Background
Bathymetric mapping is essential for monitoring inland water bodies, supporting applications such as sedimentation tracking, reservoir capacity assessment, and flood risk modeling. Traditional approaches, including boat-mounted sonar and LiDAR, are accurate but often limited by cost, accessibility, and inefficiency in shallow or constrained environments.
For this study, the research team used a UAS-based echo sounder system developed by SPH Engineering that offers a flexible and cost-effective solution. This study, conducted by Mississippi State University’s Delta Research and Extension Center (DREC), explores the feasibility and accuracy of UAS-mounted echo-sounders, complemented by LiDAR and GNSS data for validation.
Challenges
Mapping shallow waterbodies introduces unique planning and accuracy challenges:
- Access and safety. Boat surveys are difficult and risky in shallow or remote waters.
- Data gaps. Echo sounders collect depth along flight paths only. Filling in the spaces between these paths requires additional data processing.
- Lack of benchmarks. Until now, there had been no thorough comparison of UAS-collected depth data to ground-truth measurements.
- Flight planning challenge. There was limited prior evaluation of how flight line spacing and data sampling intervals affect the quality of UAV-based echo sounder results, making it necessary to test and establish best practices through this study.
Solution
To overcome these limitations, the research team employed a drone-based echo sounder system, combining modern sensors and geospatial analysis tools.
System Overview and Survey Workflow
- Platform: DJI M300 RTK drone
- Sensor: SPH Engineering ECT D052S dual-frequency echo sounder (50/200 kHz)
- Flight Planning Software: UgCS (Universal Ground Control Software)
- Survey area: 2.09-acre research pond at DREC
- Reference Data: High-resolution LiDAR and GNSS measurements post-drainage
The UAV flew parallel lines 5 m apart, with the echo sounder submerged below the surface. Using a radar altimeter and SkyHub system, the drone maintained a constant height. Data was collected and monitored in real-time using UgCS Custom Payload Monitor (CPM) application. Post-survey, the water was drained and LiDAR scans of the exposed pond bottom were performed. GNSS rover measurements were also taken in residual puddles.
Results
The results confirm that UAV-mounted echo sounders can deliver accurate depth data, especially when sampling strategies are optimized.
Accuracy metrics
To assess how well the drone system captured real underwater elevations, the team compared its depth measurements to high-resolution LiDAR and GNSS data collected after the pond was drained. Accuracy was measured using two standard indicators:
- Root Mean Square Error (RMSE): The average deviation between the drone measurements and the reference data, in cm.
- R² (Coefficient of Determination): Shows how closely the drone data aligns with reference data, with 100% indicating a perfect match.
Findings:
- Full dataset (11,439 points). The average error was ±10.2 cm, with the measurements explaining about 73% of the variation in depth (R² = 72.8%).
- Filtered dataset (95.45%). After removing outliers, accuracy improved to ±6.6 cm, and the match to reference data increased to nearly 87% (R² = 86.6%).
- Most outliers were in shallow water zones; the system performed better in deeper areas.
- Interpolation Accuracy. Among the ten tested interpolation methods, Topo to Raster stood out:
- With 5 m flight line spacing, the best results were achieved with ±6.3 cm error and 83% match to reference data.
- With 10 m spacing, performance was slightly lower, with ±7.1 cm error and 81.5% match.
Sampling strategy findings
The study also tested how flight line spacing and point intervals affected mapping accuracy. Results showed that 5–10 m flight lines and 5–10 m sampling intervals provided the best balance of data quality and efficiency. Wider spacing beyond 20 m led to higher errors and lower correlation.
Conclusion
This study confirms that UAV-based echo sounders can produce accurate depth maps for shallow inland waters, even matching ground-truth LiDAR data within a few cm. By testing over 11,000 data points and comparing ten mapping techniques, the research team identified the best methods and flight parameters for building accurate bathymetric surfaces. These insights can help teams plan better surveys, reduce time in the field, and improve data reliability.