Abstract:
Every day an abundance of movement data is generated by GPS-enabled smartphones. Big datasets containing detailed coordinates and timestamps are widely available from location intelligence companies, providing an alternative to survey-based GPS data, call detail records (CDRs) and travel surveys commonly analysed in mobility studies. This research quantifies the extent of mobility changes during COVID-19 lockdowns in Auckland, New Zealand using big data. Comparisons were made between baseline mobility and four weeks of COVID-19 alert level system restrictions between March and May 2020. Millions of rows of geolocated mobile phone location data from over 20,000 phones were cleaned and processed into common night-time location and significant stop locations (such as places of employment or recreation). The distance, spatial extent/coverage, and the number of locations were calculated using mobility metrics, including total and maximum distance travelled, the radius of gyration, and minimum convex polygon. The results were aggregated from individuals to neighbourhood scale for privacy protection and reported in median per neighbourhood. Overall, the number of significant locations dropped by over 70% during alert level four lockdown compared to before lockdown (and over 50% reduction in alert level three) in line with the closure of non-essential businesses and education facilities. As more people started working from home, the total weekly distance travelled and spread of stops reduced by approximately 90%. The Index of Multiple Deprivation (2018) was used to explore variations between different areas of socio-economic deprivation and tested for statistical significance. During baseline (pre lockdown) conditions, phones from less deprived areas travelled further distances and to more spatially dispersed locations. However, this reversed during lockdown, with phones from more deprived area demonstrating greater mobility, exemplifying the privilege and ability to remain at home during lockdown was not equally distributed among some deprivation groups and geographic areas.