Spatial-temporal analysis of small-scale determinants of the COVID-19 pandemic


The spread of COVID-19 in the second and third wave was strongly affected by social structures in the population, with higher incidence rates particularly in more deprived neighbourhoods. In Germany, spatial analyses so far used county-level data to infer on area-level determinants of COVID-19, but details on socio-economic or demographic clusters on the neighbourhood-level within cities have not yet been investigated. This project aims to fill this gap. For this purpose, we will link a comprehensive set of spatial information on socio-economic, demographic, urban, and environmental characteristics with spatially and temporally aggregated data on the COVID-19 pandemic. Based on spatial-temporal analysis of the spread of COVID-19, this project will identify the neighbourhood-specific influence of small-scale determinants using geostatistical analyses and develop prediction models using machine learning methods.
To be more specific, we will first process spatial-temporal data from a variety of sources, where especially data on COVID-19 will be provided by the COVID-19 Crisis Unit (CU) (Krisenstab Corona Bremen) located at the Federal Public Health Office Bremen (Gesundheitsamt Bremen). Data on COVID-19 will be provided for statistical neighbourhoods within sub-districts of the city of Bremen for time periods of two weeks from September 2020 to June 2021. Small-scale area-level data on socio-economic and demographic characteristics, built environment, common pollutants, weather conditions, and especially mobility flows of the population will be processed to match the spatial-temporal data on COVID-19. Mobility flows are assessed by telecom network data and will be provided by Teralytics GmbH to identify changes in mobility with respect to federal or local measures against COVID-19. Second, geostatistical tools will be applied to detect spatial-temporal clusters of COVID-19. To obtain detailed estimates of neighbourhood-specific associations of small-scale characteristics and COVID-19 over time, local geographically weighted regression (GWR) will be applied which calculates spatial non-stationary associations. Third, machine learning techniques will be implemented for the prediction of the spread of COVID-19 in neighbourhoods in the city of Bremen considering underlying area-level determinants.
Expected results of this project will provide detailed insights into area-specific key drivers of the COVID-19 pandemic and neighbourhood-specific effectiveness of federal or local countermeasures. This will support local authorities of Bremen to develop specific prevention strategies accounting for vulnerable groups, urban structures, and environmental factors. Extending data processing and analysis procedures in future projects to use available small-scale area-level in other cities or in the entire country will yield more precise insights into potential targets on an area-specific level to combat future pandemics or even prevent further outbreaks.

Funding period

Begin:   October 2021
End:   December 2022


  • German Research Foundation


Dr. rer. nat. Christoph Buck