are you wondering what the different colors and numbers below mean?
may 5th 2020
april 28th 2020
back in 2008, when the world air quality project started, all of the existing air quality monitoring stations where based on highly professional and expensive bam and toem technology. this kind of stations are always maintained by professional and highly qualified operators - and this ensures that the output of this station can be trusted.
12 years later, the bam and toem stations still do exist. but with the development of laser scattering technology and low-cost dust sensor, the bam and toem stations are now outnumbered by immense and dense low-cost sensor networks. nowdays, there are just so many of those networks - to name a few: urad, luft-daten, airqo, airbg, opensense, yakkaw, econet, airkaz, ccdc, ambente, green air, etc..
january 10th 2017
univ. of arizona ()
a common question asked is why are nitrogen dioxide (no2), sulphur dioxide (so2) and carbon monoxide (co) concentrations always so low and is measuring them useful?
to highlight the importance of monitoring these substances this article is going to explain where these pollutants come from and how to they affect our health and the environment.
to keep things short this article will just focus on no2.
august 10th 2016
univ. of arizona ()
it is already several months since the us epa updated its aqi standard for ozone 8 hours breakpoints. this is definitely a positive step towards strengthening air quality in the usa, since the us epa standard was actually among of the least conservative standard for ozone, compared, for instance, to similar standards in india, china or europe!
but more interestingly, recently took a step further by introducing the ozone formula, which allows to use the weighted 8 hours breakpoint standard in place of the 1 hour breakpoint. this change, which the airnow team did not widely communicate to the public, makes quite a change to the way ozone aqi is reported, and this is what we will be looking at in this article.
february 28th 2016
during the past months, the world air quality team has been working on analyzing several new air quality forecast models, as well as improving the air quality forecast model demonstrator.this article will present the latest forecast model demonstrator, which is based on the gridded population of the world (), and which will be applied to analyze the air quality forecast for the northern india region (including bangladesh, pakistan and nepal).
november 5th 2015
a perfect dust storm ()
october 28th 2015
it is almost two months since our colleagues from the have been making an interesting move towards promoting the idea that air quality is as important as any other business within the united nations. they started to invest in this domain, and came up with the excellent initiative to create an open footprint for affordable air quality monitoring stations.
september 20th 2015
this is a small contribution form the world air quality index project to the asmc - asean specialised meteorological centre () during this heavy period of haze in south asia.
september 18th 2015
map overlay process
we have been recently cooperating with quite a few more epas worlwide. among all the requests which we have had, it is actually to provide a web map service so that the epa themselves can include the real-time air quality data onto the map from their website.one of the obvious advantages of using such a map web service from us is to benefit from the transboundary data which we are curating, i.e. a world wide air quality information is provided without country boundary limits.
the second advantage is not only that this is a free service, but it is also completely compatible with standard map technologies from google, bing or leaflet, which allows a quick and easy integration with any existing website.
september 6th 2015
the ambient , refered as o3 and also called called ground-level or ozone impacts everyone on earth regardless of the country, as shown on the image on the right .
unlike particulate matter (pm2.5), the ground-level ozone is not emitted directly. it is instead produced through a series of chemical reactions that occur in presence of nitrogen oxides, volatile organic compounds, sunlight and high temperatures, as shown on the following visual:
quantifying the impact of this ground level ozone on health is done via the air quality index standard which each countries defines. what is interesting is that half of the world is using a standard based on milligrams measurement, while the rest is using ppb based measurement. but is this really a problem? this is what we will be looking at in this article.
september 3rd 2015
inspekcja ochrony środowiska.
(inspection of environmental protection)
the air quality data for has been available for a while, but we recently got few questions about the aqi scale used by the inspekcja ochrony środowiska, i.e. the regional inspections of environmental protection, and how it differs with the scale used on the world air quality index project. one of them is from sylwia, who asked:
i am confused with all those numbers, and why there is a difference with the numbers shown on the warsaw epa website.this is actually a very good question, since, when it comes to data, no one should take data as "granted" (refering to the excellent from on 'own your body's data'). this is also the reason why the world air quality index project exists, i.e. to explain every one how to understand the number behind air pollution.
could you please kindly inform me about the source of the data in warsaw and the reliability of the data you provide with?
