On the multiscale nature of soil moisture-temperature couplings: the role of seasonality, causation and non-linear feedbacks in land-atmosphere interactions (Invited)

Land-Atmosphere (L-A) interactions, their strength and directionality, are one of the main sources of uncertainty in current climate modeling, with strong implications on the accurate assessment of future climate variability and climate change impacts. Beside from the scarcity of direct observations, major uncertainties derive from the inherent complexity and nonlinearity of these interactions, and from their multi-scale character. Statistical analysis of L-A couplings is traditionally based on linear correlation methods and metrics. However, these approaches are not designed to detect causal connections or non-linear couplings and they poorly perform in presence of non-stationarities. Additionally these methods assess L-A couplings essentially in the time domain, despite the fact that L-A dynamical drivers can act simultaneously over a wide range of different space and time scales. This talk explores the multi-scale nature of L-A interactions, through the example of soil moisture-temperature couplings and soil-moisture memory effects. In several regions of the world, soil moisture can have a dampening effect on temperature due to evaporative cooling. By using spectral decomposition techniques and both newly developed satellite based products and re-analysis, we analyze the contribution of different time scales to the build-up of global soil moisture-temperature coupling hot spots, addressing at the same time the role of seasonality, causation and non-linear feedbacks in land-atmosphere interactions. Finally we focus on the role of fine (sub-monthly) time scales and their interplay with the seasonal scales.