Southern Africa is a water-stressed and flood-prone region widely impacted by climate and land use change. However, data scarcity and declining hydrometeorological networks cause hydrologic predictions in the region to be highly uncertain. We present an approach to constrain model uncertainty for gauged and ungauged watersheds in southeastern Africa. The work focuses on the Limpopo Basin, where transboundary water issues, combined with high risks for floods, droughts and climate change impacts, create a vital need for hydrologic predictions to improve flood forecasting and water resource management. We combine the use of satellite-based precipitation data with a top-down approach for parsimonious model selection in a forecasting framework including uncertainty. Model parameterizations in ungauged watersheds are constrained based on regionalized relationships between hydrologic response indices and watershed characteristics (physical and hydroclimatic). Some initial results are presented that demonstrate the potential of this approach for hydrologic model identification and uncertainty reduction in both gauged and ungauged watersheds in data sparse regions.