Yoshihiro MORITA

Study on Precipitation Estimation Methods Using Particle Diameter and Fall Velocity Data from Optical Precipitation Particle Sensors

Toshiro KUMAKURA

Many precipitation observation stations exist nationwide and are used for various purposes. The widely used tipping-bucket rain gauge struggles to accurately measure solid precipitation due to delays in melting during heavy snowfall. Therefore, solid precipitation observation instruments suitable for multi-point observations are needed. The optical reflection-type solid precipitation sensor (PDS) proposed by Kumakura et al. was developed with this in mind. PDS is characterized by its low cost, easy installation, and good maintainability. This study examines whether the precipitation estimation method using particle diameter and fall velocity, which was secondarily proposed by Ishizaka et al. (2013), can be applied to PDS. Additionally, the study investigates the impact of precipitation type differences on this method. For this purpose, data from the well-established precipitation particle sensor LPM was used. To account for precipitation type variations, precipitation classification was performed at the same time interval as the precipitation estimation, and classification information was assigned using CMF, a parameter that quantitatively represents particle characteristics based on diameter and velocity. Since precipitation estimation and CMF stability behave oppositely concerning calculation intervals, a 5-minute interval was selected for balance.Regression analysis was conducted to compare the calculated precipitation amount for each precipitation type with reference values. The results showed clear differences in regression coefficients, even when considering only specific temperature conditions to refine the dataset. Furthermore, the Ishizaka et al. method showed a strong correlation with observed values but had an underestimation tendency. Adjusting the coefficient for each precipitation type further improved estimation accuracy.For PDS, frequent missing data and underestimations were observed. This is likely due to the imbalance between infrequent observations of large particles passing directly in front of the sensor and the frequent detection of small or distant particles, which does not converge in a 5-minute interval. The convergence of observations depends on precipitation intensity and time intervals, highlighting the need for further investigation. Moreover, PDS also exhibited differences in regression coefficients by precipitation type, likely influenced not only by its estimation algorithm but also by the attenuation of reflected light caused by liquid water. By removing the effect of liquid water and applying type-specific coefficients, a relatively high correlation coefficient of 0.813 was achieved.