Evaluation of the West African monsoon vertical cloud structure in high-resolution models using CloudSat and CALIPSO

The West African summer monsoon (WAM) is an important driver of the global climate and locally provides most of the annual rainfall. A clear vertical structure of individual cloud layer occurrence is invaluable to forecasters and modellers to improve the understanding of the WAM. In this presentation, four years of data from the CloudSat profiling radar and CALIPSO lidar are used to create a composite zonal mean vertical cloud and precipitation structure for the WAM. For the first time, the near-coincident vertical radar and lidar profiles allow for the identification of individual cloud types from optically thin cirrus and shallow cumulus to congestus and deep convection. A clear diurnal signal in zonal mean cloud structure is observed for the WAM, with deep convective activity enhanced at night producing extensive anvil and cirrus, whilst daytime observations show more shallow cloud and congestus. A layer of altocumulus is frequently observed over the Sahara at night and day, extending southward to the coastline, and the majority of this cloud is shown to contain supercooled liquid in the top. The occurrence of deep convective systems and congestus in relation to the position of the African easterly jet is studied, but only the daytime cumulonimbus distribution indicates some influence of the jet position.

The results from CloudSat observations are used to evaluate the cloud structures in a 40-day simulation of the UK Met Office Unified Model run at 4 km resolution over a limited area covering West Africa. Using the CloudSat simulator to obtain radar reflectivities from the modeled cloud fraction and mixing ratio, a like-with-like comparison with the observations is presented, highlighting the shortcomings of the model in simulating the cloud structures of the West African monsoon. Further comparisons with model runs at different resolutions indicate the impact of increasing vertical and horizontal resolution as well as the model sensitivity to cloud parametrization.