What is AADT, why does it matter, and how does big data make it more powerful?

As agencies grapple with budget and staff limitations, big data analytics enable officials to overcome gaps in Annual Average Daily Traffic (AADT) data for local streets and low-volume roads. But first, what is AADT, and how does it inform transportation decisions?

In recent years, infrastructure improvement has become a hot topic in the US.

At the center of that interest are our streets, roads, and highways. Road conditions are a key factor in an area’s quality of life, economic dynamism, as well as access to schools, jobs, and healthcare.

Jurisdictions across the US are gearing up to improve their road networks, in part thanks to the federal Bipartisan Infrastructure Law (BIL), also referred to as the Infrastructure Investment and Jobs Act (IIJA), which makes $110 billion available for these improvements.

Behind the scenes, there’s one transport metric that is fundamental to nearly every federal funding request or routine budgeting at the state or local level: Average Annual Daily Traffic (AADT).

So what is AADT and why is it so central to transportation planning and funding? We unpack what AADT is, how it differs from other fundamental roadway metrics, and how measurement is going digital to fill data gaps and add richness to planning and modeling.

What is AADT?

Annual Average Daily Traffic (AADT) is the transportation industry’s most essential metric for analyzing and forecasting traffic volume. Among other things, it’s used for the planning and design of infrastructure, tracking traffic congestion, estimating road safety, and as an empirical measure to help allocate highway funds.

In its simplest form, AADT takes in all vehicle trips on a segment of road or highway during a yearlong interval, in both directions, and then divides the total by 365 days to arrive at the average number of daily trips.

In addition to infrastructure and transport planning, AADT data is applied in many other contexts as a way to measure traffic flow and allow for “apples-to-apples” comparisons of traffic volume. For example:

• Retailers and commercial real-estate professionals consider AADT data when evaluating sites for development and for marketing and pricing properties
• Lawyers and law enforcement officials use the metric in accident cases and investigations
• Lawmakers use AADT to inform budgeting and prioritization of transport budgets and projects

But AADT is also a simple measure, which flattens away seasonal and weekday variations in traffic patterns. Nonetheless, AADT remains the most widely-referenced benchmark for how busy a road or street is.

AADT vs. ADT

There are two main types of traffic measurement: Annual Average Daily Traffic (AADT) and Average Daily Traffic (ADT).

These measures are starkly different in how they are calculated and applied. They are often confused in casual discussions so it’s important to understand the differences.

• As described above, AADT is the total volume of vehicle travel on a road for an entire year, divided by 365.
• ADT is the average number of vehicles traveling through a location during a period shorter than a year. For example, it may be applied to a season, or a selected month or week, a short span of days, or a specific day in the year.

AADT is generally used to measure long-term trends or changes in travel demand, while ADT is more useful for short-term planning and operations.

For example, if a city wants to know how many people use a particular bridge to estimate wear-and-tear and traffic loads, it would use AADT. But if that same city needs to know how many cars will cross the bridge during a summer weekday to plan construction, it would use ADT calculated from a sample taken in the target interval.

Both measures have their advantages and disadvantages.

AADT can obscure seasonal variation (e.g., fewer people travel in winter), special events (e.g., increased travel for holidays or road closures within the measurement period), and day-to-day variation in demand (e.g., less travel on weekends). This can sometimes make it difficult to compare year-over-year changes or identify important micro-trends.

ADT can pick up these kinds of fluctuations, but they will not reflect overall demand on a road since by definition they only consider shorter periods. On the other hand, they have a limited lens: they will not reflect in any variations occurring outside of the measurement period.

As discussed below, AADT data is relatively difficult and expensive to collect since it requires real-world data spanning an entire year. Even when using calculation methods that don’t require 365 days of data, the need for accurate and continuous data is still relatively burdensome.

ADT is more forgiving from a measurement standpoint since by definition it requires the counting of traffic during shorter intervals.

How to calculate AADT

To calculate Annual Average Daily Traffic (AADT) in its simplest form as described above, analysts must know total traffic volume on the target road segment for every day in a given year.

This baseline method isn’t simple or cheap to implement since by definition it requires permanent devices known as counters or ATRs (automatic traffic recorders) detecting passing vehicles and collecting complete trips data for 365 continuous days in a year. Even missing a few hours of traffic means the calculation can be thrown off. Not to mention the recorders are not easy or inexpensive to install and maintain. During the COVID-19 lockdown, agencies saw their ability to collect counter data hamstrung by lockdown orders that kept workers off the road.