july 28th 2015
in order to improve the integration of our aqi widgets within external websites, we have developped a new widget api feed. compared to the existing widget feed, this new api feed provides much more flexibility for the integration, but also many more options for customzing the widget appearance.
july 18th 2015
real-time air quality data for is now available for indonesia. 10 stations located in major cities across indonesia are measuring in real-time pm10 pollution.
the monitoring stations are operated by (badan meteorologi klimatologi dan geofisika), the indonesian department of meteorology, climatology and geophysics.
july 16th 2015
the fleurdelisé, flag of quebec.
the air quality data for and has been available for a while, but we recently got few questions about the aqi scale used in this part of canada. one of them is from marie a., who asked:
i would like to know why the air quality index on your web site is different from the aqi on the montreal's website? do you recalculate anything to meet the us epa aqi standards? if so, how? with what data?this is indeed a very good question. and we will explain in this article the different scales in use, how the scales compare to each other, and how the convertion can be done.
july 1st 2015
thanks to the world air quality index project community, translations for the are now available for the website and web-application. many thanks especillay to the user "airs" for the excellent contribution.
june 8th 2015
beijing citizens do know for long that the best natural way to clean the pollution in the air is strong winds from the north, while winds from the south might, on the contrary, bring more pollution from hebei.
in , ecuador, there is a similar problematic, except that the pollution comes from the active . fortunately, they do not erupt that often, but when they erupt, better avoid being under the the heavy formed by the explosive eruptions (see picture).
june 3rd 2015
it is already a long time since the hong kong epa has updated its air quality index standard from the traditional air pollution index to the so called air quality health index (aqhi), but we never got a change until now to have a deeper look at it, which we will do in this article.
the hong kong epa has actually done an excellent work at explaining how the aqhi is defined, and the information one needs to understand the aqhi is available from their . the most obvious simplification is the change of the traditional 0-500 index into a new simplifed 0-10 range, sub-divded in five health risk categories, as shown below:
but there is actually much more than a range simplification: the new aqhi is also introducing the very clever concept of added health risk (%ar), which was originally , in 2007, by a team of four partners from south africa, and first used in .
may 30th 2015
the air quality forecast for south america is now available on the , in parnership with the brazilian centro de previsão do tempo e estudos climáticos (). the forecast can be accessed freely accessed from as well as . the cptec team has done an excellent work at providing a simple and detailled on the underlying concepts used by their forecast model, called (for brazilian developments on the regional atmospheric modelling system). the picture below (from the cptec) is giving an insight on the workflow for their model (source -> transport -> deposition).
may 18th 2015
the ground ozone air quality index calculation has just been updated to follow the instant cast concept, i.e. to report the pollution right now instead of the pollution from the previous hours. for more information about the instant cast, please refer to this article.the calculation for the ozone index is now using the hourly reading, instead of the previous 8 hours average, but still applying the same 8 hours aqi breakpoints formula for aqi below 100. the previous for calculating the 8-hour average concentrations is not used any more for instant reporting. above aqi 100, the normal 1 hour ozone breakpoints are used (while before, the aqi above 100 was defined as the maximum of the 1 hour and 8 hours readings).
may 16th 2015
it is already more than one year since the world health organization has released their ' database 2014'. the who team who worked on this project have done an excellent work at gathering the data for so many countries, and putting it all together in a single place accessible for all.we wrote about it in several of our past articles (like for the air quality or in ), but we never had the chance to look at the different options to visualize their data set. so, this is what we will be writing about in this article.
may 15th 2015
the indian central pollution control board has recently announced, in april 2015, the publication of india's national air quality index standard (naqi ). all the details on this standard are available from , or from this from indian institute of technology, kanpur ().this is actually a very important step forward in the air quality world awarness, since the air pollution in each country is very much specific to the country pollution type: for instance because of dust (in arabic peninsula), because of transport (new delhi) or topography (chile).
may 10th 2015
latin america, just like any other continent in our world is facing air pollution. the recent 2014 report by the on ambient (outdoor) air pollution gives some annual mean figures for several cities in latin america. without surprise, the worst scenarios are in asia, but yet, some cities in latin america have high indexes too (see table on the left).
the ranking from the who should however by considered carefully, especially considering the measurement methods used the report can be argued - for instance, for lima, the peruvian epa is arguing that the data from the who is not correct because the samples were not taken in the right places (read this article from for full feedback from several countries). this is actually for this reason that we only publish real-time information on the world air quality project since we do believe than more than ranking, it is the current conditions that are important for the world citizen.
