Surface, boundary layer and cloud couplings over land : evaluation of CMIP5 climate simulations over West Africa

The surface energy budget, boundary layers (BL) and clouds display large geographic, annual and diurnal differences over West Africa. During the monsoon, in the low levels, the atmospheric balance appears to be strongly shaped at large scale by deep convective processes within the ITCZ (intertropical convergence zone), by nocturnal advection associated with the monsoon flow on the northern warmer side of the ITCZ and by the cloud cover prevailing on the southern cooler side. These various balances are coupled with distinct low-level thermodynamics and diurnal cycles of boundary layers which are well framed by consideration of the surface temperature and thermodynamic diagrams (Gounou et al. BLM 2012 (G12), Couvreux et al. BLM 2014 (C14)). Away from the coast, as the surface becomes warmer, the specific humidity in the low levels, which is bounded by saturation, increases at first, but north of the ITCZ, where the boundary layer can display a strong daytime drying, the specific humidity overall decreases. C14 showed with a one-dimensional model that bias in simulated temperature and water vapor consistently change signs across these contrasted climates, and that biases in specific humidity cannot be simply related to biases in cloud amount and convection.

Here, the simulation of these regimes in less-constrained CMIP5 (coupled model intercomparison project phase 5) AMIP (atmospheric MIP). The differences in surface-boundary layer couplings are addressed via an analysis of the CFMIP cfSites diagnostics. These correspond to surface and atmosphere (on the model vertical levels) energy budgets (temperature and water vapor equations) provided over 30 years at high frequency (30 min) for selected grid points. More precisely, these points are located in the vicinity of observational sites where surface fluxes and/or soundings data have been collected, with a total of 10 grid points that sample the south-north climatic transect of West Africa. However, direct point to point comparisons of observations and simulations are sometimes dominated by differences in the latitudinal position of the monsoon which are of limited interest for our purpose. Therefore, the simulations are studied using the surface temperature frame of G12, C14, and the results are also further sorted according to additional parameters such as season-mean precipitation amount. Simple thermodynamic diagrams are used to analyze BL budgets and to explore and evaluate surface-BL-clouds coupling in simulations.

The results underline modeling issues which go well beyond a problem of timing of deep convection over land. They further point to the need of a more accurate simulation of surface properties and state, but also aerosols and clouds. Indeed, even if the cloud cover may be sometimes less over land than over ocean in the Tropics, its impact on the low atmospheric layers and convection occurs on much smaller time scales, during daytime, via distinct mechanisms of couplings with surface processes. Overall, the results point to the importance of 'local' physical processes on these couplings.