Deliverable 3 (linked to output 1). Health impacts

[Robinlovelace]

the output of the previous stages will be combined to enable comparison of scenarios, extending from the calibration procedure underlying HEAT and other methods used in similar tools, such as the Propensity to Cycle Tool.

[mpadge]

@Robinlovelace Exposure layer uploaded into releases of who-data. You just need to get that as net, and run

net <- merge_directed_graph(net) %>% dodgr_to_sf()

It has columns for:

• "flow", which is a first cut of expected pedestrian flows per day;
• "centrality", which is a scaled (0-1) measure of network centrality for automobile travel; and
• "exposure", which is the absolute number of lives per year lost through increased exposure associated with a 1% increase in walking (but no reduction in car travel implemented).

I'm not yet sure how the "exposure" layer scales to the city aggregate, but the values are real for now, and range up to just over 1, and very clearly show the road sections with life-threatening emissions.

The only real additional assumption made in all of that was applying a Gaussian dispersal kernel to vehicular emissions, for which I entirely arbitrarily chose a width of 20 metres.

[Robinlovelace]

Great, could you try integrating that into upthat? Not sure what the units are and wondering if it would make more sense as an additional column on the existing network data...

[mpadge]

Exposure is already simply embedded as an additional column on the existing data, so there's just a single file. The units are something I'm working on - it's very interesting pondering that, and something I'm very interesting in getting WHO feedback on. It essentially boils down to a discrepancy between most current approaches to modelling exposure, which are inherently and utterly individual-based, and so derive a variety of means to assess individual exposure. Those units on the network are, however, not individual units, rather they are aggregated across the entire population. Is exposing 1,000 people to 1/1,000th the amount of pollution ultimately as "harmful" as exposing 1 person to 1 unit of pollution? That kind of collective or aggregate approach to air pollution effects is rarely explicitly considered, obviously because the data needs are analagously 1,000 times greater. Until then ... we just need to think a slight bit harder about what those kind of units might actually mean.

[Robinlovelace]

Good questions. I think estimates of air pollution levels in standard units, e.g. in units used in WHO guidelines, is a good place to start: https://www.who.int/news-room/fact-sheets/detail/ambient-(outdoor)-air-quality-and-health

From there time/severity distributions could be created for different pollutants I imagine. But agree will be good to get feedback on this.