may 2nd 2015
air quality for malaysia has been available for a long time, but unlike other countries, only the composite aqi is provided. the aqi for the individual pollutants such as pm10, ozone ... is not available, which makes the convertion to the us epa scale more difficult.
fortunately, the information about the aqi scale used in malaysia, also known as api (for air pollutant index) is clearly explained in the document ''.
april 28th 2015
many of the countries in the balkans are providing real-time air quality readings. the data for (thrace), and have already been available on the world air quality index project for a long time. today, we are able to roll-out the data for 3 new countries: , and .
april 26th 2015
we had the honor to be invited by from the nasa applied remote sensing training () team to give a presentation about our world air quality index project at the earth observations and tools for air quality applications
april 8th 2015
when talking about air quality, the first countries that comes to people's mind are china and india, and more generally asia. even on the world air quality index project, asia has always been, since the begining, our strong focus (most likely because this is where our hq is located!).
few years ago, in 2012, we decided to extend our scope to cover the other , starting with , and then following with and , . but one of seven continents, namely africa, has been surprisingly lacking behind. we received many inquiries about this lack of data, and one of the last was from said e. who wrote:
thanks for making the world air quality index project site. it gives very useful information.
my only concern is why africa is not considered.
for information i am a moroccan citizen.
march 23rd 2015
in weather prediction, forecast models are used to predict future states of the atmosphere, based on how the climate system evolves with time from an initial state. while the forecast models are quite (and do require strong scientific and engineering capabilities), the science of analyzing those forecast models, and verifying their , by comparing actual empirical observations to predicted values, is quite straightforward.
for the domain of air quality, just like for weather prediction, it is possible to define models used to predict the future set of atmospheric pollution. there are actually , often referred as . and just like weather prediction, the same concept of accuracy analysis can be applied to atmospheric pollution predictions.
march 20th 2015
very often we receive inquiries from users asking about the numbers published on the world air quality index project, and wondering about the difference with the values expressed in other websites. and that's actually normal, since the numbers, also called air quality indexes (aqi), can be very confusing for beginners. the last message we received is the following one:
we have been checking air quality index for the tianjin, china since last year, however we found the data has such a big difference between your website and another official (air.tjemc.org.cn) one which is stated on yours as well. i am wondering which one can provide more accurate data, as it will affect the decision for outdoor or indoor. please help me with this issue. thank you!we have been writing quite a lot about this issue before, for instance from , and more recently about the nowcast. but, because we are still getting many questions, we are now writing this article to explain the difference of the aqi scales used all over the world.
march 15th 2015
among all the topics that were addressed, there is one worth writing about, namely the "nowcast" system. this system used by the us epa to convert the raw pollutants readings, expressed in µg/m3 or ppb, into the aqi (scale from 0 to 500). it is used for all the aqi values reported on the website.
the concept behind the nowcast is to compensate the "24 hours averaging", which should be used when converting concentrations to aqi. the reason for this averaging is that the specifies that each of the levels of health concern (i.e. good, moderate,... unhealthy...) is valid under a 24 hours exposure. for example, when seeing a 188 aqi (unhealthy), one need to read it as "if i stay out for 24 hours, and the aqi is 188 during those 24 hours, then the health effect is unhealthy". this is quite different from saying that "if the aqi reported now is 188, then the health effect is unhealthy".
february 23rd 2015
as there is also a huge demand for pm2.5 monitoring for (especially chiang mai and bangkok), or for more station in (ho chi minh city), the extension will also cover selected countries with existing network.
but, this time, unlike for the previous countries, we do need the help of everyone, and especially your help if you are living in one of those countries. the reason is that, this time, since the real-time air quality information is not publically available from the environmental protection agencies (epa) of those countries, the world air quality index project will start to crowdsource air quality monitoring stations in the main cities of the above countries1.
visible air pollution on the manilla and jakarta skylines
february 15th 2015
we recentely had the priviledge to start a cooperation with the .; the objective is to use the remote air quality sensing based on the nasa satelites to determine the air quality in areas where sensors are not available (e.g. over the sea, but also for countries where sensors are not yet available).
