GribStream

Guide de démarrage rapide

Connectez-vous, créez un token d’API (un plan gratuit existe) et envoyez votre première requête. Copiez-collez l’un des exemples ci-dessous (cURL / Python). Remplacez le token et les horodatages, puis exécutez. Remplacez [API_TOKEN] par votre token, sans les crochets.

1. Faire une requête simple

Consultez les données horaires de température pour le 1er mai 2025 à Times Square avec NOAA Global Forecast System (GFS).

URL :

Nous allons envoyer la requête à cet endpoint :

POST https://gribstream.com/api/v2/[MODEL]/timeseries

Renommé Auparavant /history. L’ancien chemin fonctionne toujours.

Remplacez [MODEL] par le code GFS gfs pour obtenir :

POST https://gribstream.com/api/v2/gfs/timeseries

Vous pouvez voir tous les modèles disponibles ici

Payload :

Les horodatages utilisent le format ISO 8601 UTC et nous allons demander une plage temporelle, de fromTime à untilTime.

Les coordonnées sont des paires de nombres à virgule flottante (latitude, longitude).

Les paramètres météorologiques sont sélectionnés avec des triplets (name, level, info). Ce sont des sélecteurs exacts du catalogue : name est le code de paramètre du fournisseur, level est le niveau vertical ou de surface, et info désambiguïse les variantes si nécessaire. Dans GFS, vous verrez que la température est disponible avec TMP à de nombreux niveaux. Utilisons 2m above ground pour cet exemple.

Code :

curl -X POST 'https://gribstream.com/api/v2/gfs/timeseries' \
-H "Authorization: Bearer [API_TOKEN]" \
-d '{
    "fromTime": "2025-05-01T00:00:00Z",
    "untilTime": "2025-05-02T00:00:00Z",
    "coordinates": [{ "lat": 40.758, "lon": -73.985 }],
    "variables": [{ "name": "TMP", "level": "2 m above ground", "info": "" }]
}'

import requests

API_TOKEN = "[API_TOKEN]"  # Replace with your token

payload = {
    "fromTime": "2025-05-01T00:00:00Z",
    "untilTime": "2025-05-02T00:00:00Z",
    "coordinates": [
        { "lat": 40.758, "lon": -73.985 }
    ],
    "variables": [
        { "name": "TMP", "level": "2 m above ground", "info": "" }
    ]
}

headers = {
    "Authorization": f"Bearer {API_TOKEN}",
}

response = requests.post("https://gribstream.com/api/v2/gfs/timeseries", json=payload, headers=headers)
response.raise_for_status()

print(response.text)

Sortie :

forecasted_at,forecasted_time,lat,lon,name,TMP|2 m above ground|
2025-05-01T06:00:00Z,2025-05-01T09:00:00Z,40.7580,-73.9850,,286.7000
2025-05-01T06:00:00Z,2025-05-01T11:00:00Z,40.7580,-73.9850,,286.5539
2025-05-01T00:00:00Z,2025-05-01T01:00:00Z,40.7580,-73.9850,,292.2532
2025-05-01T12:00:00Z,2025-05-01T15:00:00Z,40.7580,-73.9850,,292.2911
2025-05-01T18:00:00Z,2025-05-01T19:00:00Z,40.7580,-73.9850,,292.7923
2025-05-01T12:00:00Z,2025-05-01T14:00:00Z,40.7580,-73.9850,,290.9911
2025-05-01T12:00:00Z,2025-05-01T17:00:00Z,40.7580,-73.9850,,293.8677
2025-05-01T12:00:00Z,2025-05-01T16:00:00Z,40.7580,-73.9850,,293.2000
2025-05-01T18:00:00Z,2025-05-01T20:00:00Z,40.7580,-73.9850,,292.3354
2025-05-01T18:00:00Z,2025-05-01T18:00:00Z,40.7580,-73.9850,,293.5384
2025-05-01T06:00:00Z,2025-05-01T08:00:00Z,40.7580,-73.9850,,287.1673
2025-05-01T12:00:00Z,2025-05-01T13:00:00Z,40.7580,-73.9850,,289.4673
2025-05-01T00:00:00Z,2025-05-01T04:00:00Z,40.7580,-73.9850,,289.4754
2025-05-01T00:00:00Z,2025-05-01T03:00:00Z,40.7580,-73.9850,,290.4711
2025-05-01T00:00:00Z,2025-05-01T05:00:00Z,40.7580,-73.9850,,288.8000
2025-05-01T00:00:00Z,2025-05-01T00:00:00Z,40.7580,-73.9850,,292.9684
2025-05-01T12:00:00Z,2025-05-01T12:00:00Z,40.7580,-73.9850,,287.7405
2025-05-01T00:00:00Z,2025-05-01T02:00:00Z,40.7580,-73.9850,,291.3480
2025-05-01T06:00:00Z,2025-05-01T06:00:00Z,40.7580,-73.9850,,288.2327
2025-05-01T06:00:00Z,2025-05-01T10:00:00Z,40.7580,-73.9850,,286.2595
2025-05-01T18:00:00Z,2025-05-01T22:00:00Z,40.7580,-73.9850,,290.9539
2025-05-01T06:00:00Z,2025-05-01T07:00:00Z,40.7580,-73.9850,,287.7000
2025-05-01T18:00:00Z,2025-05-01T23:00:00Z,40.7580,-73.9850,,290.0113
2025-05-01T18:00:00Z,2025-05-01T21:00:00Z,40.7580,-73.9850,,291.8027

