Challenges of validating microscale LES air quality models in real urban environments: a case study for Prague city centre

Air quality deterioration in street canyons is an important issue strongly influencing human health, and the microscale LES models can provide very detailed information about its spatial and temporal structure. However, the proper model configuration and its comprehensive validation in real urban conditions still represent a significant challenge that has not been fully addressed. We present a validation campaign conducted in a traffic-laden area of Prague focusing on the detailed temporal and spatial structure of concentration fields. Almost one year of field measurements were collected with a carefully designed air quality sensor network complemented by reference and remote sensing air quality and meteorological observation tools. All measured data were consequently checked and verified. The detailed data of land cover, emissions, and meteorological conditions were provided to the PALM LES model, which was configured for the studied area in two nested domains. The comparison of the modeled and measured values showed good agreement during some episodes, while significant discrepancies were discovered in other episodes, mainly during stable meteorological situations. The identified reasons include imprecise meteorological boundary conditions, uncertain emission inputs, and insufficiently represented or neglected processes in the model. Also, the imprecision introduced by the advection numerical scheme was identified in some situations. We propose and test selected amends that improve the agreement of the model with observations. This experiment provides a knowledge base applicable to future validation campaigns and for further model development and application.