Analyzing GRAL model performance in street canyons: Influence of atmospheric stability schemes and simulation strategies.

Exposure of urban population to high concentrations of particulate matter is one of the most urgent worries, particularly in areas characterized by the presence of congested traffic and street canyons which confine polluted air. Therefore, robust tools are essential to simulate the effect of obstacles and reduced ventilation. The Lagrangian dispersion model GRAL is able to accurately represent the microscale spatial variability of concentrations in terms of long-term averages, however it demonstrates some deficiencies to predict detailed time series when the input is not carefully selected. In this study, the impact of certain configuration parameters was examined for the simulation of PM10 in the surroundings of a street canyon in Prague (Czech Republic) under contrasting meteorological conditions in summer and winter. First, the sensitivity of steady-state simulations to different atmospheric stability index schemes was analysed. Out of the three options compared, the solar radiation/delta-T (SRDT) method was the most effective. Secondly, the steady-state "match-to-observation" mode and transient mode were tested using the SRDT stability scheme, revealing a general reduction of unrealistic peaks of concentration encountered in the standard steady-state mode. Finally, some other factors causing bias were identified: unreliable sensor measurements, inappropriate boundary conditions, and absence of resuspension emissions.