Notez que les lignes ne sont pas ordonnées par forecasted_at ni par forecasted_time. Cela améliore l’efficacité de l’API.

2. Noms d’emplacements et alias de variables

Quand vous traitez la réponse, il peut être très pratique d’utiliser des noms plutôt que la paire de coordonnées. Vous pouvez donner à l’emplacement un nom lisible ou stocker un ID. C’est utile pour créer des visualisations et des tableaux de bord sans prétraitement. Il suffit d’ajouter une propriété name à chaque coordonnée. Dans cet exemple, nous utiliserons TimesSquare.

De même, la convention de nommage en trois parties des paramètres météorologiques peut être peu lisible (TMP|2 m above ground|). Vous pouvez donc renommer la variable avec une propriété supplémentaire appelée alias. C’est utile pour basculer par programme entre des températures à différents niveaux tout en gardant une colonne nommée temp. Pensez-y comme à AS en SQL.

Code :

curl -X POST 'https://gribstream.com/api/v2/gfs/timeseries' \
-H "Authorization: Bearer [API_TOKEN]" \
-d '{
    "fromTime": "2025-05-01T00:00:00Z",
    "untilTime": "2025-05-02T00:00:00Z",
    "coordinates": [{ "lat": 40.758, "lon": -73.985 , "name": "TimesSquare"}],
    "variables": [{ "name": "TMP", "level": "2 m above ground", "info": "", "alias": "temp" }]
}'
import requests

API_TOKEN = "[API_TOKEN]"  # Replace with your token

payload = {
    "fromTime": "2025-05-01T00:00:00Z",
    "untilTime": "2025-05-02T00:00:00Z",
    "coordinates": [
        { "lat": 40.758, "lon": -73.985, "name": "TimesSquare" }
    ],
    "variables": [
        { "name": "TMP", "level": "2 m above ground", "info": "", "alias": "temp" }
    ]
}

headers = {
    "Authorization": f"Bearer {API_TOKEN}",
}

response = requests.post("https://gribstream.com/api/v2/gfs/timeseries", json=payload, headers=headers)
response.raise_for_status()

print(response.text)

Sortie :

