lto — Long-Term Observatories of the United States
An interactive map and database cataloguing U.S. long-term environmental observatories
across seven interlocking spheres:
- Atmosphere — NOAA GML / Mauna Loa, NADP, AmeriFlux, ARM, CASTNET, IMPROVE, SURFRAD …
- Cryosphere — USGS Benchmark Glaciers, SNOTEL, CRREL, Toolik, McMurdo Dry Valleys, Juneau Icefield …
- Terrestrial / Ecological — NSF LTER, NEON (81 sites), USFS Experimental Forests & Ranges (77),
NPS Inventory & Monitoring, MAB Biosphere Reserves, NWRS, USFS RNAs, LTREB …
- Arid — Jornada Basin and Sevilleta LTERs, NEON desert domains,
Great Basin Experimental Range, Mojave Global Change Facility,
NPS Sonoran Desert Network …
- Agriculture — USDA-ARS Long-Term Agroecosystem Research (LTAR), ARS Rangelands,
USDA Climate Hubs, SCAN, KBS-AG …
- Aquatic — Ocean & Estuarine — IOOS Regional Associations, OOI, NERRS (29),
National Marine Sanctuaries, NEP, MarineGEO, NPS Coastal, NAML labs …
- Aquatic — Freshwater — USGS NWIS / WEBB / Hydrologic Benchmark Network,
GLEON, North Temperate Lakes LTER, Hubbard Brook, EPA NARS …
The site uses the long-term threshold from Peters et al. 2013 (≥10 years of record)
as the default inclusion gate.
New Mexico and ARID focus
The catalogue is national, but the site opens on New Mexico and the Southwest
drylands and carries a layer for the University of New Mexico's
ARID Institute — its partner centers, projects and people —
plus an index of every layer in EnviroData-NM, the
state's environmental data portal. The EnviroData-NM records are pointers: metadata
and links back to the portal's own download and OGC endpoints, with no data copied.
The Arizona Water Observatory's OGC API
collections are indexed the same way.
Switch the Scope filter to All U.S. for the whole catalogue. See
docs/arid.md.
Repository layout
This repo is a two-stack project (forked-and-adapted from
tyson-swetnam/cod-kmap):
- Python data pipeline (
scripts/, schema/, data/) — research subagents
(agents/R*-*.md) emit JSON facility records; scripts/ingest.py and the
load_*.py loaders write them into a local DuckLake
store of record (DuckDB 2.0, SQLite catalog at db/lto_catalog.sqlite, data
files in db/lake/, both gitignored); scripts/export_parquet.py writes
Parquet + GeoJSON to public/. The committed public/parquet/ is the canonical
copy of the data, and a fresh clone rebuilds the lake from it.
- Static MapLibre + DuckDB-Wasm site (
index.html, src/, public/) —
published to GitHub Pages with no build step.
Origins
Source repository for the engine: https://github.com/tyson-swetnam/cod-kmap
(MIT). Coastal/ocean facilities present in this database originate from cod-kmap;
terrestrial, atmospheric, cryospheric, agricultural, and freshwater coverage is
new in lto.
Reference framing for site selection and long-term thresholds:
- Peters, D.P.C. et al. 2013. Long-Term Trends in Ecological Systems: A
Basis for Understanding Responses to Global Change. USDA ARS Tech.
Bulletin 1931.
- Lugo, A.E. et al. 2006. Long-Term Research at the USDA Forest Service's
Experimental Forests and Ranges. BioScience 56(1): 39–48.
Quick start
Python 3.11 or newer (DuckDB 2.0).
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
python scripts/rebuild_lake_from_parquet.py # public/parquet/ → db/lto_catalog.sqlite + db/lake/
python scripts/ingest.py # data/raw/R*/*.json → upsert into the lake; reloads schema/vocab/*.csv
python scripts/qa.py # data-quality + schema-constraint gate
python scripts/export_parquet.py # lake → public/parquet/ + public/facilities.geojson
python scripts/export_view_caches.py
python -m http.server 5173 # then open http://localhost:5173/
To look at the site only, skip the Python steps: python -m http.server 5173
serves the committed public/ data as is.
License
MIT — see LICENSE.
