Estimating Flood Magnitudes of Ungauged Urban Msimbazi River Catchment in Dar es Salaam, Tanzania

Abstract

More often daily climate data have been used in hydrological models to estimate flood flows in small data scarce urban and rural catchments where flow peaking and recession are quick within short periods of few hours. This study assesses magnitudes of error of using daily climate data to simulate flood flow hydrographs of a small urban river catchment, Msimbazi River. Spot discharge measurements were available for Apr-May 2014 and MarMay 2015 periods.10-minutes climate records were available at one station, daily temperatures at two stations while daily rainfall records at several stations within the catchment. Visual analysis characterise rainfall events during the 2011, 2014 and 2015 floods, Thiessen-polygon was used for catchment rainfall, Hargreave-Samani model for catchment potential evapotranspiration and flow hydrographs were estimated by calibrated HBV model. Length weighted channel slope was estimated from segment slopes established from available topographical maps and used in estimating time of concentration for the catchment using Kirpich method. Results indicated that simulated flood hydrographs using 10-minutes climate inputs produced higher flood peaks for both December 2011 (peak: 471.6 m3 /s) and April 2014 (peak: 393.5 m3 /s) and expected hydrograph recession behaviour reproducing the Kirpich estimate of time of concentration of 7 hours. Simulated flood peaks using daily climate inputs were 252.3 m3 /s for December 2011 and 205 m3 /s for the April 2014 event being 53% and 52% of those simulated from 10-minutes climate inputs. Despite usefulness of 10-minute climate data, these data are required at more stations within the catchment for reliable simulation of fast receding urban floods and therefore more automatic weather stations are needed in Dar es Salaam.