Self-host the ERA5 archive via Open-Meteo (object storage) (#224)

Get the 45-year historical record off the rate-limited public Open-Meteo
archive API by running a private Open-Meteo instance that serves the
era5_seamless blend (0.1° ERA5-Land + 0.25° ERA5 for gusts) from the
compressed .om archive in object storage, mounted on the host with rclone.

- climate.py: make ARCHIVE_URL env-driven (THERMOGRAPH_ARCHIVE_URL) and pin
  models=era5_seamless on the archive fetches only, so a self-hosted instance
  serves the same 0.1° resolution; forecast path unchanged. The public API's
  default is already seamless, so dev/beta (URL unset) behave identically.
- docker-compose.openmeteo.yml: open-meteo-api + two rolling sync workers
  (era5_land 0.1°, era5 0.25° for gusts), bind-mounting the object-storage
  mount; the overlay points the app at the local instance.
- Makefile: om-up / om-down / om-backfill (one-time full-history backfill).
- Terraform: per-host openmeteo flag layers the overlay, renders OM_DATA_DIR,
  and provisions the host rclone systemd mount from the bucket credentials.
- deploy/openmeteo: operator runbook + rclone mount unit template.

data/climate.py

@@ -29,7 +29,16 @@ ARCHIVE_LATENCY_DAYS = 6           # ERA5 archive lags real time by a few days
 HISTORY_TOPUP_INTERVAL = 3600      # seconds between recent-tail refresh attempts
 FORECAST_TTL_HOURS = 1             # refetch the forward forecast hourly to track updates
 
-ARCHIVE_URL = "https://archive-api.open-meteo.com/v1/archive"
+# The archive (historical) endpoint. Overridable so it can point at a self-hosted
+# Open-Meteo instance (ERA5 in object storage) instead of the rate-limited public
+# API; the response shape is identical either way.
+ARCHIVE_URL = os.environ.get("THERMOGRAPH_ARCHIVE_URL",
+                             "https://archive-api.open-meteo.com/v1/archive")
+# The historical model: the ERA5 "seamless" blend — 0.1° ERA5-Land for
+# temperature/precip/humidity/wind, 0.25° ERA5 for wind gusts (which ERA5-Land
+# lacks). This is the public API's default; sent explicitly so a self-hosted
+# instance serves the same 0.1° resolution. Archive requests only, not forecast.
+ARCHIVE_MODEL = "era5_seamless"
 FORECAST_URL = "https://api.open-meteo.com/v1/forecast"
 # Backup archive when Open-Meteo is unavailable (e.g. its daily rate limit). NASA
 # POWER is free + keyless, global, daily from 1981. It lacks gusts + apparent temp,
@@ -324,7 +333,7 @@ def _to_frame(daily: dict) -> pl.DataFrame:
 
 def _fetch_history(cell: dict) -> pl.DataFrame:
     end = (datetime.date.today() - datetime.timedelta(days=ARCHIVE_LATENCY_DAYS)).isoformat()
-    params = _om_daily_params(cell, start_date=START_DATE, end_date=end)
+    params = _om_daily_params(cell, start_date=START_DATE, end_date=end, models=ARCHIVE_MODEL)
     r = _request(ARCHIVE_URL, params, 180, phase="history_fetch")
     return _to_frame(r.json()["daily"])
 
@@ -470,7 +479,7 @@ def _fetch_forecast_metno(cell: dict) -> pl.DataFrame:
 
 def _fetch_history_range(cell: dict, start_date: str, end_date: str) -> pl.DataFrame:
     """Fetch just a date range of archive history (used to top up the recent tail)."""
-    params = _om_daily_params(cell, start_date=start_date, end_date=end_date)
+    params = _om_daily_params(cell, start_date=start_date, end_date=end_date, models=ARCHIVE_MODEL)
     r = _request(ARCHIVE_URL, params, 60, phase="history_topup")
     return _to_frame(r.json()["daily"])
 

tests/data/test_climate.py

@@ -216,3 +216,36 @@ def test_derive_metrics_adds_wetbulb_and_absolute_humidity():
     assert "wetbulb" in out.columns
     # _derive_humidity replaces raw RH (50) with absolute humidity (g/m³, ~8-9).
     assert out["humid"][0] < 30 and out["humid"][0] != 50.0
+
+
+def test_history_fetch_targets_archive_url_with_seamless_model(monkeypatch):
+    """Historical fetches hit ARCHIVE_URL (env-overridable for self-hosting) and pin
+    the ERA5 seamless model, so a self-hosted Open-Meteo serves the same 0.1° blend."""
+    seen = {}
+
+    class Resp:
+        def json(self): return {"daily": _om_daily(3)}
+
+    def fake_request(url, params, timeout, *, phase, headers=None, attempts=climate.MAX_ATTEMPTS):
+        seen["url"], seen["params"], seen["phase"] = url, dict(params), phase
+        return Resp()
+
+    monkeypatch.setattr(climate, "_request", fake_request)
+    cell = {"center_lat": 47.6, "center_lon": -122.3}
+
+    climate._fetch_history(cell)
+    assert seen["url"] == climate.ARCHIVE_URL
+    assert seen["params"]["models"] == "era5_seamless"
+    assert seen["phase"] == "history_fetch"
+
+    climate._fetch_history_range(cell, "2026-06-01", "2026-06-10")   # tail top-up
+    assert seen["url"] == climate.ARCHIVE_URL
+    assert seen["params"]["models"] == "era5_seamless"
+
+
+def test_daily_params_carry_no_model_by_default():
+    # The shared param builder (also used by the forecast path) stays model-free;
+    # only the archive fetches add models=era5_seamless.
+    p = climate._om_daily_params({"center_lat": 1.0, "center_lon": 2.0}, past_days=7)
+    assert "models" not in p
+    assert climate.ARCHIVE_URL.endswith("/v1/archive")   # default (no env override in tests)