A comparison between low-cost single-parameter IoT sensors and traditional high-cost monitoring stations for stream water quality assessment

This study evaluates the performance of Low-Cost Internet of Things (LCIoT) sensors for measuring dissolved oxygen (DO) in urban streams, comparing their accuracy to state-of-the-art reference instruments. Through four field deployments between 2022 and 2025, both DIY and commercially available LCIoT sensor systems were tested alongside semi-permanent water quality monitoring stations used for regulatory monitoring. The results demonstrate strong agreement in temporal DO dynamics, with mean absolute errors consistently below 1 mg/L and Pearson correlations exceeding 0.90 in some cases. The Nash–Sutcliffe Efficiency (NSE) reached as high as 0.906, indicating excellent performance in capturing variance. Despite minor systematic deviations and power-related data losses, the LCIoT reliably detected diurnal and seasonal patterns and detected low and high DO events, with minimal sensor drift and low maintenance demands, even without recalibration. In specific situations, slightly larger daily ranges and deviations at extreme values were observed compared to reference instruments. While these differences did not impair trend detection or event identification, they point to potential differences in signal dynamics that warrant further study. Overall, the findings support the suitability of LCIoT sensors for long-term monitoring and large-scale deployment, particularly where budget or physical access constraints exist. Practical guidance on installation, data processing, and power management is also provided to facilitate future applications. • Low-cost IoT sensors accurately monitor dissolved oxygen in streams • Errors stayed below 1 mg/L with strong correlation to reference stations • IoT devices detected diurnal cycles, seasonal shifts, and pollution events • Affordable systems enable large-scale, long-term water quality monitoring