The first model evaluation results from the Urban-PLUMBER project are now published!
This phase evaluates land surface models at the same site as the last major urban model intercomparison by Grimmond et al., 2011.
This allows us to assess how the last decade of model development affected performance. Since the last intercomparison found that the representation of vegetation in urban areas was critical for model performance, much of the development efforts in the last decade were focussed on better representing vegetation and evapotranspiration processes.
We find this development work appears to have paid dividends, with this cohort of models performing much better in latent and sensible heat fluxes. While previously mid-complexity models (such as “canyon” schemes) performed more poorly than simple slab models, that is no longer the case. Where evapotranspiration processes are well represented, more complex urban schemes (those that more explicitly resolve processes and with more parameters describing urban characteristics) can perform well.
Importantly, we also find that human factors play an important role in performance. More experienced groups (either with a model or with the site) tended to have lower errors, particularly in the initial submissions. Although significant efforts are applied to test models rather than modellers, we conclude that human factors are likely to influence results in this (or any) model intercomparison.
Check out the paper here: https://doi.org/10.1002/qj.4589