Every dataset here is summarized on the same grid, so a single cell can be read across vegetation, citizen-science records, unique species, soils and land administrators. The grid was developed by the University of Kansas, with extensions by NatureServe: a one square mile hexagon, a cluster of seven hexagons (7 sq mi), and a cluster of seven clusters (49 sq mi). Cell IDs are the grid's own IDs, so the numbers here join back to the source tables.
| Layer | Source | Keep in mind |
| Land administrators | A.T. corridor lands and tracts (A.T. administrative areas) | Who administers mapped land in each cell, and their share. Land protected over time adds the acres of tracts conveyed by each year, from the tracts’ conveyance dates, loaded live; tracts without a date are left out. Generalized to the grid; not a boundary record. |
| Vegetation | NPS vegetation map for the A.T., by mapped vegetation type | The map stops at the edge of the surveyed lands, so most cells are only partly covered. |
| iNaturalist | Research-grade records, , for the A.T. corridor and the HUC10 watersheds. Kinds of life are by taxonomic group (plants, insects, fungi and others) | Records follow people. Few records means few visits, not few species. |
| eBird | Checklists, , for the A.T. corridor and the HUC10 watersheds | Same caution. Thinly sampled cells are faded on the bird maps. |
| EDDMapS | Invasive plant and forest pest reports from the EDDMapS points service (University of Georgia), rebuilt quarterly with iNaturalist and weekly for the spread maps | Only species on the invasive lookup and the forest pests are counted. A report that matches an iNaturalist record (same species, within a day and 50 m) is counted once, as iNaturalist. EDDMapS has no measure of search effort, so the shares of records use iNaturalist alone. |
| Facilities and landscapes | A.T. facilities map service and inventory (NPS and ATC) | Shelters, campsites, privies, parking, bridges, vistas, side trails and maintained landscapes in each cell; the first six are counted live from the map service. Trail structures such as waterbars and steps are estimated from maintainer-section totals. |
| Soils | NRCS soil survey (gSSURGO), all 14 trail states | Area-weighted soil ratings. The trail vulnerability index averages three NRCS trail ratings, from 0 (best) to 1 (worst). It rates the soil, not the trail. |
| Wetlands | USFWS National Wetlands Inventory, all 14 trail states | Wetland acres in each cell by type, and the part on A.T. lands. Small, forested and seasonal wetlands are often missed. “Possibly sensitive” flags likely bogs, fens, seeps and vernal pools for a field check. Not a delineation. |
| Restoration targets | Built in this page from the layers above | Two 0 to 100 scores for grasslands and forests. How they are scored is set out under Restoration targets below. |
| Flowering | iNaturalist research-grade records, built weekly, with newer records fetched live | Nearly fifty plants: trail-wide, A.T. Seasons, northern and southern indicators, and AMC focal species. First, peak and season length follow USA National Phenology Network measures, pooled across years. The flowering calendar by latitude follows the USA-NPN calendar view: the middle 80% and 50% of flowering-record dates and the median for each band of latitude, from 2010 on; bands with fewer than 20 records are not drawn. These are dates of photographs of the plant in flower, not measured flowering periods. Records follow hikers, so compare years by eye as leads for field checks; for measured change see A.T. Seasons (Nature’s Notebook). |
| Forest health risk | USDA Forest Service 2024 National Insect and Disease Risk Map (composite risk, 240 m) | The share of each cell's treed area, and its acres, where 25% or more of the tree basal area is expected to die from insects and diseases over 2024 to 2038 without treatment. The map combines 172 insect and disease hazard models. Counted from the 240 m cells whose centers fall in each grid cell. A strategic, broad-scale planning map, not a stand-level assessment. |
| Forest condition (NASA) | Canopy loss: OPERA Land Surface Disturbance Alert, Version 1 (NASA JPL, from Landsat and Sentinel-2, 30 m), on tree cover in ESA WorldCover 2021; updated monthly. Height and biomass: GEDI lidar, L3 and L4B gridded products, 1 km (from April 2019). Green-up and leaf-off: MODIS Land Cover Dynamics (MCD12Q2 v6.1, 500 m, yearly from 2001) | Canopy loss counts confirmed loss of vegetation (full or partial) by the month it was first seen; it follows the satellite record from 2022 or 2023, so it shows recent change, not history, and cloud or snow can delay detection. GEDI samples strips of ground, so cells with fewer than 50 footprints are blank; 1 km squares averaging over 50 m (164 ft) of canopy, which no eastern forest reaches, are left out as errors (they turn up over water); and heights average all land in the cell. Green-up and leaf-off are dates of the vegetation index rising and falling, averaged over the latest five years; the shift compares them with 2001–2005. Screens for a field visit, not stand surveys. |
| Climate change since 1980 | Daymet Version 4 R1, daily weather on a 1 km grid (ORNL DAAC), 1981–2000 against the latest 20 years; rebuilt each June | January lows, July highs, heavy-rain days (1 inch or more), the frost-free season (32°F) and snow water on 1 April, averaged over each period on Daymet's grid, then over the cell. Daymet is interpolated from weather stations and modeled for snow, so read it as broad change, not a station record; it differs a little from the PRISM and NOAA normals above. |
