Environmental study on greenery planning scenarios to improve the air quality in urban canyons

Abstract

• Regression models were developed to prediction model of PM 2.5 concentrations. • The accuracy of the optimized predictive model is 97.5%. • The simulation results show that AQI rates correlate significantly with green spaces and wind direction. • Results show AQI rates correlate significantly with green spaces and wind direction. Urban Green Spaces (UGS) offer various environmental benefits, including controlling the Air Quality Index (AQI), regulating outdoor thermal comfort, and providing suitable spaces for enhanced human health. Due to the high concentrations of pollutants in cities, especially particulate matters with a 2.5-mm diameter (PM 2.5 ), various countries have a wide range of AQI rates. This paper attempts to generalize the results from ENVI-met simulations applied to street canyon configurations in nine cities worldwide and seeks to find a quantitative model to predict ambient PM 2.5 concentrations in terms of meteorological and built environment variables for any street canyon worldwide with the same climate conditions to the simulated models. We selected nine cities from a range of most polluted cities to the least ones based on the statistics in 2019. First, we defined four different scenarios within a pattern of Green Infrastructure (GI) located on the sidewalks; also, by considering independent (greenery and wind direction) and dependent (wind speed, air temperature, humidity, and H/W) variables to find the optimized scenario throw an optimization process. The simulation results show that AQI rates correlate significantly with green spaces and wind direction, and the optimized scenario could decrease the PM 2.5 ambient concentrations up to 33% at the level 1.75 m above the ground, in which people breathe, throw dispersion and deposition of the pollutants. In terms of prediction objectives, regression models were developed to represent the importance of variables and the prediction model of PM 2.5 concentrations in the ambient conditions. The accuracy of the optimized predictive model is 97.5%. We ran a case study with different climatic and meteorological conditions, indicating that the optimized algorithm in a predictive model can be used universally with different AQI and with common climate conditions in the simulated cities.