River corridors are among the most challenging environments to survey because no single technology can capture every feature above and below the waterline. This project demonstrates how integrating airborne lidar, mobile mapping, UAV lidar, and multibeam bathymetry can produce a seamless digital terrain model for engineering, navigation, flood analysis, and long-term river management.
Redefining River Corridor Data
When no single sensor can capture the full picture, the answer lies in combining them all.
Serbia’s Danube and Sava rivers have been central to European commerce and infrastructure for centuries. The Romans carved roads into the Iron Gate Gorge cliffs, built towpaths along the banks, and cut bypass canals around its rapids. Today, the challenge is understanding the waterway’s geometry, hydrology, and infrastructure well enough to keep it functioning for the future.
That understanding is now taking shape through one of the most technically comprehensive river surveys undertaken in Europe. Over 660 kilometres of waterway, EAASI member MapSoft — a Serbian geomatics company with more than two decades of spatial data acquisition experience — has combined four complementary survey methods into a single integrated dataset, creating a seamless, continuous model from floodplain to riverbed.
The project, “Data Collection, Hydraulic and Morphological Modelling of the Danube River and Sava River in the Republic of Serbia – Lot 02,” is financed by the European Investment Bank (EIB) on behalf of the beneficiary, the Ministry of Construction, Transport and Infrastructure of the Republic of Serbia. The initiative was implemented in cooperation with Egis and Deltares, who managed the hydraulic and morphological modelling, while MapSoft executed the comprehensive geospatial data acquisition, processing, and integration.
Why no single sensor is enough
A river environment cannot be fully captured by any single survey method. Airborne sensors efficiently map floodplains and overbank terrain but have limited ability to penetrate the water surface. Sonar systems provide high-precision bathymetry of the riverbed but no information above the waterline. Boat-mounted mobile mapping captures shoreline geometry and vertical infrastructure in detail but lacks the spatial reach of airborne platforms. UAVs add ultra-high-resolution data for localized features such as bridge components and inaccessible structures. Removing any one of these data sources introduces gaps that can affect hydraulic modelling, flood analysis, and navigation planning.
MapSoft’s methodology integrates all four acquisition domains into a single, coherent survey framework.
Airborne LiDAR: floodplain coverage
Airborne LiDAR provided broad coverage of floodplains, open areas, and gently sloping riverbanks across 452 square kilometres. The system — a Teledyne Optech Galaxy EDGE LiDAR solution, paired with a Phase One iXU-RS1000 digital camera — was installed in a SILA 750C aircraft.
Vessel-based mobile mapping: the water–shore interface
Vessel-based mobile laser scanning captured the water–shore contact zone, steep or vertical banks, retaining walls, harbours, and other river structures along 3,100 kilometres — terrain that airborne data could not fully represent, particularly where scan angle, vegetation, and occlusion limit aerial coverage on steep banks. The system combined a Teledyne Optech CL-360HD laser scanner, a 360° camera, two 12-megapixel side-view cameras, and a Trimble Applanix GNSS/INS positioning system. Mobile laser scanning was prioritised in nearshore zones and complex bank geometry, while airborne LiDAR remained the preferred source for floodplains and gentler terrain.
UAV LiDAR: bridge structures
Twenty-nine bridges and other river-crossing structures were surveyed using a DJI Matrice 350 RTK equipped with a DJI Zenmuse L2 LiDAR scanner, generating high-density 3D point clouds of bridge decks, piers, abutments, and visible structural elements.
Multibeam bathymetry: the underwater corridor
The underwater component, covering 661 kilometres of riverbed, used a NORBIT iWBMS multibeam echo sounder meeting IHO S-44 Special Order requirements. Two vessels were deployed — one suited to shallow areas and canals, the other built for colder, rainier, and more demanding conditions. Near banks, bridge piers, submerged rock formations, barges, and water huts required additional passes and adjusted sonar settings, with vessel speed, sonar field of view, roll compensation, and swath overlap all controlled to reduce artefacts and maintain sufficient point density.
Integration: one continuous model
The central technical challenge was integrating four fundamentally different data types into a single, continuous terrain model. Particular care was given to the dry–wet interface, so that no artificial steps or discontinuities appeared between riverbank and riverbed data — a task simplified by the dense, partly overlapping datasets the four sensors provided for cross-checking and controlled interpolation where direct measurement was not possible.
Classification rules determined what was retained in the final terrain model: only approved terrain classes were used — airborne ground points, mobile-scan ground points, and multibeam water-bottom returns. Non-terrain features such as bridge structures, navigation aids, floating objects, and isolated spikes were excluded, while genuine morphological features — sandbars, dunes, scour holes, submerged structures, and wrecks — were preserved where they represented actual riverbed form.
Airborne and mobile laser scanning data were merged first to build the terrestrial and riverbank model; multibeam bathymetry was then integrated as the riverbed source, with overlap areas trimmed using defined source-priority rules and gaps identified through point-density and source-provenance checks. The resulting Hybrid Digital Terrain Model was produced at 0.5-by-0.5-metre resolution, with valid terrain points averaged within each grid cell and only local, terrain-plausible voids interpolated.
Final products
Beyond the hybrid digital terrain model itself, the project delivered a full set of complementary geospatial products: classified point clouds combining airborne, bathymetric, and mobile laser scanning data; UAV LiDAR point clouds of the 29 surveyed bridges; spherical and side-view imagery supporting interpretation and quality control of riverbank features; vector drawings of significant structures relevant to modelling and navigation assessment; survey coverage and metadata layers documenting acquisition dates and processed sections; and quality control documentation providing traceability of processing steps, accuracy checks, and data validation.
Together, these products form a consistent geospatial database of the river corridor, providing the technical foundation for hydraulic calculations, morphological analysis, navigation-condition assessment, flood-risk evaluation, and future river-management planning along both waterways.
Because the survey was the first integral measurement of the entire Danube and Sava corridor in Serbia, it also produced a detailed record of features that were already known to exist but had not previously been documented at this resolution, including the submerged settlement of Donji Milanovac, the former Ada Kaleh island fortress, sections of Trajan’s Roman road, the nineteenth-century Šip Canal, and the wrecks of vessels scuttled during the Second World War.
In mapping the corridor for navigation and engineering, the survey also gave Serbia its first complete look at what the rivers had been holding all along.
MapSoft and Teledyne Geospatial are both members of the European Association of Aerial Surveying Industries, EAASI. This project was originally presented at the EAASI 2025 Summit in Dubrovnik.
Another Article: Real-Time LiDAR Monitoring for Infrastructure
