Application of a quantitative precipitation estimation algorithm for the s-band radar at Irene, South Africa.

M. Sc. Eng. University of KwaZulu-Natal, Durban 2014.Flash floods are the number one cause of death and damage with regard to natural disasters in South Africa (Poolman, 2009). Thus, the South African Weather Service (SAWS) and the National Disaster Management Centre (NDMC) embarked on a collaborative project for the implementation of the South African Flash Flood Guidance system (SAFFG) in flash flood prone regions (de Coning & Poolman, 2011). The SAFFG is dependent on accurate precipitation estimates from radars and therefore much emphasis has been placed on the performance of the Quantitative Precipitation Estimation (QPE) fields. Weather radars offer the public efficient means of measuring precipitation remotely. Although the measurements are indirect radar remains the best alternative in capturing the spatial variability associated with precipitation at high temporal and spatial resolutions. A methodology proposed by Chumchean et al., (2006) was selected to be implemented and compared against the existing radar precipitation field of the Gematronik 600S S-band Doppler radar at Irene, South Africa. The methodology proposes a process that includes a rainfall classification algorithm. This algorithm separates convective from stratiform precipitation with the intent to assign different Z-R relations to the two different types of rainfall (Chumchean, et al., 2008). A technique for smoothing accumulations was also included into the algorithm, which is based on optical flow techniques (Bowler, et al., 2004). Reflectivity data from the Irene radar together with in situ rain gauge data within a 300 km radius of the radar location were obtained for the South African summer rainfall season from October 2010 to March 2011 for evaluation of the QPE field. One and twenty-four hour accumulations were compared to the corresponding rain gauge totals and the resulting evaluation scores are compared to the existing precipitation field to determine any improvements. The study showed that by applying specific Z-R relationships to both convective and stratiform precipitation yields better results than using a single relationship only. Smoothing the precipitation with optical flow vectors further decreases the QPE error at both one and twenty-four hour accumulations. Overall the dual Z-R relationship with the optical flow smoothing yields the smallest error and is an improvement from the previous algorithm