february 8th 2015
earlier this week (of february 2015), we got several inquries from hanoi citizen wondering about the air quality index which was displayed on the hanoi page ().
the main reason for the inquiry was that the air quality index displayed on the world air quality index project was around 50 (green aqi) level, while the outdoor visibilty in hanoi looked like this:
october 28th 2014
the latest version 3.2 of the world air quality index project android application has just been released, and comes with few new cool features: the embedded map which will let you navigate through the world air monitoring station directly from the app and the wind and temperature forecast tab which gives you the coming 3 days forecast. we have also been working on refreshing the visual and updating settings to support, for instance, fahrenheit temperature display.
and wind conditions
when aqi > 150
the nearest stations
october 6th 2014
like every other year since 2002, our collegues from (aka clean air initiative) are organiznig the biennial (). this time, it will take place in sri lanka, from november 19th to 21st 2014, and will be co-hosted together with the environmentally sustainable transport (est) forum.
the "integrated conference of baq 2014 and intergovernmental 8th regional est forum in asia" co-organized by the , , , , and , in partnership with the , , and .this year theme is "next generation solutions for clean air and sustainable transport − towards a livable society in asia":
recognizing that clean air and sustainable transport are essential to a livable society in asia, we call for innovative and smart solutions (policy, institution, technology and financing) that significantly reduce air pollution and greenhouse gases from energy, industry, transport, and area sources, and ensure a safe, equitable, environment and people-friendly transport system by accelerating the shift towards more environmentally sustainable transport (est) in asian cities and countries.check out the . more than 800 participants are expected to join the baq 2014, and we will be part of them, presenting our recent research on we have been conducting within our .
september 6th 2014
august 1st 2014
it *claims* to be able to detect and measure both (voc, i.e ozone), as well as the famous small pm2.5 particulate matter. but can it really do so? to know the answer, check our latest on-going experiment about the air box at
july 21st 2014
october 21st 2013
the is still going-on, and has been on for now 50 days. so, last weekend, as the air pollution in beijing was getting really low - thanks to strong north winds, we used this time as an opportunity to perform a maintenance break... on the dylos monitor itself.
the reason is that, even if the dylos monitor is definitely well designed and engineered equipment (we do mean it), it maybe not be so well conceived to subsist beijing's tough episodic air pollution.
it all started after noticing that the correlation graphs between the dylos and bam monitor tend to show some systematic signs of leveling beyond 38,000 particles (≥0.5µ). this kind of saturation symptom can be the consequence of internal air inflow bottleneck, which can be caused by "saturated" air channels. during times of heavy pollution, this phenomenon is totally normal due to the large amount of particulate matter having to go through the air channel.
so, no surprise, after opening the dylos monitor, we confirmed that the internal air channels are definitely saturated, by an accumulation of dust during the last 50 days:
-beijing dust impact on the dylos air particle counter internal structure-
the result of cleaning the internal structure, using a fine brush, would produce an astonishing amount of dust - and this, just after 50 days of outdoor experimentation. just think that this is what you are breathing...
-the real beijing dust... the one beijiners breathe during heavy pollution day!-
actually, having to perform such cleaning and maintenance on the dylos equipment is not a surprise, since even the very expensive bam equipment requires monthly maintenance by skilled operators.
it is now 10 days since the cleaning has been done, and the situation has definitely improved for the correlation between bam and dylos. from the graph below, which shows the measurements before and after the cleaning, the zone indicated with number (2) shows a very good similarity (correlation) between bam and dylos. but, under condition of heavy pollution, there are still consistent signs of leveling (deviation in measurement), as highlighted with zone number (1) and (3). the deviation seems however less worse after the cleaning (zone (3)) than before (zone (1)). we are contacting the dylos company, to consult them on this leveling phenomenon, and will update this post later.
