Flood forecasting in Manafwa river basin using GIS and remote sensing

Abstract

Floods are the most frequent and destructive water-related disasters, accounting for nearly 50% of all water-related disasters, which themselves represent about 90% of natural disasters worldwide resulting in significant loss of life, displacement, and economic hardship. Despite these escalating impacts, most flood-prone areas in Uganda and similar regions lack operational flood early warning systems, largely due to data scarcity and limited hydrological monitoring. This deficiency severely impedes effective disaster preparedness, risk reduction, and timely response. This study focuses on the Manafwa river basin, a region representative of many flood-prone basins in sub-Saharan Africa lacking operational forecasting tools. The basin encompasses several districts, with Butaleja District being the most severely affected by recurrent floods. Field investigations revealed that Butaleja’s vulnerability stems from its low-lying topography, proximity to the confluence of several tributaries, and extensive rice cultivation in floodplains, which increases exposure to inundation. Major flood-affected locations identified during the field visit include Butaleja Town Council, Doho Irrigation Scheme, Lelesi village, Namulo trading centre, and Kangalaba in sub counties of Himutu, Mazimasa and Kachonga. The study integrates satellite-derived meteorological forecast from ECMWF model with hydrological (HEC-HMS) and hydraulic (HEC-RAS) modeling to delineate watershed characteristics, identify key hydrological parameters, and simulate runoff and flood extents for multiple return periods. Model calibration and validation demonstrated satisfactory performance (R² = 0.86, NSE = 0.55, PBIAS = -8.34), indicating reliable reproduction of observed flow patterns and flood events. Key findings reveal extensive areas at risk of inundation, with flood maps clearly delineating potential flood extents for 2, 5, and 10-year return periods and flood forecasts for the month of May with peak flows of 42.57m3/s, 61.48 m3/s and 73.96m3/s for the 2, 5 and 10-year return periods respectively. For the flood forecasts carried for days 5th, 10th and 15th May, 2025, river flows obtained were 17.5m3/s, 16.7m3/s and 16.9m3/s respectively which were below the 2-year peak flow indicating no flooding during the forecast period. The flood forecasts were validated by a filed investigation in Butaleja district using the questionnaire approach, visual inspection and oral interviews with the residents. These outputs provide actionable insights for early warning, emergency planning, and land use management. These results underscore that integrating GIS and remote sensing with advanced hydrological modeling can enhance flood risk assessment in data-limited environments. This research demonstrates that modern, data-efficient modeling approaches can bridge gaps in flood forecasting where conventional gauge networks are lacking. By providing reliable flood predictions and risk maps, the study supports data-driven decision-making, strengthens disaster preparedness, and enhances community resilience. The approach contributes to the broader goal of sustainable disaster risk reduction and climate adaptation in developing countries.