Annual Average Daily Traffic (AADT)

Overview

The annual average daily traffic (AADT) table contains information about almost all major roads. For motorways and trunks, Replica also provides single-truck volume and combination truck volume [1]. Data is available for 2019, 2021, 2022, 2023, 2024, and, 2025 and is updated annually. The data is available for download through the Replica application in CSV, Geojson, and Shapefile formats.

To produce AADT, Replica leverages data about hundreds of millions of trips per week, in combination with network link volume ground truth. Read the full methodology here.

See what coverage we have in specific areas here. The coverage values in the table below are calculated based on the entire OSM network, which is highly detailed containing even service alleys. The denominator is all links by highway type in the OSM network.

Sample Download

Click here to download a sample of Replica's AADT table.

Schema

Methodology

There are two inputs to the Replica AADT model:

1. A large panel of pre-processed vehicle counts derived from connected vehicle data. We also use trajectories from connected trucks with high-level vehicle classification - medium trucks and heavy trucks. Both panels contains information about hundreds of millions of trips per week, and constitutes more than 30% of all trips that happen.
2. A set of ground truth AADT values, derived from HPMS data, conflated to the same representation of the roadway network. These AADT values are broken out by vehicle classification—total count, single-unit truck, and combination truck. We use the most up-to-date HPMS data available: 2024. [2]

We use the ground truth data to infer the “penetration rate” of the probe panel. The penetration rate is what fraction of actual trips are represented in the panel. In the initial iteration of the Replica AADT product, the penetration rate was defined as a function of the properties of the link. In this iteration, for overall AADT, the penetration rate is allowed to vary as a function of the H3 cells containing the network link and the highway type.

We use ElasticNet regression to learn spatially-varying scaling factors that convert GPS probe vehicle counts into AADT estimates. Ground truth coverage is often sparse; some geographic areas and road type combinations may have limited or no sensor data for model training. ElasticNet addresses this challenge through regularization, which enables the model to learn smooth, spatially-coherent scaling factors by leveraging patterns across similar areas and road types. The technique combines two forms of regularization: Ridge regression ensures neighboring areas have similar scaling factors (reflecting realistic geographic variation in GPS penetration rates), while LASSO regression identifies areas with insufficient training data that should fall back to broader regional averages. The model also incorporates constraints to prevent unrealistic scaling factors and uses cross-validation to optimize performance. This approach ensures comprehensive, accurate AADT predictions across the entire road network, including roads far from ground truth sensors, while maintaining spatial coherence and preventing overfitting to noisy data.

Once this is done, inference using the model is straightforward: every link with probe vehicle data gets multiplied by a coefficient based on it's h3 cell and highway type. When aggregated at the daily level and averaged, this produces a by-vehicle classification model for each network link’s AADT, nationwide. By aggregating at the hourly level instead, we can also produce an estimate for the typical volumes on a per-day of week, per-hour of day basis.

Quality Metrics

Model Validation

Replica’s 2025 Annual Average Daily Traffic (AADT) model was developed using 2024 HPMS data and 2025 permanent counter data from 10 select states. To ensure an unbiased assessment, this validation compares Replica’s outputs against FHWA Traffic Monitoring Analysis System (TMAS) data from January through June 2025. Notably, all data from the 10 training states were excluded from this analysis. By validating against the remaining 40 states, we ensure a "out-of-sample" test, verifying the model's accuracy on independent data.

The table below summarizes the distribution of Replica’s AADT estimates within specific percentage error margins relative to permanent counter volumes. Additionally, the scatter plot illustrates the correlation (R-squared= 0.98) between FHWA TMAS AADT (January–June 2025) and Replica AADT (2025), demonstrating the model's overall fit and reliability.

Appendix

[1] Functional Road Classes describe how major a road is. FRC 1: Highways and major interstates. FRC 2: Major artery. FRC 3: Major road. FRC 4 and higher: Neighborhood Streets. Single-Unit Trucks are defined as FHWA Vehicle Classes 5-7. Combination Trucks are defined as FWHA Vehicle Class 8 or greater.

[2] States for which we only have 2019 HPMS ground truth data: Washington, Oregon, Idaho, Nevada, Utah, Arizona, New Mexico, South Dakota, Wisconsin, Mississippi, Tennessee, South Carolina, West Virginia, Virginia, Pennsylvania, Vermont, Rhode Island, Maine, Alaska. For all other states, there is HPMS ground truth data for both 2019 and 2021.

[3] A strongly regularized formulation would be more likely to give equal scaling factors to the two trips mentioned, whereas a more weakly regularized formulation would attempt to satisfy the ground truth constraint however possible. In such massively underdetermined systems like these, totally un-regularized least squares will match the training data exactly and is less likely to generalize.