Land-surface-physics-based downscaling versus conventional dynamical downscaling for high-resolution urban climate change information: The case study of two cities

Abstract

This study for the first time examines the performance of the computationally-efficient land-surface-physics-based downscaling (LSP-DS) approach for high-resolution urban climate prediction by comparing it with the conventional dynamical downscaling (D-DS). LSP-DS utilizes the offline land-surface-modeling system HRLDAS, while D-DS employs the regional climate model WRF. Both approaches integrate the coupled land-surface and urban-canopy models. Simulations are driven by coarse-resolution reanalysis data to achieve 2-km climate downscaling, targeting two cities, Tokyo and Singapore. The findings reveal that LSP-DS can accurately reproduce the urban heat island (UHI) effect at high resolution, comparable with D-DS. LSP-DS even shows consistently better results for urban areas, across varying weather conditions such as heatwaves, non-heatwaves, dry, and rainy periods. Both methods show the same performance on the compound effects of heatwaves and UHI, with LSP-DS tending to simulate moderate UHI effects compared with D-DS. This study highlights the LSP-DS's potential as a computationally efficient and effective tool for urban climate downscaling, particularly to serve the next IPCC special report on climate change and cities. However, users should be mindful of the LSP-DS's limitations, such as the absence of two-way feedback with atmospheric physical and dynamical processes, when applying LSP-DS and explaining its results.