Wind direction related parameters improve the performance of a land use regression model for ultrafine particles

Özet

The performance of land use regression models (LUR) depends on the quality and the representation of the explanatory variables and the spatial parameters. Generally, those parameters are implemented in a static manner, i.e. different wind directions are not depicted in the LUR approach. We introduced Pseudo Dynamic Parameters (PDPs) that represent the interaction between wind and the urban morphology as a function of wind direction to improve LUR model performance based on 27 monitoring spots in an urban study area in Braunschweig, Germany. The explanatory parameters were categorized assuming a process chain of particle emission, dilution, and deposition. The Pseudo Dynamic Parameters were categorized under the dilution process. As a result, twelve different LUR models were calculated for two main wind direction sectors (South East and South West). The final LUR models that used only static parameters (LUR_STATIC) and the ones that used both static and PDPs (LUR_CMB) were compared. The results showed that LUR_CMB models explained 82% and 89% of the variance of UFP concentrations in comparison to LUR_STATIC that explained 68% and 79% for both wind direction sectors (SE and SW), respectively