Multi-year evaluation and application of the WRF-chem model for two major urban areas in Brazil - Part I: Initial application and model improvement

High-resolution retrospective simulations of air quality can generate O 3 and PM 2.5 concentrations to estimate their health effects and establish a baseline to assess the impacts of future climate and emission scenarios. To this end, the Weather Research and Forecasting Model coupled with Chemistry (WRF-Chem) is applied to a triple-nested domain over Brazil, with a high spatial resolution of 3-km over the metropolitan areas of São Paulo (MASP) and Rio de Janeiro (MARJ) for the period of 2012–2016. This work is presented in two-part papers. Part I describes an initial application and evaluation of WRF-Chem for August 2012 to study the impacts of improved inputs and wind speed correction parameterization options on the model performance. These simulations aim to improve the model's accuracy in reproducing the observed meteorological variables and air pollutant levels. The model results with updated land use and cover and urban fraction show a lower positive bias in temperature at 2-m. Updated elevation based on high-resolution (30-m) data reduces the positive bias on wind speed at 10-m for MARJ. The modified wind speed correction reduces the systematic bias of wind speed at 10-m for all domains. WRF-Chem using the combined emissions based on global and local inventories performs well with lower bias and better temporal and spatial representations of historical concentrations of major air pollutants than those using global emissions only. These sensitivity simulations identify the best possible model inputs and wind speed correction parameterization option for simulations for the period of 2012–2016, which will be presented in Part II paper. • Improved model skill by updating land use, urban fraction, and elevation datasets for key domains. • Reduced wind speed bias using an updated wind speed correction parameterization scheme. • Better captured pollutant levels using global and local emission inventories with new time profiles. • Enhanced agreement with O 3 and PM 2.5 observations, lowering model biases after improvements. • Validated model results with surface and satellite meteorological and chemical observations.