Key findings

• Decadal predictions significantly outperform longer-term climate projections over both five- and ten-year timescales, particularly when predicting winter rainfall that significantly differs from the local average over Southern Europe, amongst other regions.
• The improved accuracy of these predictions can largely be attributed to the fact that decadal predictions are initialized with information about the observed state of the climate, whereas long-term climate projections are not. 
• The decadal predictions better capture the North Atlantic Oscillation (NAO) and subpolar North Atlantic sea surface temperatures (SPNA-SST), which have a significant impact on European climate, including wintertime precipitation variability. 
• An innovative new approach further improves on the predictive ability of decadal predictions, which, with further refinement, could provide even more accurate forecasts. This would be particularly beneficial for sectors such as water management and agriculture. 

Background

The Mediterranean region is highly sensitive to climate change, with higher warming than the global average and decreased precipitation expected to continue through the 21^st century. This poses risks to water resource management, agriculture, adaptation planning, increasing the value of reliable and accurate forecasts for stakeholders and decision-makers.  

Decadal climate predictions bridge the gap between seasonal forecasting and long-term climate projections. They look up to ten years ahead and include observations of the current climate, which improves their ability to model near-term changes. In comparison, long-term climate projections are designed to run for multiple decades until the end of the century and focus on the climate’s response to slower-changing factors such as greenhouse gas concentrations

This study compares the ability of these two types of modelling approaches to predict winter rainfall over the Mediterranean, exploring how the rainfall differs from the regional average over two forecast periods of five and ten years. It demonstrates the value of decadal predictions for near-term climate change, which incorporate observed climate information through initialization.

The study also introduces a novel new modelling approach combining dynamical and statistical methods based on the observed relationship between precipitation and the NAO and SPNA-SST. The NAO refers to pressure variations across the North Atlantic, a known predictor of seasonal temperature and precipitation across Europe. 

Why these findings matter

Accurate forecasts of conditions that vary from the regional average are fundamentally important in the creation of climate adaptation policies and for stakeholders and decision-makers in at-risk sectors such as agriculture and water-resource management. Utilizing the most suitable model type for these forecasts helps provide the most reliable and accurate information to inform decisions. 

With further research into rainfall processes, the innovative new model proposed in this study has strong potential to enhance the accuracy and reliability of forecasts in the Mediterranean. 

Read the full study in Environmental Research Letters

Reference: Nicolì, D., Gualdi, S. and Athanasiadis, P. J., (2025) Decadal predictions outperform climate projections in forecasting Mediterranean wintertime precipitation. Environmental Research Letters, 20(3), 034034.

Keywords: decadal climate predictions, forecast quality assessment, CMIP6 multi-model ensemble, climate projections, Mediterranean, rainfall