# Traffic

Mean Annual Average Daily Trips

## Description​

High-resolution raster dataset of Annual Daily Traffic across road surface areas in the Puget Sound trough duirng 2016 and 2018.

## Overview​

We produced a continuous, high spatial resolution Average Daily Traffic (ADT) raster dataset covering the Puget Sound trough. The purpose of this dataset is for modeling water quality used in the Puget Sound Stormwater Heatmap. Here, we summarize the geoprocessing steps and the input data sources which included 2018 ADT point locations (Kalibrate Technologies, 2019) and year 2016 road lines with traffic and lane attributes (Messager et al., 2021), which was produced from 2016 US Census TIGER Lines spatial data joined with tabular data from the 2017 U.S. Department of Transportation National Transportation Atlas Database Highway Performance Monitoring System. We used Esri’s ArcGIS software and its Spatial Analyst extension for geoprocessing to develop this layer.

## Methods​

First, we grouped the road lines datasets based on their Federal Highway Administration highway functional class (FHWA, 2013) for our geoprocessing (Table 1). This was necessary to keep the extremely high traffic of Major road segments from being unintentionally allocated to Minor and Local roads where they intersect in 2-dimensional space (e.g. overpasses and tunnels). In the end, the outputs of the separate geoprocessing steps were combined, and traffic at intersecting cell locations was assigned the highest of the overlapping annual daily traffic values.

Table 1. Highway functional classes (FHWA, 2013) and the groups we assigned them for geoprocessing.

Functional Class Code (US DOT, 2013)Functional Class Name (US DOT)Geoprocessing Group
1InterstateMajor
2Other Freeway or ExpresswayMajor
3Principal ArterialMinor
4Minor ArterialMinor
5Major CollectorMinor
6Minor CollectorMinor
7LocalLocal
0UnclassifiedLocal

We created the Major roads raster dataset through a relatively simple process. Major roads included interstates and freeways, as well as their access ramps. We used the Functional Type attribute to differentiate access ramps (Types 0, 1, and 4) from the actual freeways (Types 2 and 6) during geoprocessing. We did this for similar reasons that we separated Major roads from Minor and Local. For each of the Major road groups, we used the Euclidean Distance tool to create 2-m resolution rasters covering the road area and based on the lane width described in the road lines attribute table. Next, we used Path Distance Allocation to assign known Average Daily Traffic (also from the lines attribute table) to their nearest road cell along the path of the road surface. We combined the two outputs by taking the maximum value at any given cell location.

Minor and local roads required a more complicated approach because the road lines dataset did not contain ADT, lane count,or lane width for most Local roads and because those Local roads often exist in a grid network rather than simple point- to-point layout. We assumed those unattributed roads consisted of two lanes (one per direction) and used Federal Highway Administration guidanceto assign lane widths of about 10-feet, or 3-meters(FHWA, 2013). Similar to the Major roads, we separated these roads into their Minor and Local classes to created two new 2-m resolution rasters of road area using the Euclidean Distance tool. The difference here is in the next steps for allocating Average Daily Traffic to each cell. We created our final road surface area rasters by classifying all values of the distance values to zero.

For minor roads, we assigned average daily traffic based on a point dataset of combined 2016 and 2018 traffic counts data to provide the most comprehensive coverage we could assemble for the Puget Sound region. The 2016 points were created from the 2016 road lines dataset at the mid-point of each road segment. The 2018 points data represent locations of observed Average Daily Traffic from numerous years that were modeled to estimate 2018 traffic based on demographic changes between the year of observation and 2018. Lastly, we performed a Path Distance Allocation to assign each Minor road cell the ADT value of the nearest ADT data point.

Our most complex step came was assigning average daily traffic values to each cell of
the Local roads. We approached this with three assumptions:

1. Average daily traffic drops drastically once it transitions from Minor to Local roads;

2. As distance increases from a Minor road to locations along a Local road, ADT continues to decrease but at a lower rate, nearly leveling off;

3. Minimum average daily traffi con any road is ten trips.

With those assumptions, we calculated traffic dispersing through a local road grid using a distance decay function calibrated with estimated traffic values on familiar urban residential streets of north Tacoma, WA. With the Raster Calculator tool, we used two newly created raster datasets as inputs in the distance decay function:

1. An initial (maximum) ADT raster dataset was used in the function as a constant representing the initial (and maximum) value for the decay. We created this dataset using the Path Distance Allocation tool which assigns the ADT of the nearest Minor road to each given Local road cell. We represent this here as variable a.

2. A Distance raster dataset contains the distance at every Local road cell to its nearest Minor road cell,which was calculated with the Path Distance tool. We represent this here as variable d.The decay rate constant ris calculated as a function of the initial (maximum) ADT value a, written as: r = f(a) = a* (11.03955 − 0.007743235a + 0.000001332168a<sup> 2 < /sup> )

This distance decay function outputs average daily traffic at each cell location, written as:

f(d) = a/(1 +(t d) 0.33)*

The exponent 0.33 keeps the resultant value from dropping below a reasonable amount of traffic moving through the Local road grid even at long distance from a Minor road source cell.

Next, we converted the output ADT datasets to integers and reclassified any ADT value of less than ten cars per day to ten cars per day as a minimum. Finally, we combined the resulting ADT datasets for all roads into a single dataset by choosing the maximum ADT value of those datasets at a given cell location.

## Layer Access in Earth Engine​

The javascript commands below can be used to access this layer within the Google Earth Engine Code Editor. A Google Earth Engine account is required.

// Import the layer data dictionaryvar data = require('users/stormwaterheatmap/apps:data/public')// To view data dictionary, print to the console:print('Data:', data)//Get this layer from the layer data dictionary: var layer_name = data.rasters["Traffic"]

#### Viewing​

Individual objects contain all the info used in the stormwater heatmap. To add it to the map, add the layer object.

var display_image = layer_name.layerMap.addLayer(display_image)

#### Analysis​

To get the raw image data for analysis, access the eeObject key.

var raw_image = layer_name.layer.eeObjectMap.addLayer(raw_image,{},'Traffic')

## Visualization​

### Palette​

Colors
1A3399
3B7CB8
5EBAD1
ABE5D4
DEEAB4
E0DD86
CBB64D
BF9D39
B99333
AF7E28
AB7424
A5691F
9B5516
964B12
91400
8A3308
842705
7F1900

Minimum: 0 Average Annual Daily Trips (log)

Maximum: 100000 Average Annual Daily Trips (log)

## Source​

The Nature Conservancy