forecasted_at,forecasted_time,lat,lon,name,temp
2025-05-01T05:00:00Z,2025-05-01T06:00:00Z,40.7580,-73.9850,TimesSquare,287.7400
2025-05-01T10:00:00Z,2025-05-01T11:00:00Z,40.7580,-73.9850,TimesSquare,285.7200
2025-05-01T03:00:00Z,2025-05-01T04:00:00Z,40.7580,-73.9850,TimesSquare,289.1600
2025-05-01T08:00:00Z,2025-05-01T09:00:00Z,40.7580,-73.9850,TimesSquare,285.7200
2025-05-01T20:00:00Z,2025-05-01T21:00:00Z,40.7580,-73.9850,TimesSquare,291.7000
2025-05-01T04:00:00Z,2025-05-01T05:00:00Z,40.7580,-73.9850,TimesSquare,288.4300
2025-05-01T02:00:00Z,2025-05-01T03:00:00Z,40.7580,-73.9850,TimesSquare,290.1200
2025-05-01T09:00:00Z,2025-05-01T10:00:00Z,40.7580,-73.9850,TimesSquare,285.4300
2025-05-01T06:00:00Z,2025-05-01T07:00:00Z,40.7580,-73.9850,TimesSquare,286.8900
2025-05-01T19:00:00Z,2025-05-01T20:00:00Z,40.7580,-73.9850,TimesSquare,292.5500
2025-05-01T07:00:00Z,2025-05-01T08:00:00Z,40.7580,-73.9850,TimesSquare,286.3500
2025-05-01T15:00:00Z,2025-05-01T16:00:00Z,40.7580,-73.9850,TimesSquare,292.9300
2025-05-01T11:00:00Z,2025-05-01T12:00:00Z,40.7580,-73.9850,TimesSquare,286.8000
2025-05-01T18:00:00Z,2025-05-01T19:00:00Z,40.7580,-73.9850,TimesSquare,293.4100
2025-05-01T17:00:00Z,2025-05-01T18:00:00Z,40.7580,-73.9850,TimesSquare,293.8500
2025-05-01T12:00:00Z,2025-05-01T13:00:00Z,40.7580,-73.9850,TimesSquare,288.1300
2025-05-01T21:00:00Z,2025-05-01T22:00:00Z,40.7580,-73.9850,TimesSquare,291.0500
2025-05-01T13:00:00Z,2025-05-01T14:00:00Z,40.7580,-73.9850,TimesSquare,290.2100
2025-05-01T14:00:00Z,2025-05-01T15:00:00Z,40.7580,-73.9850,TimesSquare,291.3900
2025-05-01T01:00:00Z,2025-05-01T02:00:00Z,40.7580,-73.9850,TimesSquare,291.2300
2025-05-01T00:00:00Z,2025-05-01T01:00:00Z,40.7580,-73.9850,TimesSquare,291.6000
2025-05-01T16:00:00Z,2025-05-01T17:00:00Z,40.7580,-73.9850,TimesSquare,293.6300
2025-04-30T23:00:00Z,2025-05-01T00:00:00Z,40.7580,-73.9850,TimesSquare,292.6500
2025-05-01T22:00:00Z,2025-05-01T23:00:00Z,40.7580,-73.9850,TimesSquare,289.6300

C’est plus lisible.

3. Formats de réponse

Le format de réponse est déterminé par le header HTTP Accept. L’API prend en charge trois formats de réponse :

  • Accept: text/csv - la réponse est un fichier CSV contenant les données de prévision.
  • Accept: application/json - la réponse est un tableau JSON unique d’objets.
  • Accept: application/ndjson - la réponse est du JSON délimité par lignes, avec un objet JSON par ligne.

Si aucun format n’est spécifié, le format de réponse par défaut est text/csv.

Code :

curl -X POST 'https://gribstream.com/api/v2/gfs/timeseries' \
-H "Authorization: Bearer [API_TOKEN]" \
-H "Accept: text/csv" \
-d '{
    "fromTime": "2025-05-01T00:00:00Z",
    "untilTime": "2025-05-02T00:00:00Z",
    "coordinates": [{ "lat": 40.758, "lon": -73.985 , "name": "TimesSquare"}],
    "variables": [{ "name": "TMP", "level": "2 m above ground", "info": "", "alias": "temp" }]
}'