| Soundscape | Geospatial sound modeling, National Park Service Natural Sounds and Night Skies Division (DataStore reference 2217356); rebuilt when NPS publishes a new release | A model of sound across the country at 270 m, built from recordings at 492 sites: L50 dBA for a typical summer daytime hour, which is the level exceeded half the time rather than the loudest moment. Two layers are modelled, what a person would hear and what the place would sound like with human sources removed; the difference is the noise we add. Cells carry the average of the 270 m readings falling inside them. It is a national model, so it describes lengths of trail rather than a single spot. |
| Viewshed | USGS 3DEP elevation, delivered as AWS Terrain Tiles; rebuilt by hand when an overlook is added | What can be seen from an overlook on the A.T., and so where a new building, tower, quarry or clearing would be seen from the trail. Rays are cast from the overlook across bare terrain at 10 m, allowing for the curve of the earth and the bending of light, out to ten miles; each cell is scored by how much of it falls in view. The observer is placed on the highest ground within 250 m of the published coordinate, because an overlook coordinate often marks the foot of a cliff rather than the top. Trees and buildings are not in the terrain, so this is the leaf-off case with nothing in the way — the conservative one for judging what a change would be visible from. It is computed one overlook at a time, so only cells within ten miles of a mapped overlook carry a value. |
| Trail vulnerability | Trail slope from the A.T. centerline and USGS 3DEP elevation (1 m lidar where available, 10 m elsewhere), rebuilt yearly; with the soils, weather, storm events and visitation layers above | Grade, hillside steepness (landform grade) and slope alignment are measured every 10 m along the mapped centerline. Grade is the sustained grade over about 90 m (elevations smoothed over 50 m, rise taken over 40 m), so lidar noise and a line a few meters off the tread average out. Cells holding only a short piece of trail are blended with the trail just before and after it (up to 800 m each way along the line; a quarter mile of the cell's own trail counts equally with it), so a few steep steps don't push a cell to the top. The score adds points per factor (default weights: steepest trail grade 40, hillside steepness 10, how directly the trail climbs 10, heavy rain 15, use hot spots 10, soil erosion hazard 5, wet organic soils 5, forest pests 2.5, invasive plants 2.5; tested against the trail assessments, where within each trail club section they rank recorded treadway problems at a Spearman correlation of 0.30, against 0.22 for the first weights and 0.22 for one assessment predicting the next) and scales them to 100 over the factors with data. Shares and NRCS ratings are used as they are; rain, wet soils, use hot spots (parking areas, shelters, campsites and vistas) and forest pests and invasive plants (species reported to iNaturalist or EDDMapS) are ranked among cells of the same size. Forest pests and invasive plants carry a low weight as early indicators of declining forest health and its cascading effects; like all citizen-science records they reflect where people have looked. Use hot spots stand in for use because citizen-science records follow naturalists and birders rather than hikers; they are counted in the A.T. corridor cells only. The factors follow recreation ecology research on trail soil loss: trail grade and how directly the trail climbs are the two main design factors (Olive & Marion 2009; Marion & Wimpey 2017; Meadema et al. 2019 on the A.T.). Two other key factors in that research, tread drainage features (water bars, drainage dips, grade reversals) and rock in the tread, are not modeled because no corridor-wide data exist for them; a trail with good drainage and a rocky tread can hold up on a grade the score ranks high. Each 500-foot trail segment also carries a Trail Sustainability Rating (Good, Neutral, Poor, Very poor, from grade and alignment); on the A.T. the four classes recorded 0.65, 0.27, 1.53 and 2.25 times their club section’s erosion rate. Flat trail (Neutral) is prone to mud and widening rather than erosion. It screens for a closer look and is not a trail condition assessment. |
| Trail segments | The A.T. centerline cut into 21,769 segments of about 500 feet (160 m), measured every 10 m from USGS 3DEP elevation by the trail slope build | Each segment carries its steepest sustained grade, average grade, hillside steepness, how directly it runs up the slope and an erosion risk score (ranks among all segments: grade 60%, hillside steepness 20%, how directly it climbs 20%), and, from elevation rings 30 m and 100 m around it, whether it sits in a low spot. Low, flat trail (average grade under 5%, hillside under 15%, 0.25 m or more below the ground 30 m around) is marked with a dashed line; within club sections these stretches recorded about twice the drainage problems of other trail. A cell-wide wet-tread score was tried and dropped: at a square mile it matched recorded drainage problems too weakly to rank cells. The trail score's slope factors use the steepest 10% of each cell's trail. Segments are drawn under the trail question when zoomed in; the same segments are in gis/at_trail_slope_segments.gpkg for GIS. |
| Recorded trail problems | A.T. trail assessments: 16,373 problems recorded by trail clubs and partners in 82 club sections, with their kind, severity and the fix called for (provided by APPA) | Each record is placed in its grid cell. Treadway problems are counted by kind (erosion, drainage, other) and severity, per mile of trail, and compared with the club section average: recorded problems divided by what the cell's trail would hold at its trail club section's average rate, since sections recorded at rates from 0.4 to 28.7 problems a mile. Problems found 13 to 22 years ago; many have been fixed and some trail relocated since. |
