South Africa is located in the sub tropics with an elevated plateau which is located approximately 1500 m above mean sea level (a.m.s.l). Every year, snow occurs on the mountains of Lesotho, but on occasions this snow descends to lower elevations which impacts on the livelihood of people. Severe weather originating from extra-tropical weather systems has been well documented in South Africa and yet very little research has been done to predict significant snowfall from these weather systems. The main aim of this research is to identify those weather systems responsible for snow and to understand the processes causing snow to form when these systems occur.
A comprehensive database of significant snowfall events is supplied from 1981 to 2011. The database is subjectively classified into characteristic synoptic patterns. The snow cases are then objectively classified using self-organising maps (SOMs) to obtain synoptic configurations most typically associated with significant snowfall over South Africa. Case studies which aim to explain the synoptic conditions, formation mechanisms as well as critical surface temperature and relative humidity during snowfall events are described. This is done by analysing each case study with respect to synoptic circulations, surface observations, atmospheric soundings, satellite imagery as well as atmospheric thickness.
Conclusions are drawn and critical threshold values of atmospheric thickness, surface temperature and humidity are identified when snowfall occurs.
A methodical snow forecasting decision tree is devised. It takes the synoptic classification of circulation patterns during significant snowfall, atmospheric thickness, height of the freezing level, surface temperature, and relative humidity into account. This process is explained by case studies.
It is recommended that results from this dissertation are made available to weather forecasters in South Africa and that the results are implemented in the operational forecasting environment. Further case study investigations are suggested, taking the mesoscale processes effects into account.