Calibrating low-cost rain gauge sensors for their applications in IoT-based precision agriculture

Understanding the Need for Densified Environmental Monitoring Networks

Environmental observations are crucial for understanding the state of our environment and addressing various challenges such as natural hazards, climate change, and sustainable resource management. However, current observation networks operated by authorities and research institutions often lack the necessary spatial and temporal resolution due to the high costs associated with deploying and maintaining professional-grade monitoring equipment.

In recent years, the rapid advancements in the Internet of Things (IoT) have presented a promising solution to this problem. The availability of smaller, cheaper, and more energy-efficient sensors, combined with ubiquitous connectivity, has enabled the collection of environmental data at a much higher resolution. Even if the quality and reliability of these low-cost sensors may not match that of official measurement stations, the resulting datasets can still provide valuable insights and complement existing monitoring networks.

Precision agriculture is one field that can greatly benefit from the use of IoT-enabled low-cost sensor systems. By collecting real-time data on climate, soil conditions, and water availability, farmers can optimize their agricultural practices, improve resource efficiency, and enhance crop yields. However, to effectively integrate these low-cost sensors into IoT-based precision farming solutions, it is essential to ensure the sensors’ accuracy and reliability through proper calibration.

Designing Modular IoT-Based Sensor Systems

When designing a low-cost sensor system for environmental monitoring, several key requirements must be considered:

1. Cost-Effectiveness: The sensor systems should be low-cost to allow for the deployment of a high number of nodes, enabling the densification of the monitoring network.

2. Data Quality and Reliability: The sensors must provide a certain level of data quality and reliability to ensure the usefulness of the collected information, even if the quality may not match that of professional-grade instruments.

3. Robustness and Low Maintenance: The sensor systems should be robust and require minimal maintenance to reduce operational costs and maximize the flexibility of deployment locations.

4. Energy Efficiency: The systems should be energy-efficient, allowing for long-term, off-grid operation either through battery power or renewable energy sources, such as solar panels.

5. Wireless Connectivity: Real-time or near-real-time data transmission to users or cloud platforms is essential for many applications, requiring the integration of wireless communication technologies.

6. Ease of Use and Modularity: The sensor systems should be easy to install, use, and maintain, even by individuals without specialized technical knowledge. Modularity in terms of sensor selection, power supply, and connectivity options is desirable to accommodate different deployment scenarios.

7. Open-Source Hardware: The use of open-source hardware can enhance the applicability and transferability of the sensor systems, as well as reduce development costs.

To meet these requirements, two different modular sensor systems have been developed, based on the popular open-source platforms of Raspberry Pi and Arduino.

Raspberry Pi-Based Sensor System

The Raspberry Pi Zero W (RPi0W) was chosen as the foundation for the Raspberry Pi-based sensor system due to its low power consumption and cost-effectiveness (under €10). Previous studies have shown the suitability of Raspberry Pi models for measuring various hydrological parameters, such as water levels, through camera-based techniques.

The RPi-based system is powered through the mains, allowing it to run constantly without the need to optimize for energy consumption. Sensors are connected to the input/output (I/O) pins and read out using Python scripts. Sensor data is then stored in a SQLite database and, depending on the use case, can be transmitted to a cloud infrastructure using the on-board Wi-Fi module or a USB UMTS modem.

Arduino-Based Sensor System

The Arduino-based sensor system utilizes boards from the Arduino MKR series, which feature a low-power ARM Cortex-M0 SAM D21 processor. Specifically, the Arduino MKR Fox 1200 and the Arduino MKR GSM 1400 were used, with other network options available.

The main benefit of the Arduino system is its significantly lower energy consumption compared to the Raspberry Pi. By leveraging deep-sleep modes through a low-power library, the energy consumption of the entire system can be reduced to less than 5 mW. This allows the Arduino system to be powered directly by batteries or a small solar panel, enabling long-term, off-grid operation.

The Arduino system requires two additional components: a standard SD card for data storage and an external real-time clock (RTC) module. The MKR Fox 1200 version has a limitation where the system time cannot be updated online through the Sigfox network, so the accuracy of the timestamps relies on the drift of the RTC module.

Both the Raspberry Pi and Arduino-based sensor systems are capable of connecting to a variety of low-cost sensors, such as temperature, humidity, and air pressure sensors, as well as professional-grade rain gauges like the Davis Vantage Pro2.

Evaluating the Factory Calibration of a Low-Cost Rain Gauge

The focus of this study was to assess the suitability of a widely used low-cost precipitation sensor, the Davis tipping bucket rain gauge, for an out-of-the-box use in IoT-based precision agriculture applications. This involved evaluating the quality of the factory calibration and the sensor’s performance in both laboratory and field settings.

Laboratory Calibration

The laboratory calibration aimed to assess the accuracy of the factory calibration and the consistency across multiple sensors of the same type. A total of 66 rain gauges were tested, including 37 new gauges of Type A, 20 used gauges of Type A, and 9 new gauges of the updated Type B.

