@phdthesis{oai:kitakyu.repo.nii.ac.jp:00000928,
 author = {オウ, ルイ},
 month = {2021-12-25},
 note = {This study introduced a temperature spatial downscaling method based on machine learning algorithm to downscale air temperature from 1 km to 250 m for high-resolution atmosphere urban heat island (UHI) analysis. The core of this downscaling method is to establish the regression model between urban structure and temperature, and then we used the unchanged characteristics of regression models at different scale to predict high-resolution temperature data with high-resolution resolution urban structure, thereby analyzed atmosphere urban heat island. Finally, we compared the similarity and differentiation between atmosphere UHI and surface UHI. The results indicated the following: (1) The machine learning method was proved to be suitable for the air temperature spatial downscaling predication; (2) The UHI characteristics of metropolitan areas in different climatic regions of Japan are different; (3) There are great differences in intensity and spatial distribution between atmosphere UHI and surface UHI is great.},
 school = {北九州市立大学},
 title = {Research on Spatial Downscale Temperature Prediction by using Machine Learning and its Application in Urban Heat Island},
 year = {}
}