what is the API and how is it calculated?

This post is intended to be a brief explanation of the Chinese air pollution index, API. This post will not include comparisons to other international systems; I will write about that sometime in the next couple of days.

The API is a single number indicating the air quality on a given day. The API goes from 0 to 500. The higher the number, the worse the air quality is on that day. China assigns an overall air quality to different ranges of APIs, as shown in this image:

air quality

(Note: the English translations here may be under revision.)

A “Blue Sky Day” is defined as a day for which the API is 100 or below; in other words, a day in which the air quality is either “excellent” or “good” according to Chinese designations.

The API is determined from only three pollutants: SO2, NO2, and PM10. The concentration of each pollutant is measured at various stations throughout the city over a 24-hour period (noon to noon). The average daily concentration of each pollutant is then converted to a normalized index using the following table:

api pollutant concentrations

Linear interpolation is performed between each set of points to determine an API for each pollutant from that pollutant’s concentration.

Once normalized, there are technically three APIs, one for each pollutant. (Note: Individual pollutant APIs for each measuring station are reported daily by the Beijing EPB on the Olympic air quality website.) Beijing’s single API reported daily by MEP is the simply the highest of the three averages. In Beijing, the highest of the three is almost always PM10.

Source: http://www.mep.gov.cn/quality/background.php

To convert from API to PM10 concentration (assuming PM10 is the highest on that day), use the following formulas:

For API 0-51: PM10 concentration = API/1000
For API 51-200: PM10 concentration = (API – 25)/500
For API 201-300: PM10 concentration = (API + 300)/1429
For API 301-400: PM10 concentration = (API + 225)/1250
For API 401-500: PM10 concentration = (API + 100)/1000

Again, for these formulas to be applicable the reported API must be the API for PM10.

8 Responses to “what is the API and how is it calculated?”

  1. Rob says:

    What’s the difference between the scale you have listed here, and the new roman numeral scale that, as you have observed, is now posted on the Olympic air quality site?

    I guess the roman numeral scale has no particular value? Or do you just expect that it will eventually change sometime before the Olympics?

  2. Dunny says:

    That’s interesting that of the 3 pollutants that they measure, 2 are chemical compounds (SO2 and NO2) and PM10 is simply a measure of any particulate 10 microns or smaller? I guess that’s why it’s typically the dominant pollutant, since it’s not tied to a particular compound? I would imagine that the tests for NO2 and SO2 use chemical reactions, where PM10 is just a filter? What is the dominant source of PM10 in beijing?

    Neat stuff — sorry to be a year behind on this one… I clicked on “popular posts”

    shoulda known better.

  3. Louie says:


    The conversion factors you list above are step functions. An API of 200 vs. 201 has a big difference in the calculated mass.

    On the other hand, if I plot the 6 API points and the corresponding PM mass concentrations, I get a linear function that should let you calculate the mass concentration equivalent of any API. Mass (in ug/m3) = 1.175x + 41.4, where x = given API. This would be a smoother result over the entire API range.

    Any reason not to use this instead, especially if trying to look at trend data that spans different ranges?

  4. Louie says:

    Actually, there is a problem with using a single function to convert all API data into mass.

    If you’re in the sub 50 API range, which is where we’d like to be, since you have to be in this range to fall within the WHO annual PM10 standard of 20 ug/m3, the function converts an API of 10 into something like 50, and an API of 50 into 100. It’s basically not possible numerically to have an API that falls within the WHO standard.

    So, maybe your conversion of API = PM10 mass in the 0-50 range is actually appropriate.

    But, moot point for you poor souls in Beijing anyways, who never see this range anyways… :)

  5. Vance says:


    It’s correct that PM10 vs. API cannot be described by a single linear function. I use piecewise linear interpretation between each point, as graphed here: http://www.livefrombeijing.com/2009/01/why-you-cant-average-apis/.


  6. […] recent string of blue-sky days may be a more common sight to behold in 2015. This past week, the Chinese capital released the […]

  7. Gerry Beauregard says:

    For what it’s worth… in cases where the PM10 dominates, the Chinese API is equivalent to Singapore’s PSI, with exactly the same piece-wise linear function as described here:


    Sounds like the air pollution is just dreadful in Beijing. Personally I find that once the PSI gets above 50 (into the “Moderate” range, as defined here http://www.nea.gov.sg/psi/), the air is already bad enough that I avoid vigorous outdoor exercise (e.g. trail running), and when it’s over 100 (very rare), I avoid going outside entirely.

  8. […] raw (otherwise they’d be in Parts Per Million), they are integers derived by the API formula, explained here. I suppose I could have done the math backwards ad derived the core values, but there are PhD […]

Leave a Reply