Improving Seasonal Forecasts: Evaluating Synoptic Pattern Predictability Using Weather Type Classification

Seasonal climate forecasts offer valuable insights into expected climate conditions up to several months in advance, supporting decision-making in sectors such as agriculture, energy, and disaster management. However, seasonal forecasts still struggle to capture the spatial and temporal variability of atmospheric circulation, particularly in regions prone to climate extremes. Improving their reliability requires better methods for assessing how well they reproduce key features of synoptic-scale dynamics. Classifying synoptic patterns allows for a more comprehensive statistical assessment of atmospheric circulation, enabling the identification of dominant weather regimes and their evolution over time.

In this study  we developed an updated approach to classify weather patterns in Eastern Mediterranean using z500 geopotential heights and vertical velocity data. The classification method identifies five anticyclonic and seven cyclonic types. Our findings indicate that during the cold season, cyclonic and anticyclonic weather types occur at similar frequencies, whereas anticyclonic types dominate in the summer. To evaluate the ability of seasonal forecast models to predict synoptic patterns up to three months in advance, we apply the classification to seasonal hindcast simulations produced by the state-of-the-art Advanced Research WRF (WRF-ARW) model for the period 1981-2016. The classified weather types from the seasonal model outputs are compared against those derived from ERA5 to assess forecast skill, biases, and limitations in capturing large-scale atmospheric circulation. Preliminary findings indicate that synoptic conditions in the study region can be reliably predicted three months in advance (lead time 3), with the seasonal model achieving an accuracy rate of over 75% in estimating the likelihood of anticyclonic or cyclonic conditions. This classification-based approach provides a process-oriented  alternative for evaluating seasonal forecast performance.

Acknowledgments

The work was supported by PREVENT project. This project has received funding from Horizon Europe programme under Grant Agreement No: 101081276.