Land-atmosphere Coupling Sensitivity to GCMs Resolution: A Multi-model Assessment of Local and Remote Processes in the Sahel Hotspot

Omar V. Müller, Pier Luigi Vidale, Benoît Vannière, Reinhard Schiemann, Retish Senan, Reindert J. Haarsma, Johann H. Jungclaus
2020 Journal of Climate  
Land-atmosphere interactions are often interpreted as local effects, whereby the soil state drives local atmospheric conditions, and feedbacks originate. However, non-local mechanisms can significantly modulate land-atmosphere exchanges and coupling. We make use of GCMs at different resolutions (low ∼1° and high ∼0:25°) to separate the two contributions to coupling: better represented local processes versus the influence of improved large-scale circulation. We use a two-legged metric,
more » ... ed by a process-based assessment of four CMIP6 GCMs. Our results show that weakening, strengthening and relocation of coupling hotspots occur at high resolution globally. The northward expansion of the Sahel hotspot, driven by non-local mechanisms, is the most notable change. The African Easterly Jet's horizontal wind shear is enhanced in JJA, due to better resolved orography at high resolution. This effect, combined with enhanced easterly moisture flux, favours the development of African Easterly Waves over the Sahel. More precipitation and soil moisture recharge produce strengthening of the coupling, where evapotranspiration remains controlled by soil moisture, and weakening where evapotranspiration depends on atmospheric demand. In SON the atmospheric influence is weaker, but soil memory helps to maintain the coupling between soil moisture and evapotranspiration, and the relocation of the hotspot at high resolution. The multi-model agreement provides robust evidence that atmospheric dynamics determines the onset of land-atmosphere interactions, while the soil state modulates their duration. Comparison of precipitation, soil moisture and evapotranspiration against satellite data reveals that the enhanced moistening at high resolution significantly reduces model biases, supporting the realism of the hotspot re-location.
doi:10.1175/jcli-d-20-0303.1 fatcat:fq3kbz3jvrhlbexyz3xee4tyta