Request ID: FOI-1431-2122 Date published: 25 October 2021
You asked
Follow-up to FOI-1192-2122:
Many thanks for your detailed response regarding the automated traffic counts produced for Chiswick High Road. Would it be possible to request clarification on a few questions raised by the data you have supplied?
(1) The data from the automated sensors does not seem to correlate with the manual count data previously published by TfL.
Previous communications from TfL (eg “Have Your Say” email sent 09/08/21) have referenced manual counts of cyclists on Chiswick High Road being conducted from December 2020 onward. The monthly average for weekday cyclists in April (between 6am and 10pm) has been cited as 2,125.
However, the automated data from “S62 cycle lane” and “S63 cycle lane west” yields an average for weekday cyclists in April 21 of only 1,766. Adding in the cyclist numbers from “S62 east” and “S63 east” (which is problematic - see point 2 below) gives an April weekday average of 2,716.
Therefore please could you also provide by way of comparison:
(a) the data-set from all manual traffic counts conducted on Chiswick High Road between Dec 2020 and Aug 2021; this data should include all the figures used to generate the monthly average weekday / weekend cycle counts on Chiswick High Road that have been cited in previous TfL communications.
(b) all manual counts of other vehicle types (car, bus, taxi, motorbike, LGV, OGV, pedestrians) conducted on Chiswick High Road between Dec 2020 and Aug 2021 - eg those used to generate the statistic that cyclists comprised “20% of all vehicles on Chiswick High Road in April 2021” (“Have Your Say” email sent 09/07/21).
(c) A list of the locations at which the manual counts were taken. It has been stated that the counts utilised video footage - was it taken from the same cameras as the automated counts?
(2) There seem to be a number of discrepancies in the automated sensor data which might indicate that vehicles (in particular cycles) are being double-counted between different sensors / across different lanes.
(a) the data from “S62 cycle lane” and “S62 east” would seem to indicate that c.54% (average) of daily eastbound cyclists are still using the carriageway instead of Cycleway 9. For example on 30/3, “S62 east” shows 1,088 cyclists (60% of the daily total) using the carriageway vs only 735 eastbound cyclists on C9 (as per “S62 cycle lane”), which seems unlikely to be true.
Similarly the “S64 cycle lane” data for 31/5 shows 1130 westbound cyclists; the data from “S64” then shows a further 1072 westbound cyclists on the main carriageway. Again, it seems unlikely that c.49% of westbound cyclists would choose to cycle alongside the segregated C9 path rather than on it. It is more probable that the “S64” data is double-counting a large number of the westbound cyclists from “S64 cycle lane”.
(b) There are 2 separate cameras (S64 and S66) monitoring the Chiswick High Road / Goldhawk Rd junction from different angles; presumably this will result in all westbound vehicles (inc cycles) being counted twice - for example as cycles leave King St (“S66 west”) and enter C9 (“S64 cycle lane”).
There is a similar issue at the Chiswick High Road / Heathfield Terrace junction, where eastbound cyclists crossing from the N side of the road (counted in “S59 cycle lane”) are surely being double-counted in the data from “S58 cycle lane” (which is mounted on the same lamp-post as S59).
Are there known issues with double-counting / overlapping data sets between different cameras? If so, please provide details of how TfL would propose using this data to establish an authoritative count of daily cycle / vehicle flows on Chiswick High Road.
(3) Is it possible to clarify whether the data on pedestrians included in the files “S66 path”, “S63 path” and “S59 path” includes pedestrians on both the North and South sides of Chiswick High Road, or only on the same side on which the camera is located?
(4) As per my original request, would it be possible to provide the detail on the costs (to date) of outsourcing the production of the automated traffic counts for Chiswick High Road to a third-party, and also confirm the identity of the company providing the service (presumably VivacityLabs?).
We answered
TfL Ref: FOI-1431-2122
Thank you for your further request of 30th September 2021 asking for information about traffic count data on Chiswick High Road.
Your request has been considered in accordance with the requirements of the Freedom of Information Act and our information access policy.
I can confirm that we hold the information you require. Your questions are answered in turn below:
Question 1) The data from the automated sensors does not seem to correlate with the manual count data previously published by TfL.
Previous communications from TfL (eg “Have Your Say” email sent 09/08/21) have referenced manual counts of cyclists on Chiswick High Road being conducted from December 2020 onward. The monthly average for weekday cyclists in April (between 6am and 10pm) has been cited as 2,125.
However, the automated data from “S62 cycle lane” and “S63 cycle lane west” yields an average for weekday cyclists in April 21 of only 1,766. Adding in the cyclist numbers from “S62 east” and “S63 east” (which is problematic - see point 2 below) gives an April weekday average of 2,716.
Therefore please could you also provide by way of comparison:
(a) the data-set from all manual traffic counts conducted on Chiswick High Road between Dec 2020 and Aug 2021; this data should include all the figures used to generate the monthly average weekday / weekend cycle counts on Chiswick High Road that have been cited in previous TfL communications.
(b) all manual counts of other vehicle types (car, bus, taxi, motorbike, LGV, OGV, pedestrians) conducted on Chiswick High Road between Dec 2020 and Aug 2021 - eg those used to generate the statistic that cyclists comprised “20% of all vehicles on Chiswick High Road in April 2021” (“Have Your Say” email sent 09/07/21).
(c) A list of the locations at which the manual counts were taken. It has been stated that the counts utilised video footage - was it taken from the same cameras as the automated counts?
