Usually a hub of freeway ramp traffic, data shows a central neighborhood now approaches San Francisco’s average.
So much of the outdoors has changed for the better since the gloom of pandemic arrived: quieter streets, more birds. The air quality in Hayes Valley has been so notcieably good since the city’s shelter-in-place order went into effect back in mid-March that I wanted to see how much better it really was.
Hayes Valley is a central neighborhood, once the site of a huge double-decker freeway bypass and today a traffic sewer of cars backed up Oak and Octavia to get on the 101 and backed up on Octavia and Fell to get off it. Add high volume north/south transit on Laguna and Buchanan and it’s easy to see why tailpipe emissions accumulate here.
The coronavirus situation has created an incredible natural experiment opportunity. I used the hyper-local data available from Purple Air’s excellent citizen-contributed sensor map to create two sets of data streams: one representing the city overall through the average of many outdoor sensors from around the city; and another representing Hayes Valley through the average of several local sensors. (Disclaimer: I’m not a scientist or statistician, see methodology at the end if you want to know more.)
Here’s a moving average PM2.5 particulate concentration across the whole city translated into EPA-standard Air Quality Index (AQI), from June 1, 2019 through April 15, 2020:
It’s quite variable. The huge spike in the fall was smoke from the Kincade fire in Sonoma. But notice that in the last month or two, the wiggles have fallen closer to the baseline in a more consistent way than usual. There are confounding factors here (rain, for example) but it’s not crazy to see that dip, which we don’t see previously, as being related to the shelter-in-place.
Here’s the same graph but with the Hayes Valley sensor data overlaid onto it:
The orange line (Hayes Valley) tracks the same basic shape, of course, but note that it’s almost always higher (worse air quality) than the city overall.
Except if you look closely, you’ll see something interesting — at the tail end, during the lockdown, it tracks somewhat more closely with the city line. Here’s a graph of the delta between the SF and HV AQI charts:
It’s clearer here. In a trend that began in early February and really accelerated in late March, there’s been only a negligible difference between Hayes Valley air quality and the air quality of the city in general.
This seems intuitively correct: there’s a lot less traffic moving through the neighborhood now, so it’s more similar to the “average” San Francisco location instead of a gridlock of idling cars pumping out tiny particles. I’m not a scientist or an air quality expert so there are other artifacts in this data I can’t explain. But it’s nice to see the current shelter-in-place show up in the numbers, and worth thinking about how to push towards more of this in “normal times” when auto volume returns.
Methodology: Purple Air reports many outdoor sensors throughout SF. I chose 13 of them to reflect a breadth of density, topography and geography. I chose ten sensors in and around Hayes Valley (only 3 are within the neighborhood as defined by the HVNA but I included several others just outside the boundaries to capture more data). I downloaded data for these sensors dating back to 6/1/19, then processed it using homebrew software to produce averages of the PM2.5 particulate number across one-day timespans for each sensor, and then averaged those into the basket numbers for each span. These basket averages per-day were translated into AQI values using the EPA breakpoints, which in turn underwent a 5-day moving average (to help smooth out weekday/end spikiness that are less interesting here). Graphs by Excel! More info on San Francisco’s fine-grained air quality from our Department of Public Health is here. You can contact me at my Medium name at Gmail dot com. Wash your hands. :-)