Development and Calibration of a Low-Cost, Piezoelectric Rainfall Sensor through Machine Learning

Area of Science:

• Environmental science
• Instrument development
• Hydrology

Background:

• In situ precipitation measurements are crucial for hydrology and environmental monitoring.
• Conventional instruments like tipping bucket rain gauges and disdrometers have limitations in cost and network coverage.
• Calibration and understanding instrument response are critical for accurate rainfall data.

Purpose of the Study:

• To develop a prototype of a low-cost impact rain gauge using a piezoelectric sensor.
• To calibrate the sensor's voltage signal response to rainfall intensity using machine learning.
• To evaluate the performance of the developed prototype against a commercial disdrometer.

Main Methods:

• Development of a prototype impact rain gauge using off-the-shelf and reused piezoelectric sensor components.
• Calibration of the sensor's voltage signal properties against rainfall intensity using machine learning algorithms.
• Comparative analysis of the prototype's 1-minute rainfall measurements with data from a co-located commercial disdrometer.

Main Results:

• Successful development of a functional prototype impact rain gauge.
• Machine learning effectively calibrated the relationship between sensor signal and rainfall intensity.
• The low-cost sensor exhibited fairly good performance in monitoring and characterizing rainfall events when compared to a commercial disdrometer.

Conclusions:

• The developed low-cost piezoelectric impact rain gauge shows promise for affordable and effective in situ precipitation monitoring.
• This technology can potentially improve rainfall measurement network density and reduce associated costs.
• Further research can refine the sensor and calibration for enhanced accuracy and broader application.