The “manual count data” referred to used video footage from cameras, with 2 days’ 24 hour counts (Thursday and Saturday) processed each week. This was processed between December 2020 and mid-May 2021 and was used for reporting on cycling levels until mid-May (an example being the “Have Your Say” email which referenced April 2021 data, which was an average of the 4 Thursdays’ data processed). The counts referenced from this dataset include cyclists on both the cycleway and carriageway.
From mid-May onwards these were superseded by the “automated traffic counts” referenced below and in the dataset you have received, and these will be used for reporting going forwards.
Please find attached a spreadsheet giving the weekly counts for cyclists and other vehicle types between December 2020 and May 2021, and a map showing the locations of these counts. Please note:
The “20% of all vehicles” figure referenced in your question does not include pedestrians, i.e. is just vehicles,
The 6am-10pm figures are from 6am (inclusive) to 10pm (exclusive) - to select the correct hours in the work book, you need to filter and select the data including hours 6 up to 21 inclusive in the hour range, this will provide all data between 06:00 and 21:59.
Question (2) There seem to be a number of discrepancies in the automated sensor data which might indicate that vehicles (in particular cycles) are being double-counted between different sensors / across different lanes.
(a) the data from “S62 cycle lane” and “S62 east” would seem to indicate that c.54% (average) of daily eastbound cyclists are still using the carriageway instead of Cycleway 9. For example on 30/3, “S62 east” shows 1,088 cyclists (60% of the daily total) using the carriageway vs only 735 eastbound cyclists on C9 (as per “S62 cycle lane”), which seems unlikely to be true.
Similarly the “S64 cycle lane” data for 31/5 shows 1130 westbound cyclists; the data from “S64” then shows a further 1072 westbound cyclists on the main carriageway. Again, it seems unlikely that c.49% of westbound cyclists would choose to cycle alongside the segregated C9 path rather than on it. It is more probable that the “S64” data is double-counting a large number of the westbound cyclists from “S64 cycle lane”.
(b) There are 2 separate cameras (S64 and S66) monitoring the Chiswick High Road / Goldhawk Rd junction from different angles; presumably this will result in all westbound vehicles (inc cycles) being counted twice - for example as cycles leave King St (“S66 west”) and enter C9 (“S64 cycle lane”).
There is a similar issue at the Chiswick High Road / Heathfield Terrace junction, where eastbound cyclists crossing from the N side of the road (counted in “S59 cycle lane”) are surely being double-counted in the data from “S58 cycle lane” (which is mounted on the same lamp-post as S59).
Are there known issues with double-counting / overlapping data sets between different cameras? If so, please provide details of how TfL would propose using this data to establish an authoritative count of daily cycle / vehicle flows on Chiswick High Road.
2a) The automated camera sensor on the carriageway (i.e. “S62 east”) is not currently optimised for counting cyclists because it is prioritised to other modes, and steps are being undertaken to improve the detection algorithm. The automated sensor count data should be taken from “S62 cycle lane” (countline 22096) this captures the vast majority of cyclists on Chiswick High Street and count data from this location are used. This countline does not include cyclists on the carriageway and is in a slightly different location to the “manual count locations” so will be lower than the previously cited figures.
2b) Some of the sensors are located very close to each other but each are set up to count vehicles/cycles over a specific ‘virtual’ count line. It is possible that along a route a vehicle/cycle count be picked up by more than one sensor, but potential duplication is avoided by using a single countline only.
With regards to S64 please only use S64_ChiswickHighRd (22145) as this is recording motor vehicles and cycles entering the cycle lane within the threshold of acceptance and you are right that if used with S64_ChiswickHighRd_cycleln (22098) (data is also good to use from this site) it would be double counting as the count lines essentially cover the same area but are set to record different metrics.
(3) Is it possible to clarify whether the data on pedestrians included in the files “S66 path”, “S63 path” and “S59 path” includes pedestrians on both the North and South sides of Chiswick High Road, or only on the same side on which the camera is located?
S59, S63 & S66 path sensors are recording pedestrians travelling in both directions on the path on the same side that the sensor is located.
(4) As per my original request, would it be possible to provide the detail on the costs (to date) of outsourcing the production of the automated traffic counts for Chiswick High Road to a third-party, and also confirm the identity of the company providing the service (presumably VivacityLabs?).
It is not possible to provide the costs related to the collation of this data. This is because TfL is about to enter into commercial arrangements for the procurement of similar equipment and associated maintenance services. The information you have requested is commercially sensitive as release would prejudice this procurement exercise. The information is therefore exempted under section 43 of the FOI Act. We can confirm, however, that the collection of the data with respect to the sensors on Chiswick High Road is not out-sourced. TfL own the cameras and are responsible for the positioning of the count-lines, and TfL validate the counts from these sensors and are hence also responsible for the data quality provided by the sensors. Note that the use of the section 43 exemption is subject to a consideration of the Public Interest Test. In this case, we believe the public interest favours protecting our negotiating position for the forthcoming procurement, given TfL is largely funded by public money (via fares and grants).
If this is not the information you are looking for, or if you are unable to access it for any reason, please do not hesitate to contact me.
Please see the attached information sheet for details of your right to appeal as well as information on copyright and what to do if you would like to re-use any of the information we have disclosed.
Yours sincerely,
David Wells FOI Case Officer FOI Case Management Team General Counsel Transport for London
