Precipitation modeling over southern Africa

Southern Africa presents a challenge to atmospheric modeling and, in particular, precipitation modeling.Steep topography combined with low level moisture advection results in orographic rainfall in the coastal and escarpment areas.The influence of mid-latitude lows, sub-tropical troughs, coastal lows and cut-off lows often result in extreme rainfall events.Large areas of the interior of Southern Africa are considered arid or semi-arid and as a result are sensitive to extreme rainfall events.Loss of lives, soil and general damage to property and infrastructure is a common result of extreme rainfall events in the area.The economy of the area is highly dependent on agriculture and therefore also on rainfall.

This research centres on the use of regional atmospheric models to study extreme precipitation events over southern Africa.The use of regional atmospheric models over Southern Africa has been fairly limited in the past, largely due to a lack of computational resources as well as the inadequacy of existing regional models to cope with the extreme topography of the region.Past modeling attempts using low resolution hydrostatic models have produced rainfall estimates significantly higher than reality over the steep topography of the escarpment area.High resolution non-hydrostatic models seem to offer a solution to this problem.

The Mesoscale Model version 5 (MM5) is highly suited to atmospheric modeling in this region.It is a flexible model that can be run at almost any spatial resolution with any number of vertical levels.It can be run in non-hydrostatic mode and includes a large number of physical parameterization options including a number of precipitation schemes.In addition, compilation options allow the model to execute on clusters of cheap Intel based workstations.This option is particularly attractive in the South African atmospheric modeling environment where computational resources are extremely limited.Other options such as multiple nests and even moving nests can also be used if it is thought these will increase the accuracy of the model.

Any regional atmospheric model requires atmospheric boundary forcing data.Clearly the more accurate this boundary data, the more accurately the model will be able to represent reality.The Medium Range Forecast re-analysis data produced daily by NOAA is available globally at a resolution of 1o.This data set has shown itself to accurately represent the southern African region and as result has been chosen to provide forcing data for the model runs.When MRF archive data is not available, NCEP re-analysis data is used.The research is focused on the modeling of past extreme rainfall events. A number of events are modeled and a comparison is done between the modeled precipitation and the observed precipitation during that event.Also the atmospheric dynamics generated by the model are visualized and compared to the "observed" reality of the re-analysis data.A better understanding of the dynamics of these events is thus obtained and hence our forecasting skill is improved.