Abstract. Land–atmosphere (L–A) coupling can play a crucial role for subseasonal-to-seasonal (S2S) predictability and prediction. When coupling is strong, L–A processes and feedback are expected to enhance the system's memory, thereby increasing the predictability and prediction skill. This study evaluates the subseasonal prediction of ambient surface air temperature under conditions of strong versus weak L–A coupling in forecasts produced with NASA's state-of-the-art Goddard Earth Observing System (GEOS) S2S forecast system. By applying three L–A coupling metrics that collectively capture the connection between the soil and the free troposphere, we observe improved prediction skill for surface air temperature during weeks 3–4 of boreal summer forecasts across the Midwest and northern Great Plains, particularly when all three indices indicate strong L–A coupling at this lead time. The prediction skill indeed increases as more indices show strong coupling. The forecasts with strong L–A coupling in these regions tend to exhibit sustained warm and dry anomalies, signals that are well simulated in the model. Overall, this study highlights how better identifying and capturing relevant L–A coupling processes can potentially enhance prediction on S2S timescales.
Lim et al. (Fri,) studied this question.