How this was made

The atlas starts from one point: the family's coordinates, 14.8633939 N, 74.2260639 E, on Kinnar Road beside Siddar in Karwar taluk. Every dataset was asked the same question. What is the oldest public record that reaches this point, and what does it say?

The rule for every number

• Every figure comes from a named public dataset, downloaded by a script in the pipeline/ folder of the repository.
• Scripts cache what they download, so the page can be rebuilt from the same inputs.
• Where a record has a known problem, the page says so beside the number, not in a footnote.
• Quotes from the 1883 and 1908 gazetteers were checked against the scanned page images, not just the machine-read text.

Sources, oldest first

• 1781 to 1960, maps. Bibliothèque nationale de France, John McCluer's 1791 chart, the Basel Mission map of 1857, Survey of India sheet 48 J (1915), the Imperial Gazetteer atlas (1909) and U.S. Army Map Service sheet ND 43-6 (1960). All public domain, via Wikimedia Commons and the University of Texas Libraries.
• 1796, air temperature. Berkeley Earth gridded land temperature, the 1° square over the village, cross-checked against NOAA's 20th Century Reanalysis V3 (from 1806).
• 1854, sea temperature. NOAA ERSST v5 and the Met Office HadISST 1.1 (from 1870), monthly grids. NOAA OISST v2.1 daily, from 1981, for marine heatwaves.
• 1854, cyclones. NOAA IBTrACS v04r01, North Indian basin, every storm since 1842 that passed within 300 km.
• 1878, sea level. Permanent Service for Mean Sea Level, tide gauges at Mumbai, Mormugao, Karwar and Mangalore.
• 1883, text. Gazetteer of the Bombay Presidency, Volume XV, Kanara, Parts I and II, and the Imperial Gazetteer of India, Volume XV (1908), from archive.org.
• 1891, rain. GPCC Full Data v2020 from Germany's weather service (DWD), 0.25° grid, 1891 to 2019. NOAA GHCN-Daily rain gauges at Karwar (IN009120100, from 1901) and Honavar (from 1939).
• 1972, satellite images. Landsat 1 to 9 from the U.S. Geological Survey and Sentinel-2 from the European Space Agency, via Microsoft Planetary Computer.
• 1975, people and buildings. European Commission Global Human Settlement Layer, R2023A. Census of India 2011, District Census Handbook, Uttara Kannada.
• 1984 to 2021, surface water. JRC Global Surface Water v1.4.
• 2000, forest. Hansen Global Forest Change v1.13, University of Maryland.
• Terrain. Copernicus DEM GLO-30. River and coast lines from OpenStreetMap contributors.

What the data cannot say

• The village name. The family has called it Ambejug and Umlijug. Neither appears in the 2011 census or the gazetteers. A 1960 map labels "Amble Jug" about 3.5 km to the west.
• Anything before about 1850 in numbers. Older records are drawings and words.
• Exact storm strength before satellites, roughly 1961. Many early storms have no wind speed at all.
• A sea level trend at Karwar itself, because its gauge changed benchmark in 2009.

Languages

The page reads in English, Kannada, Konkani (in Devanagari) and Marathi. The translations were drafted by machine and have not yet been checked by native speakers, so place names and a few technical words may be off. Quotes from the 1883 and 1908 gazetteers stay in their original English. Corrections go in site/i18n/, one file per language.

Rebuilding it

pip install --user numpy pandas requests pillow geopandas rasterio pystac-client planetary-computer
python pipeline/fetch_ocean.py
python pipeline/fetch_climate.py
python pipeline/fetch_sealevel.py
python pipeline/fetch_cyclones.py
python pipeline/fetch_landsat.py
python pipeline/fetch_land.py
python pipeline/fetch_history.py
python pipeline/fetch_basemap.py
python pipeline/write_place.py
cd site && python -m http.server 4831

The page itself is plain HTML, CSS and JavaScript with no build step. Charts are drawn as SVG by charts.js. Fonts: Tiro Kannada, from Google Fonts. Built with Claude Code.