2025 Volume 81 Issue 16 Article ID: 24-16196
This study uses machine learning to develop a model for estimating turbulence statistics in an urban area using building height, vegetation distribution, and elevation as input. Machine learning generally requires a large amount of training data. Therefore, we used a database of the turbulent statistics within an urban district which covers the entire Tokyo 23 wards in 2 m resolution. This database enables us to learn several thousands of 320 x 320 m2 areas. The results show that the horizontal distributions of mean streanwie velocity, spanwise wind velocity, and turbulent kinetic energy at a height of 2 m above the ground can be estimated with significant accuracy. In addition, the spatial mean values of within the test area be estimated with higher accuracy. On the other hand, the accuracy of the estimation for open space with no building or vegetaion was not good, which may be due to the influence of the outside of the test area is more strongly reflected.
