Prof. Yoshikane explained the importance of high-resolution models to predict extreme precipitation. Such an event cannot be well-represented by the global or regional numerical models. Furthermore, standard statistical downscaling methods typically fail to attain the expected accuracy. He proposed an accurate downscaling approach using machine learning to improve the global or regional model results. The method has been successfully implemented to estimate the spatial distribution of precipitation in the Southern part of Japan. Additionally, with an extended dataset, the method is also capable of discerning the effect of climate change on the distribution of precipitation. Prof. Yoshikane also explained the ongoing research in his laboratory to improve the current downscaling method further and its application to a hydrological model for simulating river discharge.