digital sea level and meteorological data. To overcome this challenge, we examine archived paper recordings from two tide gauges, Hanko for 1922–1989 and Hamina for 1928–1989, from the summer months of May–October. We visually inspect the recordings to detect rapid sea level variations, which are then digitised and compared to air pressure observations from nearby stations. The data set is complemented with events detected from digital sea level data 1990–2014 by an automated algorithm. In total, we identify 121 potential meteotsunami events. Over 70 % of the events could be confirmed to have a rapid change in air pressure occurring shortly before or simultaneously with the sea level oscillations. The occurrence of meteotsunamis is strongly connected with lightning over the region: the number of cloud-to-ground (CG) flashes over the Gulf of Finland were on average over 10 times higher during the days when a meteotsunami was recorded compared to days with no meteotsunamis in May–October. On a monthly level, statistically significant differences between meteotsunami months and other months were found in the number of CG flashes, convective available potential energy (CAPE), and temperature. Meteotsunami occurrence over the past century shows a statistically significant increasing trend in Hamina, but not in Hanko.