[4Rin1] Interactive Session 2

[4Rin1-33] Development of method of spot weather forecast using machine learning

〇Takao Yoshikane¹, Kei Yoshimura¹ (1. The University of Tokyo)

Keywords:Local Weather forecast, Machine learning, Numerical weather prediction model

Local weather forecasting is a very challenging issue because it is technically very difficult and there are a lot of problems with numerical model for now. In this study, we found the linkage between the observation and simulated regional patterns and then estimate the local precipitation, temperature, and surface wind recognizing the regional distribution patterns, which is corresponded to the weather patterns, using machine learning (A regression model of support vector machine). As the result, the model biases were drastically reduced by the innovative approach. This method could greatly contribute to realize the local weather forecast.