We analyze storms over the Northeast Atlantic Ocean and the German Bight and how their statistics change over past, present, and future. We look at data from many different climate model runs that cover a variety of possible future climate states. We find that storms are generally predicted to be weaker in the future, even though the wind directions that typically happen during storms occur more frequently. We also find that the most extreme storms may become more likely than nowadays.

Building numerical models of the Earth is a complex task that scientists and engineers around the world work on. It is important that results are repeatable accurately to help advance science. This study uses a statistical method to reduce errors when comparing two sets of simulations to see if they agree with each other. This approach helps identify if changes made to the model's code result in unexpected or unintended effects.

Many decisions require consistent climate information from seasonal to multi-year timescales. We assess seamless forecasts created by constraining seasonal and decadal predictions and compare them with initialised multi-annual forecasts. Multi-annual predictions provide the highest skill, but constrained forecasts still perform well and offer a low-cost, regularly updatable solution for delivering coherent climate information.