| Weather and climate | PRISM 1991–2020 normals, 800 m (PRISM Group, Oregon State University); NOAA nClimGrid-Daily, 1991–2020; NOAA Atlas 14; NOAA Storm Events Database, 1996 to the last full year; gridMET wind (University of Idaho), 1991–2020. Rebuilt each January; storm events every quarter | Normals, frost-free season, heavy-rain days, the 25-year storm and wind are sampled on a regular grid of points in each cell and averaged. PRISM and Atlas 14 resolve the mountains well; nClimGrid-Daily and gridMET are about 4 km, so the highest ridges are smoothed. gridMET is a research dataset, not an official record, and understates ridge-top wind. Each storm event counts in every cell whose center is in the National Weather Service forecast zone or county it was reported for, so the counts mean the same at every cell size and show where events are common rather than exact spots; a cell's count is not a count of separate storms across the trail. Map colors use one set of classes for every cell size and both extents. |
| Citizen science observers | iNaturalist, eBird and EDDMapS records (the same downloads as above), counted as observer-days: one person active in a square mile on one day | An index of naturalist and birder activity, not a count of visitors; it follows the "user-day" method used to track park visitation from photo and social-media posts, and validated against counters and official counts in several studies (see Sources). An eBird checklist within 250 m of an iNaturalist observer on the same day and square mile counts as the same party; EDDMapS reports are report-days. Most of the rise in observer-days since 2010 comes from more people using iNaturalist, not more visitors, so compare places and seasons with more confidence than years. |
| Geology | USGS State Geologic Map Compilation (Horton, San Juan and Stoeser, 2017, doi.org/10.5066/F7WH2N65) and NPS Geologic Resources Inventory geologic map footprints (July 2024), rebuilt yearly | Bedrock type, carbonate and mafic/ultramafic rock, and age come from the seamless state-map compilation, mostly 1:100,000 to 1:500,000 scale, so small units and exact contacts are not shown. Map coverage measures come from approximate footprints of the published and in-progress geologic maps the GRI found around the trail; in the HUC10 watersheds further from the trail, some maps may be missing. |
| Plants of cultural significance | iNaturalist research-grade records in the quarterly summary, matched to cultural_plants.csv (136 plants) | Compiled from the Greenbelt Indigenous Botanical Survey (Plenty Canada, Dolan and Ni 2026), Mooney (1891), Mooney and Olbrechts (1932), Tantaquidgeon (1942), the Haudenosaunee Confederacy and Cherokee and Wabanaki sources; full citations under Sources. Not exhaustive: the Native American Ethnobotany database and the longer ethnographies are not yet included. Ginseng, goldenseal, ramps and lady’s slippers are shown only at the 49 sq mi scale. A starting point for consultation, not a statement of any nation’s knowledge. |
| Forest pests | iNaturalist research-grade records (all life stages) and EDDMapS reports, built weekly, with newer iNaturalist records fetched live | Spotted lanternfly, emerald ash borer, hemlock woolly adelgid, elongate hemlock scale, spongy moth and beech leaf disease; other pests (beech scale, chestnut blight, browntail moth, winter moth, Asian longhorned beetle, southern pine beetle, oak wilt, dogwood anthracnose, Sirex woodwasp) appear once the project has 100 or more records of them; any pest under 100 research-grade records is left off the list, which runs from most to fewest records. Shown from 2010 on: the cells with a record by each year, the share of each cell’s iNaturalist records by year, the range along the trail and the cells with records by year. Records follow observers; a blank cell means nothing reported, not absence. |
| Historic resources | National Register of Historic Places (NPS) and OpenStreetMap cemeteries, rebuilt monthly | Two extents: A.T. corridor (records inside an A.T. administrative area) and HUC10 watershed (every record). The National Register and cemetery map layers use the same file and show at every zoom. Recorded sites only: small family cemeteries are often missing and some National Register locations are generalized. Find a Grave links open the cemetery’s page when its ID is known, otherwise a search. |
| Unique species | Rare species priority zones | Rates how sensitive the populations of unique species are in each cell, from plant and animal records. Shown at 7 and 49 sq mi only. A screening aid, not a survey result. |
Where do we target restoration? gives each corridor cell two scores out of 100 for restoration on A.T. corridor lands, one for grasslands and early successional habitat and one for forests. A higher score means more of the things that make a cell worth restoring. Results lists the highest-scoring cells first. The scores are a shortlist for site visits, not a site assessment.
The cell panel shows the points behind each score. A rarely surveyed cell can score low on invasives or birds simply because no one has looked, so check How many observers are out there? alongside.
iNaturalist and eBird summaries are refreshed quarterly. This page shows the run.
Reese, C. 2026. ACT - Appalachian Conservation Tool: resource data for the Appalachian Trail corridor summarized on a nested hexagon grid. Interactive dashboard, version 2026.10. https://act.atconservation.com/. Accessed .