Machine Learning Framework for High-Resolution Air Temperature Downscaling Using LiDAR-Derived Urban Morphological Features

This dataset supports the study titled "Machine Learning Framework for High-Resolution Air Temperature Downscaling Using LiDAR-Derived Urban Morphological Features", published in Urban Climate (DOI: 10.1016/j.uclim.2024.102102).

Content Overview:

• Building Label Data for Footprint Detection:

  • Amsterdam_BDG_Label.rar
  • MiamiDade_BDG_Label.rar

  These are the label datasets used for training the building detection segmentation models. They have been instrumental in accurately detecting building footprints in Amsterdam.

• Amsterdam_3D_Buildings.rar:  CityGML file of 3D building models for Amsterdam, derived from LiDAR data and U-Net3+ model.

• Morphological Features.rar: Contains urban morphological features (in raster format) extracted from LiDAR data used in the study.

• Training and Test Data for Air Temperature Estimation:

  • Train_Test_AvgTemp_Amsterdam.rar
  • Train_Test_MaxTemp_Amsterdam.rar
  • Train_Test_MinTemp_Amsterdam.rar

  This dataset includes training and testing data for estimating air temperatures in three scenarios: average daily temperature, minimum daily temperature, and maximum daily temperature for the city of Amsterdam.