Leveraging automated machine learning (AutoML) for urban climate emulation

• Location-independent urban climate emulators are developed using AutoML. • A feature importance analysis framework is proposed for AutoML models. • Location and urban surface parameters improve the emulation performance. • Forcing and location are more important than urban surface parameters in emulation. Urban climate models are critical for understanding and addressing the impacts of urban climate change and for supporting the development of sustainable cities. Yet, process-based urban climate models face limitations of high-entry barriers and substantial computing resource consumption, prompting the development of data-driven methods. In this study, we develop location-independent machine learning emulators for the daily maximum canyon air temperature. To overcome the complexities associated with model selection and hyperparameter optimization in machine learning, we apply automated machine learning (AutoML) to emulation tasks and propose a feature importance analysis framework for the AutoML models. AutoML tasks demonstrate that AutoML excels in learning the physics-based urban climate model, achieving a root mean squared error (RMSE) of 0.81 Kelvin for emulators parameterized with location information and urban surface parameters, and an RMSE of 0.91 Kelvin in the temporal extrapolation scenario. The result also indicated that the location information and urban surface parameters can improve the emulation performance. The feature importance of the emulators indicates that urban morphological parameters contribute more to the emulators than radiative and thermal parameters. The study serves as a demonstration of the potential that AutoML holds for advancing urban climate research and facilitating urban climate modeling.