Most studies about the impact of convection at large scale has been based either on observations or on cloud resolving idealized simulations (CRM). In the CRM approach, the onset of convection is not treated, as artificial initial perturbations is generally used. Also the two-way interactions between the large scale flow and convection are not considered, due to the limited extension of the domain of simulation, or to the use of periodic lateral boundary conditions.

Recently Diongue et al. (2001) succeeded to explicitly simulate the whole life cycle of an observed sahelian squall-line and its interactions with the large scale flow thanks to the use of grid-nesting technique. It offered us the opportunity to study the impact of the convection by avoiding some shortcomings of previous idealized CRM simulations. We first present our method to estimate the apparent sources due to convection from the CRM allowing to account for horizontal transports that were neglected up to now. This diagnostic is then applied to the above simulation at different periods of the system life cycle to evaluate the sources that should be parameterized by a "perfect" convective scheme for different resolution ranging from 50 km to 300 km. Main results concern the importance of the rain evaporation for this sahelian system and of the transport by convective eddies. Due to the large size of this well structured convective system, the horizontal transport by eddies can not be neglected especially at low and upper levels. This major result applies to all variables; temperature, humidity and momentum. The sources of momentum is significant and mainly acts in the direction of propagation of the squall line but depends on its position inside the large scale mesh. At the system rear, it allows to explain the intensification of the African Easterly Jet (up to 3m/s/hr for a 300 km scale) and the lowering of its altitude. In the upper troposphere the momentum apparent source due to convection is even stronger (up to 6m/s/hr) and coherent with an acceleration and a deceleration of the Tropical Easterly Jet ahead and behind the system respectively. Physical interpretation of these results and their implication on the convection parameterization problem will be discussed at the conference.