Luftdaten (air quality data)

[heersje]

Dear Jachym,

I would be interested to have a block which can display pm 2.5 and pm10 fine particles as measured by your own cheap self-built sensor (30 EURO - 40 USD) as comparisson to public measuring (e.g. AQI block) . There are more and more people measuring this see http://deutschland.maps.luftdaten.info/#6/51.165/10.455 it is active in several countries and languages.

kind regards

Jean-Paul

[Kristi]

I would also be very interested on a widget that covers these data!

If anyone's interested and knows how to make one there's some useful info on this link about the APIs:

https://github.com/opendata-stuttgart/meta/wiki/APIs

I also made a model widget some time ago for my girlfriend that's a programmer. She's been quite busy recently with school and work stuff so she hasn't began to work on it. If anyone feels like it I would be thrilled. Otherwise I'll obviously share my widget when I have it. The color legend is in Albanian in this example but it could be easily changed. The ID would obviously need to be changed depending on the station used.

You could take a look at the document I prepared here: https://docs.google.com/document/d/1Wtg ... sp=sharing

[Jachym]

With regards to the website - do you have a link to some API? in machine readable format?

With regards to comparison with pro stations - this is more complicated because each country has its own network, just like mine, and so it would require separate scripts for each country, and usually this data is not easily converted into machine-readable input

[Jachym]

One more thing though...

I know you guys do not want to hear it, but these cheap air quality sensors are close to useless. Without proper callibration (our stations are calibrated daily with a reference gas), without high-quality technology you simply cannot achieve good results (maybe in the future, but not now).

I just looked at what sensor this website uses and found some comparison with the reference instruments. You can see it at the graph below. You decide if this is good enough for you. The yellow line is the real value, the grey is what the sensor shows. Very often not even the trend (increasing/decreasing) is correct and when it is, it is 2-3 times off the real value.

The SDS011 Air Quality Sensor experiment2.png (50.37 KiB) Viewed 13507 times

The SDS011 Air Quality Sensor experiment.png (50.75 KiB) Viewed 13507 times

[Kristi]

With regards to the website - do you have a link to some API? in machine readable format?

I'm not sure if I understood your question correctly but there's a couple of .json files with data from all sensors. We'd have to manually set the "id" of our station.

http://api.luftdaten.info/static/v1/data.json - all measurements of the last 5 minutes for all sensors
http://api.luftdaten.info/v1/sensor/{apiID}/ - all measurements of the last 5 minutes for one sensor
http://api.luftdaten.info/v1/filter/{query} - all measurements of the last 5 minutes filtered by query

type={sensor type} : comma separated list of sensor types, i.e. 'SDS011,BME280'
area={lat,lon,distance} : all sensors within a max. radius
box={lat1,lon1,lat2,lon2} : all sensors in a 'box' with the given coordinates
country={country code} : i.e. 'BE, DE, NL, ...'

http://api.luftdaten.info/static/v2/data.json - average of all measurements per sensor of the last 5 minutes for all sensors
http://api.luftdaten.info/static/v2/data.1h.json - average of all measurements per sensor of the last hour
http://api.luftdaten.info/static/v2/data.24h.json - average of all measurements per sensor of the 24 hours
http://api.luftdaten.info/static/v2/data.dust.min.json - same as data.json but dust sensors only
http://api.luftdaten.info/static/v2/data.temp.min.json - same as data.json but temp./humidity/air pressure sensors only

[Kristi]

Regarding the quality of the sensor I understand your point and I totally agree that this can't be a mean to replace local competent authorities, but that's not what it's supposed to be. It's supposed to be a citizen program to monitor at least trends in particle matter concentration. As you might know individual single values don't really matter when it comes to calculating indexes even for the sophisticated sensors. But if you take the hourly or daily averages as a whole the data is interesting and I don't think it deviates more than 10/15% from the real value, which is a great deal considering this sensor with only $17 comes at what, 1/100th of the price? And when it comes to comparing it with sensors of PurpleAir I think it's an even better one and it doesn't have that ridiculous price. For example in a city like mine where there are no public measurements or a national agency and I don't really have the means to set up one, this could be a very useful tool.

[Jachym]

Regarding the quality of the sensor I understand your point and I totally agree that this can't be a mean to replace local competent authorities, but that's not what it's supposed to be. It's supposed to be a citizen program to monitor at least trends in particle matter concentration. As you might know individual single values don't really matter when it comes to calculating indexes even for the sophisticated sensors. But if you take the hourly or daily averages as a whole the data is interesting and I don't think it deviates more than 10/15% from the real value, which is a great deal considering this sensor with only $17 comes at what, 1/100th of the price? And when it comes to comparing it with sensors of PurpleAir I think it's an even better one and it doesn't have that ridiculous price. For example in a city like mine where there are no public measurements or a national agency and I don't really have the means to set up one, this could be a very useful tool.

I know, but look at the graphs... it does not even pick up the trends

[andyk1]

Regarding the quality of the sensor I understand your point and I totally agree that this can't be a mean to replace local competent authorities, but that's not what it's supposed to be. It's supposed to be a citizen program to monitor at least trends in particle matter concentration. As you might know individual single values don't really matter when it comes to calculating indexes even for the sophisticated sensors. But if you take the hourly or daily averages as a whole the data is interesting and I don't think it deviates more than 10/15% from the real value, which is a great deal considering this sensor with only $17 comes at what, 1/100th of the price? And when it comes to comparing it with sensors of PurpleAir I think it's an even better one and it doesn't have that ridiculous price. For example in a city like mine where there are no public measurements or a national agency and I don't really have the means to set up one, this could be a very useful tool.

Hello Kristi!

Not wanting to stick my nose in where it does not belong I have been building these sort of sensors in my spare time written in python lang.... for years always trying for different and better results. I rarely ever got the same results when ordered by 3-4 or 6 at a time none would match the others so how was I going to figure out which to use as a reference model to finish my code. Blowing different types of gases, smoke, co2 etc the results drifted till I realized I was chasing a dead horse and wasting my time. I considered getting a purple air when they first came out to see how it worked but luckily seeing the results nearly matched my home built sensors and then realizing they also used a bme280 temp, hum, pressure or altitude sensor it was the exact same sensors I became fairly competent in building and writing code for. Calibration was difficult unless I wrote a simple cailibration mode for each 10 degree (F) sorry not a metric guy here but am learning... of rise or fall but as always the next day the drift was to great and maddening where the sensors where failing to hold calibration. I did manage to figure out the results where due to wind speed and of course we all know it is never consistent. I do know the limitations of this cheap sensors also but not the point. A good quality, reliable and repeatable reading is what is needed over and over and the cheap sensors are just not up to the task. If you can't calibrate and expect it to hold it for even a day then it's a waste of time and money. When a sensor is +/- 10 to as much as 70% off at times then there's no reliability factor. I see your point it mostly a reference thing or to me a toy but that's all it is as you can't trust what your seeing. Maybe someday when tungsten strands become much cheaper (Hair thin wire that does not retain heat or cold more than a few seconds and match ambient temps very rapidly) it may become a viable hobby.

Andy