Moisture transport over southern Africa

proThe southern Africa continent is subject to significant inter-annual variability of precipitation, subject in part to teleconnection influences, regional sea-surface temperatures, and antecedent vegetation and soil moisture. A key impact on society is the variability in water resources, compounded by an only moderate degree of skill in seasonal forecasting skill due to limited understanding of the regional moisture sources and the dynamics underlying the seasonal processes.

To begin to address this problem a climatology of the atmospheric moisture transport for the past 40 years is developed based on NCEP reanalysis data. A kinematic 3-d trajectory model is developed to run on a parallel computing PC-cluster, and is driven with 6-hourly data from the NCEP reanalysis for all levels in the atmosphere. For each time step, forward and backward trajectories are calculated over a number of days for all starting points on a 0.25 degree horizontal grid, and for multiple layers in the vertical. The NCEP data for each point on the trajectory is derived with interpolation in 3-d from the coarser 2.5 degree assimilation. The lower boundary of the model is defined by a 5 minute DEM. The focus of the model is to evaluate the long-range transport mechanisms on the synoptic scale (as opposed to the detailed local-scale and the boundary layer). The kinematic model is initially evaluated in terms of sensitivity to different interpolation techniques and time step, using MM5 simulation data as a "high resolution" benchmark boundary condition data set. Subsequently 40 years of NCEP data are used to generate the trajectory climatology.

Given the large data volumes generated, visualization of the trajectories is problematic. To address this, the data are generalized with a non-linear Self-Organizing MAP (SOM). The SOM has a particular ability to identify arch-type modes within a continuum of multi-dimensional data, and allows a relatively easy assimilation of the trajectory data set. The SOM arch-types thus represent generalized 3-d transport patterns, which are then visualized with a number of graphical forms. To address the critical summer convective season, the SOM analysis is initially undertaken only for the December-January-February months. The transport mechanisms are then evaluated in terms of characteristic transport patterns over the greater sub-continent under wet or dry seasonal conditions. These are investigates in light of the phase of ENSO to investigate the teleconnection impact on the regional processe