CAMBIKE SENSOR

Dear CamBike enthusiasts,

This week we can show you some amazing new results. We have analyzed hundreds of hours of data and below is a sneak preview. To get the full story, you should come to our presentation at Sensors day, it will be worth your time! The conference will be held this Friday in Fitzwilliam College, University of Cambridge. If you are interested, you can just drop by (the presentation starts at 12:10 and lasts about half an hour). The entry for you is FREE - you don’t need to register. Just mention that you are a CamBike volunteer at the entrance and they will let you in. We hope to see you there!


Data sneak preview

All the time in front of a computer was well spent - after analyzing several megabytes of textfiles, we were able to generate wonderful plots. We first looked at the time when the volunteers were using the sensors. Clearly we can see “our rush hour”, mostly between 7 and 8 am and 4 and 5 pm. We can also see the lunchbreak at 12pm!

The number of data points we included in the analysis plotted against the time of day.

Now that we know WHEN people are moving, we would also like to see HOW FAST. Easily done, here is the plot. As you can see, we get quite a lot data points when the bikes are not moving. We can also see that most of our volunteers cycle at a speed of about 20km/h.

The number of data points we included in the analysis plotted against bike speed.

But does speed influence the accuracy of our measurements? We were initially worried about the influence of wind and therefore added the tubes to the sensor inlet. Let’s see if that reduced the correlation!

The concentration of particulate matter for different bike speeds. Blue error bars correspond to the standard deviation of the PM values.

It seems so. We looked at the PM10 values corresponding to different speeds. Because we included measurements from different places and times of day in the average, the standard deviation is quite high. But we can see that the trend is relatively stable.
But this plot only showed PM10, what about PM2.5? Well, of course we can plot PM10 and PM2.5 values against each other:

PM2.5 plotted against PM10. Blue error bars correspond to the standard deviation of the PM2.5 values.

We can see that there seems to be a linear dependence between PM2.5 and PM10 for PM10 values between 5 and 20 and between 20 and 35 - at least in Cambridge over the last 2 months. We measure the amount of different sized particles, and PM measurements therefore reflect the composition of air and in addition strongly depend on humidity. We can therefore NOT use these results to always infer for example PM10 values from PM2.5 measurements - that’s an opportunity for further work!
And last but not least, can we say anything about the inlfuence of traffic on the measured PM values?

PM10 at different times of day.

Indeed, we see a difference between mornings, evenings and midday, but more data (also from †he night) is going to paint a clearer picture.


SailBritain

A CamBike Sensor is currently on its way from central London to Ipswich (by boat). A great thank you to everyone from SailBritain who made this possible (and made the picture). We are curious about the results!

A CamBike Sensor on sea.


Distribution

Distribution is ongoing. We are building sensors for you as fast as we can, but it is still a lengthy process. We will let you know as soon as the next batch is ready to hit the road!


What’s next?

We really hope to see as many of you as possible on Friday! We will show even more data and some pretty cool maps. Next week, the team will discuss further steps with focus on the distribution. Also, one of the sensors is shortly going to be tested in Argentina by Norberto Vidal - a big thank you and we hope the sensor arrives soon!


Until next time,

The CamBike Sensor Team


P.S. If you know anyone who might also like to stay tuned, let them know to just drop us an email!