| dc.description.abstract | The impact of environmental variables like air pollution and greenness on health is studied using personal human exposure assessment. Studies rely on of static exposure assessment, only using the residential location of a population, or apply data-rich techniques, which rely on GPS or portable air pollution measurement devices to measure the exposure for often a small population over a limited time span. Long-term population wide-exposure assessments have often low data availability because of a lack of human mobility data. In this study, a method is established for a high-resolution personal exposure model for NO2 and greenness in a sparse space-time activity data situation. Downscaling methods and exposure assessment techniques are developed and evaluated to study environmental exposures in Utrecht and Madrid. To establish a high-resolution personal exposure assessment, population datasets are downscaled using OpenStreetMap building data from Utrecht and Madrid. Because there is no global human mobility data set available, human mobility is simulated for three social-economic groups, commuters, homemakers and students, with different mobility patterns. Activity is represented in zones of potential activity and using weighted buffers to represent how likely a place is visited. The exposure assessment shows that commuters have the highest NO2 exposure and lowest NDVI exposure and commuters and students have a lower range of exposure compared to homemakers. When population data is downscaled there is an negligible difference in residential exposure between the two population datasets. Together with the considerable difference in the mean error between the different downscaling methods, this indicates that the location of the residences is more important than the resolution of the population data. Sensitivity analysis shows that there is only a minor effect (max 2%) in changing the weights of the buffers while changing the size of the buffers influences the exposure up to 18%. This study indicates that downscaling of environmental information is of major importance for exposure assessments while the weights used for different buffer sizes has relatively limited effect on calculated exposure values. | |