Even when permanent counters are in place and putting out continuous AADT traffic data, sometimes special events such as roadwork or adverse weather can distort averages with outlier days of abnormally low or high trip counts.

Finally, and most commonly, there are simply too many road segments in any jurisdiction to allow for comprehensive AADT data from permanent counters.

In the video below, Keith Nichols explains why Hampton Roads TPO commonly encounters data gaps with traditional AADT data collection methods, and how the agency supplements traditional methods for long-range transportation planning.

Traditional methods of closing AADT data gaps

Statistical methods, complemented with temporary data collection, are commonly deployed to address common data-collection issues. When officials turn to these methods, they do so knowing they are sacrificing accuracy for savings.

It’s worth understanding how these methods work together, and some of the tradeoffs.

Short-term expansion

One industry-standard method for arriving at Annual Average Daily Traffic (AADT) in the absence of permanent counters is known as “short-term expansion.”

In this method, a road section’s traffic is calculated on the basis of a temporary counter collecting two days or more of data. That incomplete short-term data is then “expanded,” or scaled up, to calculate AADT. To do this, analysts derive scaling “factors” from a nearby permanent traffic counter that has a year’s worth of data.

Ideally, that permanent counter used as a reference has same-year data. But it’s not uncommon for transport analysts to be forced to rely on permanent counter data from past years when calculating AADT using this method.

Obviously, even when the short-term expansion model bakes in some math, this tactic relies on the differences between the target road’s short-term traffic data versus the same-day measurements collected by the nearby permanent counter on a different road.

If these differences are not consistent across the year, i.e. if the short-term data was not taken on representative days, AADT accuracy will be compromised.

In the instances where permanent counters are in place but there are small gaps in the annual data or road closures and other outlier events that may throw off the averages, a separate established method is to limit the number of days for which complete datasets are required.

The AASHTO method

In one industry-recognized approach, officials collect total traffic volume on seven separate days in each month that correspond to the different days of the week. This method leaves planners with 84 days of data to work with that nonetheless will account for variations in traffic across different weekdays and on weekends.

Then, they take an average for each day of the week sampled across the year, giving them seven averages, and then they take the average of those averages for an AADT.

The American Association of State Highway Transportation Officials (AASHTO) has promoted this technique, known as an “average of averages” method.

Other methods

Many other estimation methods can be used depending on what data and counters are available and affordable. In fact, the proliferation of methods adds to the complexity faced by transport planners as they consider approaches for data collection and AADT calculation.

In fact, one in-depth review of the relevant academic literature on AADT estimation identified 30 separate methods just for estimating AADT on low-volume roads.

Adding to the difficulty, these techniques were hardly one-size-fits-all.

“Some AADT estimation techniques are only applicable in specific locations,” write the authors, Edmund Baffoe-Twum and Eric Asa of the West Virginia University Institute of Technology, along with Bright Awuku from North Dakota State University. “Others require significant data to provide accurate estimates. Several processes to adjust models for a location may be needed for other locations.”

Traditional methods vs. traffic analytics

Arguably the biggest change for Annual Average Daily Traffic (AADT) in recent years is the availability of instant up-to-date AADT estimates right from a computer through traffic-analytics providers.

On-demand traffic-analytics platforms reduce the need for expensive and sometimes hazardous fieldwork and are able to fill data gaps whenever they arise.

As a result, jurisdictions have more flexibility in the extent to which they rely on permanent and temporary counters for AADT metrics.

For example, recently StreetLight helped fill gaps in AADT data by the Indian Nations Council of Governments in Tulsa, Oklahoma. Due to budget issues, many jurisdictions in the area had stopped reporting traffic counts to INCOG, a metropolitan planning organization or MPO. In minutes, StreetLight was able to generate traffic counts for all of the untracked road segments.

In the video below, we explore other ways agencies commonly use on-demand traffic analytics platforms to leverage AADT metrics, such as developing crash rates and understanding the traffic impact of road or lane closures.

How do traffic-analytics platforms come up with their AADT measurements?

Typically, traffic analytics rely on connected devices and Internet of Things data, and then layer in parcel data and road-network data for a complete picture. With this data, it’s possible to create analytics that model vehicle trips on a stretch of road in the absence of temporary or permanent counters.

