Cost-Effective Radar Sensors for Field-Based Water Level Monitoring with Sub-Centimeter Accuracy

TL;DR

The paper addresses low-cost, autonomous water level monitoring using FMCW mmWave radar sensors in real-world environments. It evaluates three TI radar sensors with a simple windowed filtering and aggregation approach to estimate the distance to the water surface from non-contact measurements. The results show sub-centimeter RMSE in field tests, with the IWR1443 providing the most accurate, stable performance over a 3-day automated deployment, while all sensors achieve centimeter-scale accuracy with minimal calibration. The work demonstrates practical deployment potential for drones and robotic platforms and provides open data and code to support replication and extension.

Abstract

Water level monitoring is critical for flood management, water resource allocation, and ecological assessment, yet traditional methods remain costly and limited in coverage. This work explores radar-based sensing as a low-cost alternative for water level estimation, leveraging its non-contact nature and robustness to environmental conditions. Commercial radar sensors are evaluated in real-world field tests, applying statistical filtering techniques to improve accuracy. Results show that a single radar sensor can achieve centimeter-scale precision with minimal calibration, making it a practical solution for autonomous water monitoring using drones and robotic platforms.

Figures (9)

• Figure 2: Experimental setup.
• Figure 3: An example of several radar measurements accumulated in one plot as 2D point clouds. The $Y$ axis shows the vertical distance to the point, while the color denotes the radar signal intensity from high (red) to low (blue). Filtering the points by intensity results in a smaller cloud specifically around the water mass under the sensor.
• Figure 4: Schematic of the experimental setup. The radar sensor (dark red) is mounted at the end of a fixed horizontal pole and emits radio signals (dashed arcs). The signal can penetrate surfaces such as water (dotted blue) and ground (dashed grey). Reflections from metallic structures may follow indirect paths (blue arrows), producing false returns with large apparent $y$ coordinates (dark blue circles). Additional false returns may arise from sensor noise near the sensor (pale blue circles) or from within the water column (pale red circles). Valid surface returns are shown in bright red circles.
• Figure 6: Water level deltas across the entire duration of the automated deployment, computed relative to the initial measurement of each sensor. The distance deltas of the AWR1843 (yellow) and IWR1443 (red) are compared to inverse groundtruth depth deltas (blue).
• Figure : (a) AWR1843