Methodology
PerfectLake v0.1 uses a 59-lake launch roster selected from a frozen 75-lake source superset. A lake appears only after public, attributable data is layered with a hand-written narrative and reviewed; selection alone does not publish it. Every attribute we show carries its source — an attribute without provenance is a bug, not a feature.
The launch target is 59 fully reviewed lakes. While curation is underway, we would rather publish fewer fully reviewed lakes than approximate about thousands.
Data sources
- Bird observation data from eBird (Cornell Lab of Ornithology)
- © OpenStreetMap contributors
- NOAA National Centers for Environmental Information
- Hydrography from the USGS 3D Hydrography Program (3DHP)
- Water-clarity data from the U.S. EPA National Lakes Assessment 2017
- Water-clarity tier derived from Sentinel-2 L2A imagery (blue/red + NDCI over water pixels, 2021–2024 growing season) via the Microsoft Planetary Computer; Copernicus Sentinel data
- Dark-sky tier from NASA Black Marble VNP46A4 (VIIRS/NPP nighttime lights), NASA LAADS DAAC
- Median home value from the U.S. Census Bureau American Community Survey (table B25077)
- Property-tax proxy from U.S. Census Bureau American Community Survey tables B25103 and B25077 (2024 5-year estimates)
- Climate descriptor derived from NOAA NCEI 1991–2020 U.S. Climate Normals
- Place metadata from Wikidata