Sortie :

forecasted_at,forecasted_time,lat,lon,name,temp
2025-05-01T05:00:00Z,2025-05-01T06:00:00Z,40.7580,-73.9850,TimesSquare,287.7400
2025-05-01T10:00:00Z,2025-05-01T11:00:00Z,40.7580,-73.9850,TimesSquare,285.7200
2025-05-01T03:00:00Z,2025-05-01T04:00:00Z,40.7580,-73.9850,TimesSquare,289.1600
2025-05-01T08:00:00Z,2025-05-01T09:00:00Z,40.7580,-73.9850,TimesSquare,285.7200
2025-05-01T20:00:00Z,2025-05-01T21:00:00Z,40.7580,-73.9850,TimesSquare,291.7000
2025-05-01T04:00:00Z,2025-05-01T05:00:00Z,40.7580,-73.9850,TimesSquare,288.4300
2025-05-01T02:00:00Z,2025-05-01T03:00:00Z,40.7580,-73.9850,TimesSquare,290.1200
2025-05-01T09:00:00Z,2025-05-01T10:00:00Z,40.7580,-73.9850,TimesSquare,285.4300
2025-05-01T06:00:00Z,2025-05-01T07:00:00Z,40.7580,-73.9850,TimesSquare,286.8900
2025-05-01T19:00:00Z,2025-05-01T20:00:00Z,40.7580,-73.9850,TimesSquare,292.5500
2025-05-01T07:00:00Z,2025-05-01T08:00:00Z,40.7580,-73.9850,TimesSquare,286.3500
2025-05-01T15:00:00Z,2025-05-01T16:00:00Z,40.7580,-73.9850,TimesSquare,292.9300
2025-05-01T11:00:00Z,2025-05-01T12:00:00Z,40.7580,-73.9850,TimesSquare,286.8000
2025-05-01T18:00:00Z,2025-05-01T19:00:00Z,40.7580,-73.9850,TimesSquare,293.4100
2025-05-01T17:00:00Z,2025-05-01T18:00:00Z,40.7580,-73.9850,TimesSquare,293.8500
2025-05-01T12:00:00Z,2025-05-01T13:00:00Z,40.7580,-73.9850,TimesSquare,288.1300
2025-05-01T21:00:00Z,2025-05-01T22:00:00Z,40.7580,-73.9850,TimesSquare,291.0500
2025-05-01T13:00:00Z,2025-05-01T14:00:00Z,40.7580,-73.9850,TimesSquare,290.2100
2025-05-01T14:00:00Z,2025-05-01T15:00:00Z,40.7580,-73.9850,TimesSquare,291.3900
2025-05-01T01:00:00Z,2025-05-01T02:00:00Z,40.7580,-73.9850,TimesSquare,291.2300
2025-05-01T00:00:00Z,2025-05-01T01:00:00Z,40.7580,-73.9850,TimesSquare,291.6000
2025-05-01T16:00:00Z,2025-05-01T17:00:00Z,40.7580,-73.9850,TimesSquare,293.6300
2025-04-30T23:00:00Z,2025-05-01T00:00:00Z,40.7580,-73.9850,TimesSquare,292.6500
2025-05-01T22:00:00Z,2025-05-01T23:00:00Z,40.7580,-73.9850,TimesSquare,289.6300

Code :

curl -X POST 'https://gribstream.com/api/v2/gfs/timeseries' \
-H "Authorization: Bearer [API_TOKEN]" \
-H "Accept: application/json" \
-d '{
    "fromTime": "2025-05-01T00:00:00Z",
    "untilTime": "2025-05-02T00:00:00Z",
    "coordinates": [{ "lat": 40.758, "lon": -73.985 , "name": "TimesSquare"}],
    "variables": [{ "name": "TMP", "level": "2 m above ground", "info": "", "alias": "temp" }]
}'

Sortie :

