Machine learning emulation of urban land surface processes

Meyer, D. ORCID: https://orcid.org/0000-0002-7071-7547, Grimmond, S. ORCID: https://orcid.org/0000-0002-3166-9415, Dueben, P. ORCID: https://orcid.org/0000-0002-4610-3326, Hogan, R. ORCID: https://orcid.org/0000-0002-3180-5157 and Reeuwijk, M. ORCID: https://orcid.org/0000-0003-4840-5050 (2022) Machine learning emulation of urban land surface processes. Journal of Advances in Modeling Earth Systems, 14 (3). e2021MS002744. ISSN 1942-2466 doi: 10.1029/2021MS002744

Abstract/Summary

Can we improve the modeling of urban land surface processes with machine learning (ML)? A prior comparison of urban land surface models (ULSMs) found that no single model is ‘best’ at predicting all common surface fluxes. Here, we develop an urban neural network (UNN) trained on the mean predicted fluxes from 22 ULSMs at one site. The UNN emulates the mean output of ULSMs accurately. When compared to a reference ULSM (Town Energy Balance; TEB), the UNN has greater accuracy relative to flux observations, less computational cost, and requires fewer input parameters. When coupled to the Weather Research Forecasting (WRF) model using TensorFlow bindings, WRF-UNN is stable and more accurate than the reference WRF-TEB. Although the application is currently constrained by the training data (1 site), we show a novel approach to improve the modeling of surface fluxes by combining the strengths of several ULSMs into one using ML.

Item Type	Article
URI	https://reading-clone.eprints-hosting.org/id/eprint/103329
Identification Number/DOI	10.1029/2021MS002744
Refereed	Yes
Divisions	Science > School of Mathematical, Physical and Computational Sciences > Department of Meteorology
Publisher	AGU
Download/View statistics	View download statistics for this item