Project context and challenge
The Msimbazi River runs through the heart of Dar es Salaam and floods regularly, affecting dense neighbourhoods and one of the city’s main roads. To design effective flood measures, engineers need an accurate picture of the riverbed and the surrounding floodplain. That data was lacking.
Traditionally, detailed elevation models are made with LiDAR scanners mounted on aircraft, which is expensive, or with time-consuming field surveys. Drones now offer a promising alternative: they can carry either cameras for photogrammetry or compact LiDAR scanners. However, no thorough comparison of the two techniques existed for the conditions found in Sub-Saharan Africa, where floodplains are often covered with reeds, papyrus, mangroves and dense vegetation.
The World Bank asked CDR to deliver reliable survey data for the Msimbazi and, at the same time, to find out which drone technique works best for flood risk studies in such settings.
CDR’s approach and contribution
Together with Shore Monitoring and Research, CDR carried out an integrated survey campaign. Ground control points and benchmarks were set out, river cross sections were measured with RTK-GNSS and an echo sounder, and water levels and salinity were recorded. The entire area was surveyed twice by drone: once with LiDAR and once with photogrammetry.
The two resulting elevation models were then compared in detail using GIS analysis. The team assessed 17 sample areas, from urban neighbourhoods to mangroves, reed beds and farmland, looking at how many reliable ground points each technique produced, how evenly they were spread and how accurate they were. The results were scaled up to the whole basin using nine representative land cover classes.
Vegetation types were mapped across the basin, and for each type the density of ground points and the height differences between the two models were quantified, giving a clear and objective basis for the comparison.
Result
The World Bank received an accurate, fit for purpose dataset of the Msimbazi River and its floodplains for detailed flood studies. The comparison showed that drone LiDAR clearly outperforms photogrammetry in vegetated areas, because the laser reaches the ground between and below the plants. Photogrammetry remains a faster and cheaper option for open and urban areas.
The study concludes with practical guidance on when to use which method, based on project size, vegetation cover and practical issues such as importing equipment.
These insights help clients choose the right survey method for future flood projects in the region, saving time and cost while making sure that flood models rest on reliable data.
