Introduction to using seasonal correction factors

As a recommendation for these guidelines, we recommend the use of caution in going to the extent of assigning road groups to sections with a view to giving seasonal correction factors to count data. This should not be a default position but more an approach deliberately chosen by the asset manager as being appropriate to specific parts (or all) of their network.

Be sure the introduction of these measures is adding value and not just confusing the picture. There is a level of uncertainty and variation in count data by its very nature. Therefore any estimates should be treated with an appropriate level of confidence, as discussed earlier. Unless a clear pattern or link can be ascribed from the profile to the count site then consideration should be made as to its value. If counts vary significantly through the year, then maybe a manual assessment is preferred to an “automatically assigned” process.

There are also methodologies to mitigate the seasonal and weekly variations and provide more consistency to the count data. Examples include limiting count programme times to with school terms. One can also prescribe count sites be counted the same time each year for example for sites.

Continuous Measurements and Seasonal Adjustments

There are two purposes for continuous monitoring identified as part of this process:

• Monitoring to inform seasonal changes through the year at points on the network to improve our knowledge of network characteristics for asset managers and planners for example
• Monitoring used to adjust counts into an AADT estimate (scope of these guidelines).

The first is an important part of the traffic count strategy. Continuous monitoring can be used as control sites to understand seasonal and weekly impacts at key sites. These can be permanent sites recording continuously or sites counted with a high level of frequency. Hasting DC, for example, uses a form of continuous monitoring where they have a rota of sites counted for one week of every month for a year to monitor seasonal changes across the year such as ebbs and flows of horticulture activity on the network. Such information is useful for traffic planners and asset engineers alike.

Second is the ability to use continuous monitoring to adjust our counts into AADT estimates. This is the use we discuss as part of these guidelines. Obviously, traffic volumes will ebb and flow for a variety of reasons throughout the year. The use of adjustment factors can adjust a weekly count to a typical average flow for the year. However, different sites will vary through the year for different reasons. The continuous sites therefore need to reflect the sites they are being linked with to add any degree of improvement. This is difficult to achieve without a programme of sites with a statistically reliable correlation to the links you want to adjust. This is expensive to initiate and expensive to maintain. It could be applied to a selection of key routes and be viable. But in this case, it would be recommended as a specific response on certain networks, rather than a general recommendation of minimum practice for each local authority.

The other key consideration is accuracy. Mike Alford from Auckland Transport undertook an analysis focussing on this issue and found no real pattern from which to apply widely across many links. The issue of a poorly linked adjustment is that the adjustment is not helpful and may make the AADT estimate worse. It also has the added complication of appearing more accurate through application of the adjustment. Unadjusted volumes are clear in what they are and the level of confidence therefore they should hold.

The seasonal variation factors detailed in RAMM for the original traffic estimation tool show the road types adjustments show a variation around +-10% for urban road sections. Rural sections show an adjustment of +-10 – 15% except for the recreational road types. These would be peak seasonal routes accessing summer holiday spots or winter ski areas for example.

The graphs are based on data reported in the Transfund New Zealand Research Report No. 205 Guide to Estimation and Monitoring of Traffic Counting and Traffic Growth released in 2001. This report looked at the calculation of AADT using adjustment factors followed in the RAMM TCE tool.

For typical urban and rural corridors, the uncertainty of the AADT is not outweighed by the application of the seasonal adjustment. Most Councils pragmatically avoid counting over holiday periods as part of their contractual requirements and use those counts as the basis for their estimates.

The asset manager can set up their own profiles from continuous sites on their networks or set a profile from experience they see as applicable.