A Multivariate Characterization of Land–Atmosphere Feedback Regimes Using Flux Site Observations

Abstract Land surface water and energy balances are important for understanding the coupling between land and atmosphere in Earth’s climate system. This study employs principal component analysis (PCA) and a variation on Granger causality to investigate land–atmosphere (LA) interactions among soil water content (SWC), surface latent heat flux (LE), sensible heat flux ( H ), and net radiation (RAD) at flux observation sites across seasons. The time scales of variability for each surface variable are key: For daily data, H is a crucial variable in LA interactions across all seasons, while LE becomes important for LA interactions mainly during warm seasons. SWC, despite its direct link to the surface water budget, appears less influential than energy at daily time scales, as its variability is mainly at longer time scales. Effects of RAD vary seasonally, being more significant in energy-limited regimes during spring and summer. The familiar SWC:LE correlation metric is expanded into a two-dimensional LA coupling matrix by including SWC: H relationships. This matrix facilitates fine classification of LA feedback providing a clearer understanding of water- and energy-limited regimes. Spatial analysis of observation sites shows significant LA interactions mainly in midlatitudes, influenced by solar radiation. The global distribution of LA feedback regimes underlines the complexity of defining such regimes, which vary according to geographic location, local climate, and land/vegetation types. Meanwhile, climate models and reanalyses fail to capture many observed aspects of LA interactions. Such insights are vital for enhancing the predictability of climate models and comprehending the intricate interplay between different surface conditions in LA interactions.