[MTT47-P06] Increasing temporal depth of urban Land-Use Regression Models by wind-data driven dynamic buffer generation
Keywords:Wind, Land Use Regression, Spatial Data, Meteorology
Traditionally, LUR models are used to generate annual or seasonal concentration averages and are not able cover deeper temporal variability. We aim at increasing temporal depth by using a novel approach in generating dynamic buffers for predictor calculation using both monthly wind-speed and wind-direction averages. The traditional circular buffers are replaced by wedges, whose orientation and radius are bound to these meteorological variables.
Here, Hong Kong’s diverse territory is taken as a study site, and our model is trained with 5 years data from 16 government air quality monitoring stations and deployed portable sensors network. We compare our novel approach to both traditional LUR models and a sophisticated urban chemical transport model (ADMS-Urban).
Results will highlight the opportunities in using LUR models for monthly air-quality pollution concentration as an alternative to chemical transport modelling.