Development and performance analysis of an automatic weighing rain gauge

Precipitation has widely been recognized as a fundamental component of the global water cycle. Accurate measurement of precipitation is very necessary for the main input into hydrological models. Hydraulic structures are then required to adequately design for efficient management of water resources. Several types of automatic rain gauges have been used in recent years, such as ultrasonic and laser rain gauges, but tipping-bucket rain gauges are still the common choice. Particularly, the tipping-bucket rain gauge can provide a better temporal resolution for the rainfall intensity. However, questions still remain on the accuracy of graphical representation for the actual rainfall. In this study, a real-time and automatic monitoring instrument was developed for the weighing rain gauge with high precision for precipitation. Three parts were composed of collector, weighing, measurement, and control subsystem. These subsystems were applied to multi-scenario conditions and performed well under the complex field. As such, the instrument was able to realize sample collection and measurement, data transmission and calculation, remote control, and diagnosis synchronously, compared with the traditional. The A/D conversion chip was utilized in the STM32 single-chip microcomputer to amplify the voltage signal of the weighing sensor. Subsequently, two important parameters of rainfall and rainfall intensity were achieved at a minute level with a resolution of 0.01 mm. Finally, a peristaltic pump was selected to verify the calibration of the developed instrument. The target intensities of rainfall were set as 0.02, 0.08, 0.17, 0.25, 0.50, 0.67, 0.83, 1.67, and 3.33 mm/min. The samples with the rainfall intensity of 0.83 mm/min were measured 30 times, and the rest were run five times. The results showed that the average rainfall intensity was 0.85 mm/min, where the histograms of target rainfall intensity presented a normal distribution, indicating higher precision of developed instrument than before. The best fitting linear regression was also represented by a slope with the R2 value close to 1. Additionally, the average error of the designed instrument was -1.32%, while the highest accuracy was 98.67%, and the relative error of less than 5% accounted for more than 85% of the total samples. The measured data of the developed instrument was also much larger than that of the tipping-bucket rain gauge under simulated rainfall conditions. The high resolution and sensitivity to light rain were contributed to the increase of effective rainfall rate and total rainfall. Finally, the performance of the developed instrument was verified under field conditions in the Wangdonggou watershed for one consecutive year. It was found that the annual rainfall was 522.80 mm, particularly concentrated from May to September. Correspondingly, the single rainfall ≤5 mm was the predominant contributor in natural precipitation, accounting for 74.11% of the total number of rainfall events, whereas, the single rainfall of >10-25 mm was the most important to total rainfall. Consequently, the self-designed instrument can widely be expected to automatically monitor the large variation of rainfall in most complex fields.