(click on the picture to open the real time version)
september 9th 2013
for some historical reason, the aqi convertion formula used for some of the cities on the world air quality index project were not using the . (this is only relevant for cities in asia, like japan and taiwan. other continents (america, europe and australia) have always been using the 2012 standard.
the main change is this updated epa standard, published on december 14th 2012, compared to the previous one (published in 1999), is to reduce (limit) the pm2.5 aqi level 50 from 15µg/m3 to 12µg/m3, and this in order to enforce more strict air quality standards.
since september 9th 2013, all the countries are now using the same pm2.5 aqi conversion scale, which is reflected in this table:
september 8th 2013
the bam working principle is quite simple: it measures the decrease in the number of beta particles (electrons) transmitted through a thin layer of particulate matter (pm). as the thickness of the pm layer is increased, the lower the number of beta particules can go through.
the only problem with this kind of bam equipment is that the cost, estimated to 120k cny (15k eur). in china only, there are around 500 of those (arround 10 station for each city), and yet, this is not enough as there are still many cities without any pm2.5 readings. in thailand, there are 21 automatic (i.e. with hourly reading) stations, but none of them provide pm2.5 data. and in vietnam, there are only 3 stations for the entire country.
so, while waiting for all the world governments to invest in brand new beta attenuation monitors, is there anything that could be done? then answer is yes! there are actually relatively affordable equipment that can be used to measure the pm10 and pm2.5 based air quality. the most famous one is the dylos air particule counter:
if you want to know how the dylos air particule counter is working, please refer to the excellent reverse engineering work from , or simply from dylos's website (). you can also refer to the resarch paper to understand the theory behind the measurement.
please, note that you must get the pro, with pc interface version. the pro is needed to be able to measure particules as small as pm2.5, as the non-pro version can only detect pm10 particules or larger. the pc interface version mandatory as it provides a serial interface, needed for automatically extracting the data from the dylos monitor to the pc.
if you buy the equipment from china, from taobao or jndong (jd.com), please be carefull and make sure to order the right version. the most common version in the dc 1100 without pc interface. only few shop on taobao provide this right version, and at the present time of writing, jd.com does not even provide the correct version. if you need help for buying for equipement from china, you can contact us with the subject "dylos equipment order", and we will help you for the ordering.
in order to automatically extract the data to send it to the server, a raspberry pi ($56) and a serial to usb cable ($9) is used, totalizing a bom of $375. (you can buy all this material from amazon by clicking on the images below)
| dylos dc1100 pro|
with pc interface
|serial to usb cable||raspberry pi|
the documentation and scripts for connecting the dylos particules counter to the raspberry pi, and automatically uploading the dylos data to the world air quality index project server are available freely from github: . if you have any questions, you can contact us with the subject "dylos experiment".
also, please note that when you setup your dylos particule counter outside, you need to be carefull where you locate it. ideally, it should be in a place which is not too windy, as the wind can have a significant impact on the readings (the reason is that the dylos counter is using a simple fan to get the air inflow, and the volume of air inflow per minute needs to be constant in order to have accurate readings ~ this simple fan is one of the reason why this kind of material is so affordable compared to the professional metone bam sensors).
in order to compare the dylos reading with the bam reading, we use the bam data provided by beijing chaoyang agricultural exhibition hall (朝阳农展馆), as our station is located in the sanlitun soho area.
and here is the real-time result: the line in blue is the dylos reading (updated every minute), and the line in red the bam data (updated every hour). the first graph is to compare the pm10 reading, while the second graph compares the pm2.5. also, please note that the bam data is provided in µg directly, i.e. not converted to the aqi grades.
(click on the picture to open the real time version)
as you can see, the similitude between dylos and bam reading is quite high, but yet not always fully comparable. in the second part of this post (which we will write after two month, once we have collected enough data), we will provide more statistical data about the readings.
also, if you are interested in joining this experiment, by setting up your own dylos station, please contact us. we will provide all the scripts and detailled setup, as well as to add you the sensor database, so that you can see your readings in real-time.
as a follow up of this post, you can check the new post about the .
june 25th 2013
with the recent affecting singapore and now , we got many questions about why there is a difference between the data which can be read from singapore nea website () and the world air quality index project .
for instance, here is what can be read today in nea website:
may 20th 2013
shenyang city aqi data is now based on the maximum of the , and . for more infomation, please refer to the article about
february 7th 2013
for historic reason, the world air quality index project has been using the china mep (ministry of environmental protection) designation for the air quality levels. this is now past history, and the designation will follow, from now on, the epa (environmental protection agency) .
the table below summarizes the differences in designation between the two standards, as well as the translations for the various languages.
