There hasn’t been any rigorous study about the accuracy of the SDS011 so far. However, there are evidences that the measurements can represent pretty well the reality.
First, you can see pretty good matching between some sensors and the monitoring stations on a local level, as it is shown on figure 1.6 below.
Second, such matching has also been observed on an international level, as you can see on figure 1.5 below.
Third, the National Dutch Institute for Public Health and the Environment (NDIPHE) has conducted an experiment to test the accuracy of the data collected by the SDS011.They spread 110 DIY sensors in the end of 2017 in December in The Netherlands. After one month of measurements, they compared the data collected with the ones from the 40 official monitoring stations located in the Netherlands. They concluded that the correlation was pretty good after slight calibration of the data (see figure 1.7 below) and hope they can use the sensors for official air pollution monitoring in the future. Watch the video about this test here.
However, the sensors’ measurement get abnormally high when humidity gets high, and too low with dry conditions. You should thus not take consideration of the measurements when it rains nor on very dry days. In those situations the data should be either dropped or calibrated. You should also always compare the measurements with official ones.
Those clues can still let us assume that the data collected is good enough to raise awareness among citizens and give an idea of the pollution of the air. However, it is important to make a serious research about the sensor’s accuracy is the future, to know if the data is usable or not for scientific research such as epidemiology or other.
4 Comments
R.L. Kars · November 14, 2018 at 6:42 am
Overnight November 12th to 13th, pm2,5 increased gradually about 20 points (12 to 32). The only explanation is the 100 % relative humidity. This should not be possible if there are no pm2,5 producers. I think the model should be adapted for humidity affect.
Toon · November 14, 2018 at 6:57 am
Hi indeed it is a well known issue with cheap/optical pm sensors that they’re influenced by humidity. Our research and the datasheets of the SDS011 even states that humidity levels from +75 % have an influence on the readout of the sensor.
Research has been done to work out a model to compensate for these errors, but our conclusion is that this is not possible. We propose to exclude high humidity data from our dataset before we analyse the data. Above 75% rel. humidity readings of the sensor start to fluctuate too much.
Because this is an optical sensor you will never know the origin of the particle, it could be a minuscul water droplet, dust, soot, …. That is also the main reason why high relative humidity has a big influence on the sensor, it starts measuring water droplets.
Francisco Nogueira · April 7, 2019 at 4:11 pm
I’m just wondering though if this sensor is still worth using for indoor air/ dust measurement. Particularly in summer dry weather.
Toon · April 8, 2019 at 2:02 pm
The SDS011 is developed for air conditioning systems. So the answer is YES. It is better suited for indoor use then outdoor use.