Regional flood frequency model for Namibia and Zimbabwe.

Regionalization was introduced to hydrology in the near past. Heterogeneity among the sites in the region, use of historical data and correlation between sites in a region are the limitations of this concept. This study tries to minimize their effects and assesses the significance of some of the limitations. Namibia and Zimbabwe are delineated into homogenous regions for hydrological studies. In order to achieve this objective physiographic, climatic and drainage characteristics are used to preliminarily identify homogenous regions. Using statistics obtained from the available annual maximum series these regions are tested for homogeneity. These tests are mainly based on regional weighted average L-moments. Based on this study Namibia is divided in to 3 and Zimbabwe in to 6 homogenous regions. Before starting the analysis, the data were checked for temporal and spatial independence. Those sites showing temporal dependence were completely abandoned from further analysis subsequent analysis was done for both spatially. A procedure, based on a combination of description and predictive ability tests, adopted for selecting a suitable flood frequency model with the corresponding method of parameter estimation is described. This procedure has resulted in pearson type 3 with probability weighted moments to be the preferred distribution and method of parameter estimation for all established regions respectively.