|aqi range||old mep designation||new epa airnow designation||new chinese translation||new japanese translation||new korean translation||new russian translation||color coding|
|51-100||good||moderate||良||良||초록||неподходящее для особо восприимчивых|
|101-150||lightly polluted||unhealthy for sensitive groups||轻度污染||軽微汚染||노랑||плохое|
|201-300||heavily polluted||very unhealthy||中度污染||中度汚染||빨강||очень нездоровое|
please note that for now, this is only the designation for the website which is updated. the designation for the android application will follow soon.
february 3rd 2013
actually, this is a very good question! if you have to wear a mask in beijing, then better to make sure that it really filters the bad particulates. since i did not have the answer to his question, i just dropped an email to francis chu from the singapore ® mask manufacturing company, who replied in almost no time:
yes, our totobobo mask is able to cut down 99.85% of 0.1 microns particles.for the readers not familiar with particles size, pm10 refers to particles smaller than 10 micrometer, and for pm2.5, it refers to smaller than 2.5 micrometers. so, filtering out all the particulates smaller than "0.1 microns (same as 0.1 )", means that it can filter particles 25 times smaller than the harmful pm2.5! so, no worries klaus, the newest technologies are definitely efficient!
you can download the lab report:
for more information about particles size, you can refer to the page. if you want to learn more about how air filters are actually tested, or how the various mask brands compare to each other, check this very informative entry.
and if you just want to get your own totobobo mask, .
february 2nd 2013
we got a very good and relevant question this week from severine p., who asked about pm2.5 concentrations versus pm10 concentrations. what severine asked in her mail was:
i don't understand why pm2.5 concentrations are often higher than pm10 concentrations.severine's question is totally correct: pm2.5 are, by definition, particulates smaller than 2.5 micrometers, so indeed, they should also be included as smaller than 10 micrometers.
aren't pm smaller than 2.5 micro included in pm smaller than 10 micro?
thank you for your answer
but the assumption about concentration is not correct: pm2.5 concentration can be smaller than pm10, but yet, the converted aqi value can be higher. for instance, an aqi of 50 for pm2.5 corresponds to 15.5 mg/m3, while it corresponds to 55 mg/m3 for pm10.
so, even if the full pm2.5 15.5 mg/m3 are added to the pm10 concentration, the pm10 aqi still remains much lower than the pm2.5 aqi, for instance, earlier this week, the 东城东四 pm10 concentration was 366, resulting in 216 aqi, while the 东城东四 pm2.5 was 348, resulting in 398 aqi.
in order to double confirm that our understanding is correct, we did contact the world renown air quality expert from , who wrote us back:
you are right on these. the differences are due to the differences in the break points for pm2.5 and pm10 and how the epidemiological data is correlated to each of the fractions.
so, the next questions is about the relevance of the pm10, especially in china. it is totally right that most of the time (empirically confirmed), pm2.5 is the dominant value in the aqi. so, do we still need the pm10 measurement? is it still relevant? and which conditions are reflected when pm10 aqi is higher than the pm2.5? we asked the question again to , who replied:
there are now new studies presenting evidence that pm2.5 is more harmful than pm10. physically, this makes sense - smaller than particle, more probability that it will go deeper into the lungs and harm us. this is also one of the important reasons for who to push for all countries to have standards for pm2.5.many thanks, dr sarath, for the quick and professional answers.
to the question, do we still need to have pm10 measurement, when we are doing pm2.5 - yes we still do. while the fraction of pm2.5 is higher in the pm10 fraction, for the most cities with pollution from transport and other combustion sources, an often neglected non-combustion source is dust re-suspension (from road dust and dust storms), which forms part of the coarser fraction (pm2.5 to pm10). if we suddenly stop measuring pm10, we will be neglecting this source.
most cities still measure only pm10 - in china and in india. so, another reason we cannot remove it from the equation.
january 19th 2013
shanghai aqi data is now based on the maximum pm2.5 aqi data for 3 stations: ,
the first station is maintained by , while the two last ones are provided by , which is governmental center.
if you have concerns about some stations reporting lower pm2.5 data compared to others, then refer to the following graphics; they show the relative pm2.5 reporting for the last 5 days for each of the above mentioned stations:
january 11th 2013
for more information, please refer to the shanghai aqi update explanations. the following graphics show the relative pm2.5 aqi data for beijing stations on which the overall beijing aqi is based.
january 11th 2013
for more information, please refer to the shanghai aqi update explanations. the following graphics show the relative pm2.5 aqi data for chengdu stations on which the overall chengu aqi is based.