This poster presentation focuses on the spring to summer transition of the West African monsoon through diagnostics from the NCEP/NCAR (USA) reanalyses, 8 AGCM (ARPEGE-Climat, France) long-term simulations over the period 1968-1998 and in-situ observations from the CRU (UK) dataset. The selected NCEP variables are the monthly temperature, humidity, geopotential and the 3 wind components on 11 standard isobaric levels in the troposphere (1000hPa-100hPa), along with the radiative (long and short wave) and heat (sensible and latent components) fluxes at the surface and soil moisture (surface and deep reservoirs). These atmospheric diagnostics refer mainly to the spring /summer transitions occurring before the abnormally wet and dry July-September rainy seasons using several simulated and observed rainfall indexes. The results show that horizontal (mostly meridional) energy gradients in the boundary layer control significantly the main spring to summer changes in monsoon dynamics as tested with selected indicators in terms of convergence/divergence of the humidity fluxes, vertical ascent, shallow / deep convection partition and of course regional rainfall indexes. The course of events can be summarized as follows: 1- At the beginning of spring, the Moist Static Energy (MSE) meridional gradients are the strongest over the continent and are maintained by sensible heating and cooling. Then they relax and the monsoon air mass is transported northward into the continent. 2- These gradients are mainly controlled at regional scale by the seasonal evolution of precipitation and soil moisture fields . 3- Before an abnormally wet (dry) rainy season, negative (positive) rainfall and soil moisture anomalies are normally observed over Sudan-Sahel: this maintains stronger than normal MSE and humidity gradients between the Sahelian and Guinean areas in late spring 4- The MSE meridional gradients tend to be significantly stronger (weaker) and to relax later (earlier) but more (less) intensely in summer. This increase (decrease) monsoon dynamics as tested through the aforementioned atmospheric indicators and rainfall indexes.

This could explain why, for improving rainfall predictions over West Africa, numerical simulations must prescribe an interactive (dynamical) soil moisture all along the time integration while statistical approaches have just to take into consideration the atmospheric MSE contents in spring.