Intercomparison of WRF Urban Configurations for Optimizing Long-Term High-Resolution Simulations in the Chicago Metro Area

Regional Climate Models (RCMs) are some of the best tools to study the effect of climate change on urban areas and to evaluate the impact of adaptation and mitigation strategies. Constant model development seeks to produce more accurate results often adding complexity at the expense of higher computational costs. Current models offer various possible configurations in terms of the physics applied and dynamic setups, with advantages and disadvantages for different situations. Given this increasing range of alternatives, it is often unclear which RCM configuration is more suitable for any given case, and many studies don’t provide a clear explanation of why they choose a specific setup. Here, we test whether using more advanced Urban Canopy Models (UCMs) will produce large enough improvements in urban climate representation compared to simpler configurations to justify the increased computational cost. We also test whether a more precise selection of urban parameters increases or shortens this gap. We used the RCM WRF with three different configurations: the more complex UCM BEPBEM, the simpler and well-tested SLUCM, and the newly added SLUCM using distributed urban parameters and anthropogenic heat as inputs. We ran historical simulations over three summer months in the Chicago Metro Area at 1km resolution and evaluated the model’s representation of urban climate characteristics against reanalysis data and observations. In addition to using performance metrics based on point comparisons, we analyzed the correct representation of urban dynamics using pattern evaluation (EOF analysis). We focus our results on evaluating urban effects on temperature and precipitation, especially the accurate representation of extreme events (heatwaves and heavy precipitation). We hypothesize that using a simpler model with a more accurate and comprehensive urban parameter input set will produce results close to those of the most sophisticated model configurations, with lower computational costs, allowing for higher resolution and more extended simulations.