Scalable rain—snow discrimination in glacierized regions using tower observations and time-lapse imagery

Precipitation plays a crucial role in controlling glacier mass balance, yet accurately distinguishing rainfall from snowfall in high-altitude glacierized regions remains difficult due to the limits of in situ observations. To address this challenge, a hybrid approach integrating meteorological tower measurements with time-lapse camera imagery was developed for rapid and reliable precipitation phase identification, applied on Yulong Snow Mountain from August 1, 2019 to July 11, 2021. Based on the Exponential Double-Temperature Threshold Method (EBTM), the results revealed a sharp phase transition from snowfall to rainfall between −1.5 and 3 ℃. Additionally, a critical threshold of 0.25 ℃ was identified using the Mathematical-Statistical Method (MEM). Comparisons with other widely used methods suggest that EBTM is the most accurate for quantifying the amount for different precipitation phase, and the MEM provides a quick identification the phase classification. It was noted that all models show reduced performance within the narrow transition range of −1 to 2 ℃. Overall, this study demonstrates that the proposed hybrid method is cost-effective and adaptable, thus holding broad potential for application in other high-altitude, data-scarce regions.