Battle of the Indices: AQI vs PSI

3 min read

Singapore Air Quality Index

When the PSI number climbs in Singapore, people like to share the link to the AQICN site.

What is AQICN? The individuals behind this project are not disclosed however they state they are based in Beijing.

One of the major reasons people like the AQI site is because whenever they think the PSI is too low, they go to AQI, see a higher number and say, “Aha! I knew it. NEA is lying to us! NEA is useless!”

But AQI is not actually the same measurement as PSI. That 230 number above is not 230 PSI. It’s 230 AQI which is a function of PM2.5 (a measure of very fine pollution particles). The PSI is 143 (which is a function of 6 different particles). If you thought these were measuring the same thing, it would seem to be a massive discrepancy.

So why do people complain about PSI so much? There is only so much you can do with a single number. It cannot account for all altitudes (eg void deck vs 17th floor), all locations, spikes due to short term wind changes, some pollutants increasing while others decrease, etc.

The only truly local real-time measurement you have is your nose. Everything else will be off for some reason or another.


The other issue is that NEA does not provide a PSI forecast but AQICN provides a 3 day outlook. The idea of a forecast seems to provide more comfort than the forecast itself which has severe limitations:

  • it is not precise: it is only a rough indicator of good/bad
  • it cannot predict short term spikes
  • it is not a forecast of PSI so most people don’t really have a feel for what it is predicting.

Today, for instance, the haze spiked. Officially the 3hr average peaked at 161.  But it probably was 180+ for about an hour near where I live before clearing up quickly.

A forecast cannot tell you about these spikes even though such a spike may be what triggers a health crisis. In my case it triggered me shutting all the windows and turning on the aircon. My main monitoring device? My nose.

So when would an NEA forecasts actually be helpful?

  • Is it for deciding whether to leave your windows open or not? No. Use your nose.
  • Is it for deciding to carry a mask? Maybe. But there may be spikes. Keep one handy until the hazy season is over (by mid October).
  • Is it for planning a picnic or outdoor activity tomorrow? Yes. This may be a good case for it but be sure to add  a grain of salt.
  • Is it for planning an activity further out? No. There is limited information available either way.

So why don’t they provide a forecast like AQICN?

AQICN is only trying to estimate future PM2.5 and the forecast they provide is tantamount to licking your finger and sticking it in the air.

In order to forecast PSI, NEA would have to forecast 6 different pollutants up to 36 hours out. This is extremely difficult if not impossible. Furthermore, NEA is unlikely to be willing to undertake such a rough calculation because of the criticism it would come under if the numbers were off. As it is, there is constant criticism of the PSI number based on actual pollutant readings. Estimating a PSI which so few people trust would seem to be a fool’s errand.

Open Door Policy

I don’t envy NEA’s task – having to come up with a number that gels with every individual’s subjective experience of pollution.

Saying that, as long as NEA keeps its data private, people will not understand that the devil is in the detail. They should make the raw readings available to the public and even put it in an API. Let the community take a crack at making it as useful as possible. It will generate a lot of conversation about how difficult the problem is, but may also make it possible for open source tools to be developed which lets people get a more personalised PSI reading.