Land surface processes relevant to subseasonal-to-seasonal prediction

Subseasonal-to-seasonal is a “sweet spot” for land-surface feedbacks on the atmosphere; those feedbacks are most impactful on forecasts at subseasonal time scales. The role of land-surface interactions with the atmosphere is discussed in terms of the physical processes as currently understood and the implications for improved prediction. The potential for improvement stems from predictability provided by relatively slowly varying land-surface states like soil moisture, snow cover, and vegetation. A history of the evolution of land-surface models at operational forecast centers is also provided, along with a discussion of land-surface data assimilation to initialize forecast models and sources of data for global assimilation. We conclude that significant improvements in forecast skill can be made in the short term by treating land and atmosphere as a coupled system throughout the model development process and by better applying available observations to calibrate, validate, and initialize land-surface states.