While certain atmospheric flow regimes such as blocking patterns can degrade forecast skill, multiple recent studies demonstrate that long-term weather prediction and extended-range forecasting models achieve significant predictive accuracy through advanced dynamical methods, data assimilation, and machine learning techniques.
The retrieved literature contains multiple studies confirming the effectiveness of numerical weather prediction and machine-learning models for medium-range, extended-range, and seasonal forecasting (e.g., papers 0, 4, 9, 10). While paper 8 highlights flow-dependent limitations during specific blocking regimes, the weight of evidence across the set supports the overall effectiveness of these models, with certain limitations and dependencies noted.