Solving the challenges of interpolating NO<sub>2</sub> from SPRINT data and modelling population movements in agent-based modelling

This study addresses two critical challenges in urban air quality exposure simulation and offers solutions. The first challenge is to generate nitrogen dioxide (NO2) fields across the city of London from available stations that provide Spatially Poor but Rich In Time (SPRINT) data. We first used Inverse Distance Weighting (IDW) to spatially interpolate NO2 at each half-a-day step. Each station had a list of hourly NO2 values for each time step, from which one NO2 value was selected by a stochastic process to generate the field. We also added weightings of up to a factor of 3 to London's NOx emissions to account for emissions from sources other than vehicles. We cross-validated the modelled data with the station data and found beta parameter of 1.5 to be the most appropriate 'power' parameter. The second challenge investigated the use of a fractional origin-destination (OD) matrix to see how to overcome errors when assigning destinations to a small set of population. We tested that ‘the nested bin strategy’ worked well for our 6,078 London resident agents. To enrich the dynamics to represent people’s non-work mobility patterns, we included visits to recreational areas during weekends and festive periods. This, in turn, provides a more comprehensive representation of urban mobility. Solutions to each challenge can provide more accurate assessments of pollution exposure, leading to better informed public health interventions.