The real technical challenge for deriving AADT from traffic analytics is not just in collecting and organizing the large volume of location data, but in the next steps:

• Algorithms are needed to match this raw data to vehicle trips
• The dataset, for greater accuracy, is enhanced with additional sources of data such as US Census Data or street-map data to account for changes in demographics and road networks
• The resulting AADT model must be tweaked for greater accuracy by testing results against real-world “ground truth” data, which should encompass different road and vehicle types, e.g. heavy trucks

In the case of StreetLight, which provides AADT metrics for 4.5 million miles of roadway in the US and Canada, all these steps were important and detailed in the whitepaper, “AADT 2022: New Methodology, Validation, and Improvements.”

By comparing their own AADT metrics to AADT produced by thousands of permanent counters nationwide, StreetLight was able to determine that their AADT figures fall within the 95% prediction interval for all range of road types.

Challenges in measuring AADT: local and low-volume roads

AADT measurement is increasingly being shaped by many jurisdictions’ need for more granular and complete coverage of road systems.

What’s driving this demand? The answer is a set of interrelated traffic, demographic, and environmental concerns that are only growing stronger with time.

First, local streets are increasingly seeing overflow traffic from overburdened highways and multi-lane roads.

“Estimating AADT on local streets becomes a necessity as local street traffic continues to grow and the capacity of arterial roads becomes insufficient,” write Peng Chen, Songhua Hu, Qing Shen, Hangfei Lin, and Chie Xie. They are the authors of a 2019 paper on AADT measurements in the Seattle area, published in the Transportation Research Record.

Second, jurisdictions are also newly anxious to track traffic increases on low-volume roads in rural and semi-rural areas. Many of these areas have seen a major influx of short- and long-term visitors and new residents arrive over the last two years.

A 2022 analysis by StreetLight showed how this trend had impacted the resort town of Jackson Hole, Wyoming. The analysis showed that AADT on a section of unpaved road —an access point to Grand Teton National Park — had already significantly surpassed 2019 levels in 2021, even after dipping dramatically during the COVID pandemic.

AADT counts for Moose-Wilson Road from StreetLight suggest 2021 traffic volume had already exceeded 2019.

Thirdly, environmental and road-safety concerns are also behind the need for more comprehensive AADT data. Since a significant proportion, if not a majority, of vehicle miles in a state are driven on these roads, it’s impossible to form a complete picture of emissions or accident trends without it.

Despite the demand, coverage of low-volume and local streets poses a formidable challenge for traditional AADT calculation.

As we’ve seen, it is cost-prohibitive to deploy permanent counters widely on local streets or low-volume roads. Not to mention, rural roads are far-flung and cover many miles of sometimes difficult terrain. Local roads in urban areas are dense and highly varied in traffic patterns, which would mean putting counting stations on virtually every corner.

Many jurisdictions, in the cases where they have the budget to take the measurements at all, turn to temporary counts and estimation methods to measure AADT on these segments.

For these reasons, traffic analytics–based AADT metrics are a cost-effective and simple solution for filling data gaps on local and low-volume roads.

For example, StreetLight’s AADT metrics include urban and rural roads — even unpaved roads, as seen in the case study from Jackson Hole. Analyses covering hundreds of low-volume road segments can be run in minutes.

Annual Average Daily Traffic (AADT) and technology

Traffic-analytics platforms relying on big data approaches are only the latest tech innovations to transform how Annual Average Daily Traffic (AADT) is collected and calculated.

In the 1930s, AADT was based solely on manual counts, which required considerable manpower and intensive fieldwork, according to David Albright in a 1991 article on the history of AADT measurement. Counting devices only began to be used in the 1940s, and became widespread only after a couple of decades, with methods continuing to evolve in the years since.

More recently, computers and algorithms have helped run some of the sophisticated statistical models used by transport engineers and planners for AADT estimation. As discussed, StreetLight itself uses advanced software algorithms to process data, tie it to vehicle trips, and enhance it with U.S. Census and OpenStreetMap data.

While conceptually AADT has remained the same metric over all this time, and surprisingly resilient as a keystone metric in transport planning, technology has completely transformed data collection, calculation, and estimation methods. It’s a good bet that technology, and specifically big data and software-driven algorithms, will continue to drive innovation around AADT in coming years.