Querying data

All forecast endpoints accept the same set of parameters and work with both GET (parameters in the query string) and POST (parameters in a JSON body). Use POST when you request many coordinates, to avoid URL-length limits.

Endpoints

Endpoint	Purpose
`GET/POST /weather/v2/query`	Convenience endpoint; auto-routes to operational or historical based on the requested period
`GET/POST /weather/v2/point/operational`	Latest forecast for one or more points
`GET/POST /weather/v2/point/historical`	Historical forecasts for one or more points (period required)
`GET/POST /weather/v2/area/operational`	Latest forecast over a grid
`GET/POST /weather/v2/area/historical`	Historical forecasts over a grid (period required)
`GET /weather/v2/meta`	Model metadata (latest reference time, availability)
`GET /weather/v2/download/{token}`	Download a generated NetCDF file

All paths are relative to https://api.rebase.energy.

Parameters

Parameter	Required	Default	Description
`model`	yes	—	Model identifier, or several joined with `+` for a multi-model query
`latitude`	yes	—	A single latitude or a comma-separated list (points), or the grid latitude axis
`longitude`	yes	—	A single longitude or a comma-separated list
`variables`	yes	—	Comma-separated variable names; use `Variable:height` for a level
`start-date`	for historical	—	Start of the period (`YYYY-MM-DD` or `YYYY-MM-DD HH:MM`), in `timezone`
`end-date`	for historical	—	End of the period; may be in the future for operational forecasts
`timezone`	no	`UTC`	Timezone for input dates and returned timestamps
`reference-time-freq`	no	`6H`	Frequency of forecast runs to include; clamped up to the model's native run frequency
`valid-time-freq`	no	—	Output time resolution within a forecast (e.g. `1H`); values are interpolated to it
`forecast-horizon`	no	`latest`	`latest`, `full`, or a list of lead-time hours (see below)
`interpolate`	no	`false`	`true` = spatially interpolate to the exact coordinates; `false` = nearest grid point
`output-format`	no	`json`	`json` or `netcdf`
`output-schema`	no	`list`	For JSON: `list` (pandas-style) or `xarray`
`type`	no	`points`	`points` (paired lat/lon) or `grid` (lat × lon mesh)
`wait`	no	`false`	For NetCDF, block until the file is fully generated before responding

The point/* and area/* endpoints fix type to points and grid respectively; on the generic /query endpoint you set type yourself.

GET vs POST

import requests
headers = {"Authorization": "your_api_key"}
url = "https://api.rebase.energy/weather/v2/point/historical"

# GET — lists are comma-separated strings
resp = requests.get(url, headers=headers, params={
    "model": "DWD_ICON-EU",
    "latitude": "60.1, 59.3",
    "longitude": "16.5, 18.1",
    "variables": "Temperature, WindSpeed",
    "start-date": "2023-08-01",
    "end-date": "2023-08-03",
})

# POST — lists can be JSON arrays; same (hyphenated) parameter names
resp = requests.post(url, headers=headers, json={
    "model": "DWD_ICON-EU",
    "latitude": [60.1, 59.3],
    "longitude": [16.5, 18.1],
    "variables": ["Temperature", "WindSpeed"],
    "start-date": "2023-08-01",
    "end-date": "2023-08-03",
})

Points vs grid

Points (type=points): latitude and longitude are paired element-by-element, so the lists must be the same length. Two values each → two points.
Grid (type=grid): latitude and longitude define the axes of a mesh; every latitude is combined with every longitude. Grids are best retrieved with output-schema=xarray (or output-format=netcdf). The area/* endpoints use grid mode.

Forecast horizon

forecast-horizon controls which lead times you get back:

latest (default) — a single, continuous "best available" time series built from the most recent runs (each valid time taken from the latest run that covers it). Ideal for "what's the current forecast for the coming days". Served by the operational path.
full — the entire forecast horizon of each run in the requested period, keyed by both reference time (run) and valid time.
a list of lead-time hours — e.g. [0, 1, 2, 6, 12, 24] returns only those lead times from each run. On GET, pass it as a string: forecast-horizon=[0, 6, 12, 24].

Reference time vs valid time

Reference time (ref_datetime) is when a forecast run was initialised. reference-time-freq selects how many runs to include (e.g. 6H = every 6 hours), and is automatically clamped up to the model's own run frequency.
Valid time (valid_datetime) is the time a value applies to. valid-time-freq resamples/interpolates the output to a target step (e.g. 1H).

Height levels

Request a specific level with the Variable:height notation, e.g. WindSpeed:100 (100 m wind speed) or a native level like Temperature_Height:80.0. The harmonized :10 / :100 levels are documented in Post-processed variables; native level coordinates per model are in Models.

Multi-model queries

Combine models by joining identifiers with +, e.g. model=DWD_ICON-EU+NCEP_GFS. The response is a single continuous series that fills each valid time from the first model that provides it. Multi-model queries are restricted to:

post-processed (harmonized) variables only,
forecast-horizon=latest,
output-format=json, output-schema=list, type=points.

Interpolation

By default (interpolate=false) values come from the nearest model grid point. Set interpolate=true to spatially interpolate to your exact coordinates.