Thirdtrail

thirdtrail

Python client for the thirdtrail APIs: sun, moon, twilight, solar irradiance and tides for any point on Earth. Sun and moon times are corrected for the elevation you pass in.

pip install thirdtrail

from thirdtrail import Client

tt = Client(api_key="tt_live_…")

for day in tt.ephemerides(lat=53.3498, lon=-6.2603, elevation=20): print(day.date, day.sun.rise, day.sun.set)

Why elevation

Most sunrise APIs answer for an observer standing at sea level. The higher you are, the further you can see, and the earlier the sun clears the horizon.

At the Cliffs of Moher (214 m) the difference is about three minutes, and it moves sunrise, sunset and every golden- and blue-hour boundary with them:

Golden hour lasts 20 to 40 minutes, so three minutes is a noticeable share of it. Pass elevation whenever you know it.

What you get

day.sun.rise                        # datetime, with its timezone offset intact

day.sun.golden_hour.morning.start day.moon.phase.name # "Waxing Gibbous" day.moon.phase.illumination # 0.7376

Timestamps arrive as aware datetime objects, not strings. Anything the client does not model is still available on day.raw and day.response, so you can use a new server-side field before the client catches up.

Endpoints

tt.ephemerides(lat=…, lon=…, elevation=…, days=7)     # sun + moon + twilight

tt.prayer_times(lat=…, lon=…, school="mwl") # five daily prayers tt.solar(lat=…, lon=…, timestamp=…) # position + irradiance tt.solar_path(lat=…, lon=…, interval="5m") # the sun's arc, sampled tt.solar_dli(lat=…, lon=…, from_date=…, to_date=…) # Daily Light Integral tt.tide_heights(lat=…, lon=…, step="10m") # water height series tt.tide_extremes(lat=…, lon=…) # high and low waters tt.tide_window(lat=…, lon=…, elevation=…) # tides + sun + moon tt.tide_constituents(lat=…, lon=…) # harmonic constants

Tides take no elevation, with one exception

tide_heights and tide_extremes have no elevation parameter on purpose, because how high *you* stand doesn't change the ocean surface. If you pass one, you get a TypeError instead of having it silently ignored.

tide_window is the exception, because half of what it returns is sun and moon data:

for day in tt.tide_window(lat=52.9715, lon=-9.4309, elevation=214, days=3):

print(day.date, day.sun.golden_hour.morning.start) for t in day.low_waters: print(" low water", t.time, t.height_m)

Tide heights are metres relative to mean sea level, not chart datum, and they exclude weather-driven surge. Not for navigation.

Errors

Each kind of failure has its own exception class, so you can catch the one you care about instead of parsing error messages:

from thirdtrail import PlanLimitExceeded

try: tt.tide_extremes(lat=51.5, lon=-2.7, days=90) except PlanLimitExceeded as e: tt.tide_extremes(lat=51.5, lon=-2.7, days=e.limit) # e.limit == 31

AuthenticationError (401), PermissionDenied (403, a tier you are not on), PlanLimitExceeded and InvalidRequest (400), RateLimited (429), DataUnavailable (503), ServerError (5xx). All subclass ThirdtrailError.

Transient failures and 429s are retried with backoff. Plan limits are not retried, because the same request would fail the same way every time.

Keys

Create one at thirdtrail.life/apis/keys/. A key belongs to one subscription, so it works with one API; a Tides key calling a solar endpoint gets PermissionDenied.

Every API has a free tier.

Not covered yet

The client doesn't cover the bulk (POST) endpoints, iCalendar feeds or webhook management yet. For those, call the REST API directly; the full contract is published as OpenAPI 3.1.

Using an AI agent? There is an MCP server that exposes the same data as tools.

Licence

MIT