Using groups in updating estimates for volume growth

The aspect of accounting for growth to update estimates by averaging the growth rates from sections with a count history seems relatively sensible on the surface. However it assumes that growth rates across a group are relatively similar which, to be fair, is seldom the case. Analysis has shown that even a group of similar road types in a same location struggle to show any consistent growth pattern that could be said to represent the group. This may as much be to do with the variability of count data from week to week. None the less it is a valid point. To a certain extent, the growth rates are typically very low, less than 3% a year and as such have little impact. It can be argued also that it is as good as any other method and you have got to use something.

Grouping maybe better served possibly for higher volume corridors with more regular count programmes and more stable flows along the road to monitor and adjust for growth. But then, the estimate adjustment may be better served by looking along the road itself than to a wider group analysis.

Others would argue that this grouping approach is not sensible due to the variability within the group and suggest that no growth should be allowed for unless count data can be used as the basis to justify sufficiently a change to the estimate for that link.

Another example would be “fixed estimate” sections such as no-exit roads which it is argued have no growth in the absence of any intensification. These routes could be placed in a single group to represent that behaviour.

Parent-child relationships are another approach, albeit from a different direction for grouping. This was the basis of the RAMM Traffic Count Estimation Tool.

Groupings can be linked to continuous sites if these are present on or can be linked to the network. The RAMM set up proposes to have this grouping undertaken separately if the user wishes to adopt seasonal correction factors and adjustments to the estimate dependent on when the count was undertaken and possible. In either event, grouping can account for specific seasonal sections of the network affected by summer activities e.g. coastal communities or winter activities e.g. (ski resort townships).

Groupings can be split, depending on the user’s preferences as they see fit. Groups can be set up based on growth and no-growth aspects. Often growth aspects will be split by ONRC hierarchy and/or geographically to represent fast growing areas such as urban fringe, housing development or land use change.

Note though the strategy needs to populate enough counts in each group to be able to form enough guidance to undertake the analysis.

Using groups in updating estimates for traffic mix

Often underestimated is the importance of traffic mix in traffic monitoring information. There is often a focus on volume to within say 10% or 5% but we then have a default traffic mix which can under or over-estimate traffic loading of heavy vehicles by 100 or 200%. So while, grouping is some circumstances may provide little value in terms of volume growth, they can provide insight to traffic mix and improving this in the estimation process.

Groups allow us to form sections with similar characteristics and, as much as for growth, look at a typical traffic mix for those road sections where there is little other information. So traffic groups can provide value and a greater granularity in traffic mix updates.

Selecting Groups

In determining the link groupings, it is important to understand the methodology by which you want to update the estimates. That means there is an iterative process and the granularity of the groupings will align with the process. As discussed above there are several different approaches.

It is also essentially a manual process although parts can be automated and requires a sanity check at the end. However, it does not take long once the base information is assembled.

The different splits that could be employed are detailed in the table below along with a discussion. They can be combined as need for the characteristics and complexity of your network as well as the methodology you wish to adopt. The groupings sections within the links should also reflect these splits.

It is recommended at this early stage that traffic groups however are maintained. While it would essentially be a manual process to generate these, there are options to automate it but outside any RAMM tool. Use of GIS to assign groups to district plan land use types and ONRC categories would be suggested methods that do save time. Grouping is a useful method of covering both traffic growth and vehicle mix, and therefore encouraged as part of a recommended practice. It is acknowledged that in terms of growth that it adds little to accuracy, particularly for lower volume routes. However it does also add value for addressing traffic mix and both aspects should be considered in setting up the groups for the network.

The complexity level adopted is entirely up to the user and should reflect their preferences and therefore methodology for updating the estimates.

Calculating Growth Rate for a Traffic Group

Essentially, one of the purposes of groups is to update old estimates by a growth percentage, where we have no other basis from which to calculate growth. Again there are several methods that can be adopted:

• Assume a set annual percentage growth rate set by the user: A good example would be 0% growth for fixed estimate sections such as no-exit roads as discussed previously.
  • A sub-example maybe a set sample of carriageway sections or control sites used by the district to monitor and set growth rates. While not necessarily using the set rate themselves, their growth rate may be used on other groups to define the growth rate.
• Calculate a growth rate based on those sections within the traffic group that have a valid and consistent count history. It is suggested at least three sites in the last 5 years be used as a guide, with a variation of a maximum of say 20% or 50 vpd of the average of the counts. It is recommended that the year of the count be used as the basis for setting the slope of the trend rather than the difference in days between counts.
  • Calculate the growth rate by averaging the growth rate of carriageway sections with enough valid data to calculate a trend. In any event, once the growth rates are calculated, remove any outliers that would skew the group result unjustifiably.
  • Be wary of only using two counts particularly where they are close together as this may lead to implying a very steep rate of growth. Be careful of this in low volume groups as well where there maybe large swings in % change but this is more to do with variability in the count data from week to week. For example going from 20vpd to 30vpd the following year does not imply a 50% annual growth rate.