Context for Estimate updates

Traffic estimation is important, feeding into several different aspects of transportation planning and asset and network management. These guidelines are to help the asset manager, seeking a process to follow to assign estimates to their network and some of the concepts involved. It is not a comprehensive guideline to assist the experienced practitioner nor cover more advanced and complex techniques that would be justified on a large, complex road network.

There are three key factors to be satisfied in our traffic estimation process, underpinned by a sound traffic count strategy. It is important to understand satisfactory confidence in our estimates and having this at the time they are made. This is important when we are considering inclusion of factors like seasonal correction or how growth is considered. Three key confidence areas are:

1. Achieving satisfactory confidence in total trips across my network (VKT)
   1. The estimates for the busier links particularly have sufficient confidence that our VKT is representative of our network. The lower hierarchy links have less influence and therefore require less confidence.
2. Achieving satisfactory confidence in my AADT estimates for each link
   1. The (predominantly) rotational sites particularly spread the link counts across the network to give appropriate confidence in the estimate for links in each hierarchy. As the hierarchy typically decreases, links are less critical and are counted with less frequency.
   2. There will be routes and links in lower hierarchy categories that we would like more confidence in understanding and estimating the AADT and hence more count sites. Some examples are bus, quarry and forestry routes and areas of land use change or high growth.
3. Achieving satisfactory confidence in the type of vehicles on my road network (traffic mix)
   1. Both volume and traffic mix are important. Loading on pavement and tracking freight movements are important components in traffic monitoring. The estimation process should satisfactorily accommodate both volume and mix to both network and link level.

A further factor is journey knowledge or how are the journeys being undertaken across our network. This issue is becoming increasingly important when we start considering place functions of different links and sections. This issue is not discussed as part of this report as it reflects more on how we capture traffic data.

Furthermore, there are three types of links we must be considering in how we update our estimates. Links are carriageway sections with similar traffic characteristics of volume and mix.

1. Links directly counted recently
   • Counts undertaken recently on a link are obviously the best source of information for updating an estimate for that link. Consideration is therefore needed as to an appropriate process to turn that moment in time record to a representative annual traffic estimate.
   • The balance is around adequate confidence in the number or age of “recent” counts such that the link still falls into this category or it needs to use alternative methods detailed in the other two link types.
2. Links with count data adjacent enough to be influenced by recent counts
   • Once we have estimates for sites we have counted recently, we can then assess links directly affected. This can vary from links on a road next to a counted link or between counted links. Great South Road in Auckland for example will have many counted links and estimates for the remaining links can be determined from interpolation.
3. Links with no recent or adjacent count
   • Effectively estimates are initialised and updated based around similar link sections or assumptions. Some, on long term rotational programmes, will be reset when a count is undertaken. But other links may never be counted. So, an appropriate methodology reflecting the required confidence in the estimate is required.

Desirable Qualities in an Estimation Process

Attributes to look for in a process:

• Provides satisfactory confidence in the accuracy of the network estimates
• Easy to set up
• Simple to maintain
• Simple to undertake
• Manual tasks minimised
• Covers volume and mix
• Aligns with the traffic count strategy best practice approach (core and rotational sites)
• Scalable i.e. can be used as a base from which more customised processes can be introduced
• Can apply to any data system.

Estimation Process

The overall process is described below, starting with a current database with roads divided into carriageway sections. Some road sections will have a series of count data including a latest count with an estimated Annual Average Daily Traffic (AADT) tied to a certain year. This “year” for the estimate may vary throughout the database. Some may just have an estimate but no count history, the estimate could be recent or dated. Some sections may have no estimate being new road or carriageway sections to the database. Others may have an estimate, but it is not current due to network changes that have meant a step change in volume (and/or mix) has occurred.

At the start of this process cycle, the estimates maybe referenced to different base dates through the network. So, the generation of links is best left until we have updated all the estimates to the same base year, essentially at the end of the process cycle.