Wasted Wilderness: the Pacific Crest Trail, 2023

The closer to a road, the more waste

The study measured how far each survey started from the nearest road. Joined to the open data, the surveys within 100 m of a road found 23 times as much waste per km as the surveys more than 5 km from one. The study found the same slope with its own model: 13 pieces per km 10 m from a road, 0.4 per km 10 km away.

20.5

pieces per km within 100 m of a road (15 surveys).

7.3

per km 100 m to 1 km from a road (58 surveys).

2.4

per km 1 to 5 km from a road (129 surveys).

0.9

per km more than 5 km from a road (50 surveys).

Pieces per km, by distance from the survey start to the nearest road

Within 100 m

20.5308 in 15

100 m to 1 km

7.3423 in 58

1 to 5 km

2.4309 in 129

Over 5 km

0.945 in 50

Pieces in the open data, divided by the study's surveys in each band (each survey is 1 km; surveys that found nothing count too). The study's own table, with the two 100-piece surveys left at 100, gives 19.6, 7.2, 2.3 and 1.0.

The busiest kilometre was not a kilometre

At a sledding spot beside the Angeles Crest Highway, 36 m from the road, the surveyors logged 101 pieces in about 50 m and stopped at the survey's limit of 100. Their note: lots more trash nearby.

On a dirt road along the Los Angeles Aqueduct they reached the limit after 660 m. These are the only two surveys that hit it, both in Southern California, both on or beside a road.

In the replay, the camera tilts into the mountains at both places and walks them piece by piece.

Where the waste was, along the trail

Every survey that found waste, from Mexico (left) to Canada (right): taller bars, more pieces; warm colours, closer to a road. Grey ticks under the line are surveys where the study recorded nothing. The busiest stretches sit near Los Angeles and in southern Oregon; the High Sierra and Washington, far from roads, had the least.

The Sierra: skipped for snow in May, walked last

1234

CaliforniaCAOregonORWashingtonWA

Mexico, km 0Canada, km 4,270

1Km 32: Day two, near the Mexican border. 65 pieces in 1 km.
2Km 590: A sledding spot by the Angeles Crest Highway. 101 pieces in 50 m.
3Km 832: A dirt road along the Los Angeles Aqueduct. 100 pieces in 660 m.
4Km 2,809: Southern Oregon. 87 pieces in 1 km.

Within 100 m of a road100 m to 1 km1 to 5 kmOver 5 kmNothing recorded

Pieces per km surveyed, south to north

Southern California

8.5544 in 64

The Sierra

1.590 in 62

Northern California

3.6120 in 33

Oregon

6.4270 in 42

Washington

1.261 in 51

Each stretch's pieces divided by its surveys in the study's table (the Sierra is the section the snow closed, km 1,132 to 2,144).

What the pieces were

Soft plastic

25724%

Hard plastic, foam and rubber

18017%

Paper and glass

17916%

Metal

17816%

Everything else

14613%

Food

858%

Sanitary waste

606%

What the paper and the posts said, checked

The 2025 paper, the blog post and the news stories, against the open data. The paper's numbers come from the study's own survey table, which is public on GitHub; the open data is every piece as it was logged in the app.

Source	What it said	The data
Beyond Plastics, March 2023	A 1 km survey for every 10 miles of the 2,650-mile trail.	The study's table has a survey start every 10 miles from mile 10 to mile 2,650, with 13 left out for closures and danger: 252 rows. The paper counts 251.
The paper, 2025	12.5 pieces per km on average; about 53,000 on the whole trail (12,000 to 130,000).	The table averages 12.47 per km. One survey carries most of that: the sledding spot, where 100 pieces in about 50 m were scaled to 2,133 for the full km. Without it the other surveys average 4.0. The open data, counting pieces as logged: 1,085 over 251 surveys, 4.3 per km.
The paper, 2025	13 pieces per km near roads, 0.4 per km 10 km from one.	20.5 per km within 100 m of a road and 0.9 more than 5 km away, using the study's road distances.
The paper, 2025	Soft plastic 36%, metal 11%, sanitary waste 10%, hard plastic 8%, as the average share per survey.	The same way: soft plastic 37%, metal 11%, sanitary waste 9%, hard plastic 9%. Counting every piece together instead: 24%, 16%, 6% and 10%.
The paper, 2025	Waste in 65% of surveys; 21 surveys on snow found 94% less.	The table marks 88 of 252 surveys as finding nothing (35%), and 21 on snow averaging 0.88 pieces against 13.53 without.
The paper, 2025	Two surveys hit the 100-piece limit before 1 km.	101 pieces on April 28 and 100 pieces on May 7: the sledding spot and the aqueduct road.
San Francisco Chronicle, March 2024	More than 1,000 pieces of trash.	1,085 pieces in 913 records.
San Francisco Chronicle, March 2024	260 survey areas, about 2 or 3 a day.	252 in the study's table. The open data has 185 surveys that found waste, on 117 days: 1.6 a day, up to 4. Surveys that found nothing left no record in the app.
Advnture, March 2024	The San Gabriel Mountains outside Los Angeles were the dirtiest.	Yes: the busiest survey of all is the sledding spot on the Angeles Crest Highway in the San Gabriels, 101 pieces in about 50 m.
Advnture, March 2024	40% of the zones had no litter.	35% in the study's table; the paper says 35%.
Advnture, March 2024	Used toilet paper was the biggest offender, more of it further north.	Not in the counts: sanitary waste is 6% of pieces, behind soft plastic. Per km it was highest in Southern California (0.5) and Oregon (0.3), with 0.1 in Washington.
The Cool Down, August 2024	Around 200,000 pieces on the trail, about 75 per mile.	An early estimate. The paper settled on 53,000, or 12.5 per km (about 20 per mile); the open data's raw count is 4.3 per km.

