Yugo Arisaka

Development of AI Technology to Determine Precipitation Type and Intensity Using Precipitation Images

Minjiao LU

There is a growing need for precise measurement and observation of precipitation type and amount for various purposes such as weather forecasting and disaster prevention. However, the current observation equipment is expensive and installation at multiple locations is challenging. Recently, advancements in AI technology have led to the development of a system that can identify precipitation types and relative intensities using precipitation images. By capturing images and measuring precipitation intensity, the AI can determine the type of precipitation instantly. The AI analyzes patterns in the images and can differentiate between different types and intensities of precipitation with sufficient training data. The study found that using images with a black background can shorten the learning time for the AI. Additionally, standardizing the shooting distance and depth of field can help accurately determine the size, thickness, and density of the precipitation in the images. The study achieved a 100% accuracy rate for precipitation judgments using deep learning on 30 images. However, more images and addressing the issue of different precipitation intensity levels are necessary for future improvements.