Atlantic–Pacific winter warming as an early indicator of major hurricane activity over the North Atlantic

Seasonal forecasting of major hurricane (MH) activity in the North Atlantic remains challenging, particularly at extended lead times. A key obstacle is the ENSO predictability barrier, which limits forecast skill for outlooks issued in spring — well before the August–September–October peak season. As Atlantic major hurricanes increase in frequency and their downstream impacts increasingly affect European weather through extratropical transition, improved seasonal forecasting carries relevance well beyond the Atlantic basin. This study addresses this challenge by leveraging Causal-Effect Networks (CENs) to identify robust boreal winter teleconnection precursors of seasonal MH frequency, and by incorporating both local and remote sea surface temperature (SST) drivers into a statistical forecast framework.

Using the PCMCI+ algorithm applied to ERA5 reanalysis and HURDAT2 hurricane data over 1980–2023, we identify February SST anomalies in the Atlantic and central North Pacific as the two key predictors of seasonal MH counts. Unlike traditional correlation-based approaches, causal analysis allows us to isolate robust, potentially causal links while filtering out spurious associations arising from confounding variables and autocorrelation.

Our multiple linear regression model, trained on 1980–2009 data and independently tested on 2010–2023, achieves a correlation of 0.70 between predicted and observed MH counts during the test period — substantially exceeding the skill of operational April forecasts and comparable to June forecasts issued by major forecasting agencies. Crucially, these forecasts are available as early as March.

We demonstrate that both precursors influence hurricane season conditions in the Main Development Region (MDR) through distinct but complementary interseasonal teleconnection pathways. Atlantic February SST anomalies drive warming through a wind–evaporation–SST feedback linked to the negative phase of the North Atlantic Oscillation, reducing trade winds and vertical wind shear while enhancing MDR SSTs and low-level convergence. North Pacific February SST anomalies initiate a Pacific Meridional Mode-like feedback, generating easterly wind anomalies in the tropical Pacific that further reinforce a La Niña-like SST pattern by hurricane season — ultimately creating favorable thermodynamic and dynamic conditions in the MDR.

These results demonstrate how cross-basin teleconnections rooted in boreal winter SST anomalies shape the subsequent hurricane season environment, and how causal analysis can serve as a valuable tool for identifying physically meaningful predictors for improved early seasonal hurricane forecasts.

This work is published open access: https://doi.org/10.1002/qj.5048