Subseasonal predictability of cold spells in the central United States

Abstract Wintertime cold spells in the central United States pose significant socioeconomic and environmental risks, yet actionable subseasonal forecasts beyond 2 weeks remain a challenge. Here, we evaluate central United States cold spell predictability at weeks 3–4 using reanalysis and a suite of subseasonal model hindcasts, focusing on the physical mechanisms underlying forecast success and failure. We find that prediction skill declines rapidly beyond week 2, but forecasts retain sufficient resolution and sharpness to identify precursor signals linked to performance. Skillful and confident week 3 forecasts are preceded by a weakening stratospheric polar vortex (SPV), persistent high‐latitude blocking, and robust stratosphere‐troposphere coupling. In contrast, confident false alarms are often associated with unrealistic North American winter weather regime transitions and weak or absent stratospheric signals, particularly under strong SPV conditions. By stratifying confident forecasts by SPV state at initialization, we demonstrate that weak SPV conditions enhance forecast reliability, whereas strong SPV states can degrade it. These findings reveal that the vertical structure and timing of SPV variability play a critical role in modulating extended‐range predictability of central United States cold spells, and capturing both SPV variability and weather regime persistence is essential for improving weeks 3–4 cold spell forecasts. Overall, the study offers a mechanistic standpoint and guidance for interpreting and developing more reliable extended‐range predictions of extreme winter cold in the central United States.