[{"forecasted_at":"2025-05-01T00:00:00Z","forecasted_time":"2025-05-01T00:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":292.96841796875}
,{"forecasted_at":"2025-05-01T06:00:00Z","forecasted_time":"2025-05-01T11:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":286.553857421875}
,{"forecasted_at":"2025-05-01T00:00:00Z","forecasted_time":"2025-05-01T05:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":288.80002441406253}
,{"forecasted_at":"2025-05-01T00:00:00Z","forecasted_time":"2025-05-01T04:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":289.4754296875}
,{"forecasted_at":"2025-05-01T12:00:00Z","forecasted_time":"2025-05-01T17:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":293.86765136718753}
,{"forecasted_at":"2025-05-01T12:00:00Z","forecasted_time":"2025-05-01T13:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":289.46726074218753}
,{"forecasted_at":"2025-05-01T12:00:00Z","forecasted_time":"2025-05-01T14:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":290.99106445312503}
,{"forecasted_at":"2025-05-01T18:00:00Z","forecasted_time":"2025-05-01T19:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":292.7922607421875}
,{"forecasted_at":"2025-05-01T18:00:00Z","forecasted_time":"2025-05-01T18:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":293.53837890625005}
,{"forecasted_at":"2025-05-01T18:00:00Z","forecasted_time":"2025-05-01T20:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":292.33542480468753}
,{"forecasted_at":"2025-05-01T12:00:00Z","forecasted_time":"2025-05-01T15:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":292.29106445312505}
,{"forecasted_at":"2025-05-01T18:00:00Z","forecasted_time":"2025-05-01T21:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":291.802685546875}
,{"forecasted_at":"2025-05-01T06:00:00Z","forecasted_time":"2025-05-01T09:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":286.70000000000005}
,{"forecasted_at":"2025-05-01T06:00:00Z","forecasted_time":"2025-05-01T06:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":288.23271484375}
,{"forecasted_at":"2025-05-01T06:00:00Z","forecasted_time":"2025-05-01T07:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":287.70000000000005}
,{"forecasted_at":"2025-05-01T18:00:00Z","forecasted_time":"2025-05-01T22:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":290.953857421875}
,{"forecasted_at":"2025-05-01T06:00:00Z","forecasted_time":"2025-05-01T08:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":287.1672607421875}
,{"forecasted_at":"2025-05-01T06:00:00Z","forecasted_time":"2025-05-01T10:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":286.2595458984375}
,{"forecasted_at":"2025-05-01T12:00:00Z","forecasted_time":"2025-05-01T12:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":287.74045410156253}
,{"forecasted_at":"2025-05-01T12:00:00Z","forecasted_time":"2025-05-01T16:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":293.20000000000005}
,{"forecasted_at":"2025-05-01T00:00:00Z","forecasted_time":"2025-05-01T02:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":291.348046875}
,{"forecasted_at":"2025-05-01T00:00:00Z","forecasted_time":"2025-05-01T03:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":290.47107421875}
,{"forecasted_at":"2025-05-01T00:00:00Z","forecasted_time":"2025-05-01T01:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":292.25318359375}
,{"forecasted_at":"2025-05-01T18:00:00Z","forecasted_time":"2025-05-01T23:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":290.0113037109375}
]

Code :

curl -X POST 'https://gribstream.com/api/v2/gfs/timeseries' \
-H "Authorization: Bearer [API_TOKEN]" \
-H "Accept: application/ndjson" \
-d '{
    "fromTime": "2025-05-01T00:00:00Z",
    "untilTime": "2025-05-02T00:00:00Z",
    "coordinates": [{ "lat": 40.758, "lon": -73.985 , "name": "TimesSquare"}],
    "variables": [{ "name": "TMP", "level": "2 m above ground", "info": "", "alias": "temp" }]
}'

Sortie :

