Implications for future research
We argue that NDVI and other vegetation indices are not just unable to differentiate vegetation types but are also easily confounded by urbanicity and thus air pollution. When investigating associations where knowing the amount of greenspace or the distance to greenspace is not enough, like with allergic outcomes, we suggest that more detailed vegetation measures be considered. In our study, number of trees appeared to be a good proxy for number of allergenic trees, but this unlikely to always be the case across study areas. Digital tree registry data are scarce but available for some cities, including New York, Frankfurt am Main, Vienna, Hamburg and Melbourne. Similarly, rich datasets are available for different areas, like the maps of 12 key allergenic taxa in the UK for the year 2012 in 1 x 1 km resolution produced by McInnes et al. 53