About this site

Sources

• NASA Goddard Space Flight Center eclipse pages by Fred Espenak: Besselian elements, path tables, greatest eclipse and duration, and the decade tables. As NASA asks: "Eclipse Predictions by Fred Espenak, NASA's GSFC".
• NASA science (science.nasa.gov) for the 2027 geography and for safety guidance.
• American Astronomical Society (eclipse.aas.org) for safety guidance.
• Spain's Instituto Geográfico Nacional (eclipses.ign.es) for municipal times in Spain, and its CartoCiudad geocoder for town points.
• US National Geospatial-Intelligence Agency GEOnet Names Server for city coordinates outside Spain and the US.
• US Census Bureau: 2024 place gazetteer for US city points, 2024 cartographic boundaries for states.
• Map outlines: Natural Earth (public domain) and the Census boundaries, for drawing only.

Every value in the dataset was re-fetched on October 12, 2026 and matched against its source before the site was built. The whole dataset is on our data page.

How the checker works

The checker uses NASA's polynomial Besselian elements for each eclipse and the standard method from the Explanatory Supplement to the Astronomical Ephemeris (1961), also set out in Jean Meeus's Elements of Solar Eclipses. Coordinates are decimal degrees, north and east positive (south and west use a minus sign). If the solver doesn't converge, or an event falls outside the time range NASA's elements cover, the checker shows an error instead of a result. It finds the moment of greatest eclipse at your location, then the start and end of totality, on the IAU 1976 Earth ellipsoid at sea level, and converts to Universal Time with NASA's own ΔT. It treats the Moon as a smooth sphere, as NASA's tables do, so it doesn't include the few seconds that lunar mountains and valleys add or remove at the edges.

How we checked it

• Against NASA's path tables. At every center-line point NASA prints (102 for 2027, 28 for 2044, 100 for 2045), our totality duration matches NASA's within 0.19 seconds. At every edge point NASA prints, our edge of the shadow falls within 0.12 km of it.
• Against NASA's own calculator. On a grid of 2616 points in totality across southern Spain and northern Morocco, our eclipse type, totality duration and time of maximum match the script behind NASA's interactive eclipse maps to within 0 seconds. That's the only region and the only eclipse we compared this way.
• Robustness. For 20,000 pseudo-random locations worldwide per eclipse, every result was either a finite classification or an explicit error (0 errors, 0 non-finite). That checks the code doesn't break, not that it's accurate there.
• Against Spain's IGN. For all 105 municipalities IGN lists in totality, 74 agree with our duration within 5 seconds (median difference 3.5 s). The largest difference, 26.7 s, is at a town near the edge of the path. IGN includes terrain height, uses its own solar radius and its own ΔT, and we don't know the exact point it uses for each town. Times of maximum differ by 4 to 7.3 s, consistent with a different ΔT.

Display rules. We show durations to the second only where a place gets at least 45% of the nearby center-line duration; closer to the edge we round to 10 seconds. That threshold comes from the 2027 IGN comparison (every town above it agreed within 5 s) and is a display heuristic, not an error bound for other places or eclipses. "Too close to call" means within about 3 km of the edge measured as projected clearance in NASA's eclipse plane, an approximation rather than a distance over the ground. When the Sun is within 1° of the horizon at maximum, or below it at second or third contact, durations are labeled geometric. For Spanish towns we show IGN's own numbers. The full test output is in validation.json; its "scope" section says exactly what each test covers.

States in the 2045 path

NASA's decade table names only "s US", so we joined NASA's path-table edge points with straight lines and tested the band against Census state boundaries every 0.05 degrees. States touched by only a sliver are marked borderline.

What we leave out

• Weather and cloud predictions. We don't make them.
• Travel booking and tour links.
• Our own safety advice. The safety page is quotes only.
• Eclipse times for every town. The checker accepts any coordinates, but the tables cover a curated list.

Corrections

Found a mistake? Contact us through North Coast AI with a link to the official source.

Ads and analytics

This site may show Google AdSense ads and uses Google Analytics to count visits. See our privacy policy. Coordinates you type into the checker stay in your browser.

Who makes this

Path of Totality is published by North Coast AI LLC in Ohio, built with AI assistance and reviewed by Bill Burkey. Data checked October 12, 2026.