Walk the trail

The story first. Then the whole trail lights up on a relief map of the West, 2,650 miles from Mexico to Canada, with the people behind the study. Then the walk, north up: a circle for each survey sized by what it found, the km walked beside the dot, across the snow skip to Northern California while snow falls over the Sierra they skipped, on to Canada, and back for the Sierra by plane. Five times the camera comes down and tilts into the mountains to walk one survey piece by piece while the live screen shows each photo, each survey within 212 m of a road: day two near the border, the sledding spot on the Angeles Crest Highway, the aqueduct road, southern Oregon and the Columbia River, where the map turns to aerial imagery. Photos of Macy and Tori show along the way, the last one their high-five at the Canadian border. The bars keep score by stretch, by distance to a road and by material.

Press play for March to September 2023 in 212 seconds. It ends with photos from the trail. Click any circle or piece for what it was, when and where; the labels under the bar jump to Mexico, the snow skip and Canada; and the map can switch from Night to Satellite, Topo or Light. The finds play as soft notes over quiet music; the speaker button turns the sound off.

Open the map full screen

The people behind it

Victoria McGruer, Macy Gustavus and Win Cowger in Rubbish shirts, logging a piece of trash in the app

Tori, Macy and Win logging a find in the Rubbish app. Photo: Moore Institute for Plastic Pollution Research.

Victoria McGruerResearcher and surveyor; first author of the study

Macy GustavusResearcher and surveyor

Win CowgerResearcher; led the study at the Moore Institute for Plastic Pollution Research

Emin IsrafilRubbish technical support; co-author of the study

Every piece was logged offline in the Rubbish app, with a photo, a type and a location, then exported for the study.

The study's other authors: Kendra Z. Hess, Hazel Vaquero, Krystle Moody Wood, Jorge Gonzalez-Estrella, Victoria M. Fulfer and Shelly Moore. Photos of the team: Moore Institute for Plastic Pollution Research.

About these numbers

The open-data file holds 913 records from the trail between March 28 and September 24, 2023: 1,085 pieces (a record can hold more than one). Records are grouped into the app's sessions, one per survey in most cases: 185 surveys found waste.
Per km: pieces divided by the number of surveys, using the study's own table of 252 survey starts (summary.csv on GitHub, © 2023 Win Cowger, MIT License), so surveys that found nothing count too. Each session is matched to the nearest survey start within 5 km. Distance to a road is the study's, from the start of each survey to the nearest road in OpenStreetMap.
The study's counts differ from the raw open data in two ways: surveys with nothing found were given a small value between 0 and 1 for its statistics, and the two surveys that hit the 100-piece limit were scaled up to a full km. The open data is the count as logged.
The trail line in the replay is the Pacific Crest Trail as mapped in OpenStreetMap (© OpenStreetMap contributors), scaled to its official 2,650 miles for the state lengths and to the study's 4,270 km for the km along the trail. The relief and the 3D ground in the dives come from the Mapzen terrain tiles on AWS Open Data. In the Columbia River dive the map crossfades to aerial imagery from the U.S. Geological Survey's The National Map (USGS Imagery Only; USDA NAIP orthoimagery at that scale; public domain; credit: USDA, USGS The National Map: Orthoimagery). On the live map the viewer can pick the mode: Night (the default and the video's), Satellite (the same USGS imagery everywhere), Topo (USGS Topo, The National Map's public-domain topographic map; credit: USGS The National Map: National Boundaries Dataset, 3DEP Elevation Program, Geographic Names Information System, National Hydrography Dataset, National Land Cover Database, National Structures Dataset, and National Transportation Dataset; USGS Global Ecosystems; U.S. Census Bureau TIGER/Line data; USFS Road data; Natural Earth Data; U.S. Department of State HIU; NOAA National Centers for Environmental Information) or Light (OpenFreeMap's own Positron colours). Both USGS maps stop at a scale of 1:9,028, so in the close-ups the imagery is softer and the topo map gives way to the light map. The map credits change with the mode. The tracks in the close-ups are the surveyors' own recorded routes, with short gaps inside one survey joined; between them the walked line follows the trail. In the close-ups every piece is its own dot: a photo that logged several pieces shows them as a small bunch around its point, at least 4 m apart, so each one can be seen.
The records are open data: RubbishOpenData, pacific-crest-trail-2023.csv (CC BY 4.0). No names, accounts or photos; locations to about 11 m.
The trash photos on this page and the 193 on the replay's live screen were checked one by one at full size: hands holding a find are shown; faces, legs, shoes, people's shadows and writing that reads as a name or a slogan are not. They are cropped square and carry no location data. Clicking a circle or a piece on the live map shows one of these screened photos when it has one, and never any other. The photos of Tori, Macy and Win are from the Moore Institute for Plastic Pollution Research; Emin's is from Rubbish. The four large photos of Tori and Macy in the replay (Tori with balloons, the two of them in the snow, Macy with a sample, and their high-five at the Canadian border) are from the Moore Institute for Plastic Pollution Research's No Trace Trails page, cropped square with no shoes, no shadows and no one else in them, and carry no location data.

On the Pacific Crest Trail, the trash was by the roads: 20.5 pieces per km within 100 m of one, 0.9 more than 5 km away.

The closer to a road, the more waste

Where the waste was, along the trail

What the paper and the posts said, checked

Walk the trail

The journey

The people behind it

From the trail

Read more

About these numbers