What Have Studies or Urban Greenhouse Gas Emissions Taught Us about Urban Meteorological Simulations?

Accurate and precise estimates of greenhouse gas emissions from cities require accurate and precise meteorological reanalyses. The transport properties of greatest interest to greenhouse gas budgets are boundary layer depth, wind speed and wind direction. We present a synthesis of studies focused on the ability of the Weather Research and Forecast model (WRF) to reproduced the observed state of the atmospheric boundary layer in and around the city of Indianapolis, Indiana. We show that simulated ABL properties in our default WRF reanalysis contain biases that a function of time of year, except for surface layer winds which are always biased high in the model. We demonstrate that ABL dispersion is significantly overestimated close to surface sources, potentially causing errors in the interpretation of tower-based GHG observations. We explore the causes of these biases by varying the parameterizations we employ in the WRF model, and by creating a more realistic description of the land surface. We find that the urban surface energy balance and momentum fluxes are significantly biased. The surface energy balance errors are reduced, but not eliminated, by including a more realistic representation of the fractional vegetation cover within the city. These modifications, however, do not cause a large change in the ABL simulations, suggesting that the errors in the urban ABL simulation are strongly influenced by the simulation of the rural meteorological conditions. We further demonstrate persistent spatial patterns in ABL biases across the US Midwestern region, but show the potential for minimizing these biases with a carefully pruned model ensemble. Finally, we quantify the random errors in simulated ABL properties, and show that assimilation of locally observed ABL winds significantly reduces errors in the simulated winds, and that carefully chosen model ensembles can potentially represent the random error in our reanalyses.

These results show the benefits of integrated ABL observational systems in evaluating urban ABL simulations, and suggest that evaluation of errors in both rural meteorological simulations, and surface layer winds and dispersion are of high importance for improving future reanalyses. Finally, the seasonal and regional nature of ABL biases suggest that no single configuration of the WRF modeling system is likely to minimize errors in all environments, and that model ensembles represent a more realistic path forward in characterizing urban GHG transport and its uncertainties. These findings can be directly translated to studies of urban air quality, and to simulations of urban climate.