The key findings from the laboratory calibration are as follows:

1. Type A Gauges (Old Design): The mean water amount required for one tip was 0.174 mm, with a standard deviation of 0.013 mm. This is lower than the 0.2 mm stated by the manufacturer. Additionally, significant differences were observed between the left and right tipping buckets, leading to larger errors for small precipitation events.

2. Type B Gauges (New Design): The mean water amount required for one tip was 0.194 mm, with a standard deviation of 0.004 mm, which is closer to the manufacturer’s specification. The differences between the tipping buckets were much smaller compared to the Type A gauges.

These results suggest that the factory calibration of the Type A gauges may have been performed on only one side of the tipping mechanism, leading to the observed imbalance. The Type B gauges, with their single-bucket design, appear to have a more consistent factory calibration.

Field Performance Evaluation

In addition to the laboratory tests, a field study was conducted at the meteorological site of the TU Dresden, where 20 identical Type A rain gauges were set up in an array and compared to three professional-grade reference instruments (Hellmann, OTT Pluvio, and Young tipping gauge).

The key findings from the field study are as follows:

1. Cumulative Precipitation: The low-cost gauges showed an average undercatch of 11.1% compared to the reference Hellmann gauge over the 8-month study period. This was better than the results of the OTT Pluvio (18.4% undercatch) and the Young tipping gauge (28.4% undercatch).

2. Correlation with Reference Gauges: The correlation between the low-cost gauges and the reference instruments improved with increasing accumulation intervals, reaching high values (0.973-0.991) at the daily timescale. However, at shorter intervals (15 minutes or less), the correlations were much lower, necessitating the use of multiple low-cost gauges to improve the reliability.

3. Influence of Snow and Partial Blockages: The low-cost gauges without heating were affected by snow accumulation and partial blockages, leading to significant deviations from the reference measurements during these events.

These field results indicate that the low-cost rain gauges, when used in an out-of-the-box manner, can provide valuable precipitation data that is generally better than the performance of some professional-grade instruments, especially for longer accumulation intervals. However, the factory calibration should be checked, and the sensors may require recalibration or the use of multiple units to achieve the desired data quality and reliability for precision agriculture applications.

Recommendations for Calibrating Low-Cost Rain Gauges

Based on the laboratory and field evaluations, the following recommendations can be made for calibrating low-cost rain gauges, such as the Davis tipping bucket sensors, for their use in IoT-based precision agriculture:

1. Check Factory Calibration: Perform a laboratory calibration to verify the factory-stated precipitation volume per tip. This is especially important for the older Type A gauges, as the laboratory tests showed significant variability in the factory calibration.

2. Assess Bucket Imbalance: Evaluate the differences between the left and right tipping buckets. If the difference exceeds 0.05 mm of precipitation (25% of the nominal volume), recalibration is recommended to ensure accurate measurements, particularly for small rainfall events.

3. Recalibrate if Necessary: If the factory calibration deviates significantly from the expected 0.2 mm per tip or the bucket imbalance is too high, recalibrate the gauges using a controlled lab setup and a high-precision scale.

4. Use Multiple Gauges: Consider deploying a network of low-cost rain gauges, as this can help improve the reliability and reduce the impact of localized factors, such as partial blockages or small-scale precipitation variability.

5. Implement Automated Calibration Checks: Develop procedures for periodically checking the calibration of the low-cost rain gauges, either through automated lab tests or by comparing the sensor data with a reference instrument in the field.

6. Address Snow and Partial Blockages: Equip the low-cost rain gauges with heating elements or other mechanisms to mitigate the effects of snow accumulation and partial blockages, which can significantly degrade the measurement accuracy.

By following these recommendations, farmers and precision agriculture practitioners can effectively integrate low-cost rain gauge sensors into their IoT-based monitoring systems, ensuring reliable precipitation data to support data-driven decision-making and optimize their agricultural practices.

Conclusion

The integration of low-cost sensors into IoT-based precision agriculture systems holds great promise for enhancing environmental monitoring and supporting sustainable farming practices. However, to fully realize the benefits of these technologies, it is essential to address the challenges associated with the quality and reliability of the sensor data.

This study has demonstrated the importance of calibrating low-cost rain gauges, such as the widely used Davis tipping bucket sensors, to ensure their suitability for precision agriculture applications. The laboratory and field evaluations have revealed the need to carefully assess the factory calibration, address imbalances in the tipping mechanism, and potentially recalibrate the sensors to achieve the desired data quality.

By following the recommended calibration procedures and implementing strategies to mitigate the impact of environmental factors, farmers and precision agriculture practitioners can effectively leverage low-cost rain gauge sensors to densify their monitoring networks and make more informed decisions for improving resource efficiency, crop yields, and overall sustainability.

The open-source sensor systems presented in this article, along with the insights gained from the calibration analysis, provide a valuable foundation for integrating low-cost environmental monitoring solutions into IoT-based precision agriculture ecosystems. As technology continues to evolve, these modular and adaptable sensor platforms can be further enhanced to meet the growing demands of the agricultural sector and contribute to the broader goal of sustainable resource management.