Kai Sasaki

Improvement of a Precipitation Estimation Method Using an Optical Reflective-Type Sensor

Kumakura Toshiro

Precipitation measurements are routinely conducted and are essential for weather forecasting and climatological records. In particular, accurate snowfall observation is important for preventing traffic accidents and congestion caused by snow accumulation. In Japan, tipping-bucket rain gauges are widely used at AMeDAS stations; however, they are not suitable for solid precipitation such as snow. Heated tipping-bucket gauges with melting mechanisms can become buried during heavy snowfall. Although disdrometers such as the LPM and 2DVD are suitable for measuring solid precipitation, they are generally expensive and not appropriate for deployment at multiple observation sites. Consequently, it remains difficult to achieve accurate and spatially dense measurements of solid precipitation during heavy snowfall events.
In this study, we investigated a precipitation estimation method using an optical reflective-type sensor called PDS, which is inexpensive and suitable for multi-site deployment. The PDS consists of four horizontally aligned emitters that project near-infrared light. When hydrometeors pass through the sensing area, the reflected light is detected by receivers mounted beneath the emitters. Because the PDS cannot directly measure precipitation amount, particle diameter and fall velocity are calculated from the maximum voltage and transit time derived from the waveform signal. Precipitation amount is then estimated using a precipitation estimation formula based on these parameters.
Previously, simple calculation formulas were used to estimate particle diameter and fall velocity. In this study, more sophisticated formulas were introduced to improve precipitation estimation accuracy. The estimated precipitation was compared with measurements obtained using an electronic balance, which was regarded as the reference value. Regression analysis was conducted, and accuracy was evaluated using the coefficient of determination (R²). Experimental data were obtained from indoor artificial snowfall experiments conducted at the Shinjo Cryospheric Environment Laboratory.
The results showed that the coefficient of determination improved from 0.401 with the conventional method to 0.476 with the proposed method. Furthermore, cumulative precipitation over time estimated by the new method exhibited behavior closer to that measured by the electronic balance. These findings indicate that employing more advanced calculation formulas for particle diameter and fall velocity improves precipitation estimation accuracy.