{"forecasted_at":"2025-05-01T12:00:00Z","forecasted_time":"2025-05-01T12:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":287.74045410156253}
{"forecasted_at":"2025-05-01T00:00:00Z","forecasted_time":"2025-05-01T05:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":288.80002441406253}
{"forecasted_at":"2025-05-01T12:00:00Z","forecasted_time":"2025-05-01T15:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":292.29106445312505}
{"forecasted_at":"2025-05-01T06:00:00Z","forecasted_time":"2025-05-01T06:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":288.23271484375}
{"forecasted_at":"2025-05-01T12:00:00Z","forecasted_time":"2025-05-01T16:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":293.20000000000005}
{"forecasted_at":"2025-05-01T18:00:00Z","forecasted_time":"2025-05-01T22:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":290.953857421875}
{"forecasted_at":"2025-05-01T00:00:00Z","forecasted_time":"2025-05-01T01:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":292.25318359375}
{"forecasted_at":"2025-05-01T00:00:00Z","forecasted_time":"2025-05-01T02:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":291.348046875}
{"forecasted_at":"2025-05-01T00:00:00Z","forecasted_time":"2025-05-01T03:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":290.47107421875}
{"forecasted_at":"2025-05-01T12:00:00Z","forecasted_time":"2025-05-01T13:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":289.46726074218753}
{"forecasted_at":"2025-05-01T00:00:00Z","forecasted_time":"2025-05-01T04:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":289.4754296875}
{"forecasted_at":"2025-05-01T12:00:00Z","forecasted_time":"2025-05-01T14:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":290.99106445312503}
{"forecasted_at":"2025-05-01T00:00:00Z","forecasted_time":"2025-05-01T00:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":292.96841796875}
{"forecasted_at":"2025-05-01T06:00:00Z","forecasted_time":"2025-05-01T07:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":287.70000000000005}
{"forecasted_at":"2025-05-01T12:00:00Z","forecasted_time":"2025-05-01T17:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":293.86765136718753}
{"forecasted_at":"2025-05-01T06:00:00Z","forecasted_time":"2025-05-01T08:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":287.1672607421875}
{"forecasted_at":"2025-05-01T18:00:00Z","forecasted_time":"2025-05-01T18:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":293.53837890625005}
{"forecasted_at":"2025-05-01T06:00:00Z","forecasted_time":"2025-05-01T09:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":286.70000000000005}
{"forecasted_at":"2025-05-01T18:00:00Z","forecasted_time":"2025-05-01T19:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":292.7922607421875}
{"forecasted_at":"2025-05-01T06:00:00Z","forecasted_time":"2025-05-01T10:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":286.2595458984375}
{"forecasted_at":"2025-05-01T18:00:00Z","forecasted_time":"2025-05-01T20:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":292.33542480468753}
{"forecasted_at":"2025-05-01T18:00:00Z","forecasted_time":"2025-05-01T21:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":291.802685546875}
{"forecasted_at":"2025-05-01T06:00:00Z","forecasted_time":"2025-05-01T11:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":286.553857421875}
{"forecasted_at":"2025-05-01T18:00:00Z","forecasted_time":"2025-05-01T23:00:00Z","lat":40.758,"lon":-73.985,"name":"TimesSquare","temp":290.0113037109375}
4. Sélectionner des heures précises plutôt qu’une plage temporelle

Il est parfois utile de sélectionner une liste d’heures précises plutôt qu’une plage temporelle. Pour cela, utilisez l’argument timesList dans la requête. Sur /timeseries, chaque valeur est une heure de validité à retourner, correspondant à forecasted_time.

Par exemple, consultez la météo pendant le Nouvel An 2022, 2023, 2024 et 2025.

Code :

curl -X POST 'https://gribstream.com/api/v2/gfs/timeseries' \
-H "Authorization: Bearer [API_TOKEN]" \
-d '{
    "timesList": [
        "2022-01-01T00:00:00Z",
        "2023-01-01T00:00:00Z",
        "2024-01-01T00:00:00Z",
        "2025-01-01T00:00:00Z"
    ],
    "coordinates": [{ "lat": 40.758, "lon": -73.985 , "name": "TimesSquare"}],
    "variables": [{ "name": "TMP", "level": "2 m above ground", "info": "", "alias": "temp" }]
}'

Sortie :

forecasted_at,forecasted_time,lat,lon,name,temp
2022-01-01T00:00:00Z,2022-01-01T00:00:00Z,40.7580,-73.9850,TimesSquare,283.1841
2024-01-01T00:00:00Z,2024-01-01T00:00:00Z,40.7580,-73.9850,TimesSquare,279.0649
2023-01-01T00:00:00Z,2023-01-01T00:00:00Z,40.7580,-73.9850,TimesSquare,281.7180
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,40.7580,-73.9850,TimesSquare,281.8039
5. Sélectionner des emplacements avec une grille

Au lieu de consulter des données pour des coordonnées spécifiques, vous pouvez définir une grille avec minLatitude, maxLatitude, minLongitude, maxLongitude et step, où step est la résolution en degrés.

Voici un exemple visuel avec notre démo de champs de vent

Code :

curl -X POST 'https://gribstream.com/api/v2/gfs/timeseries' \
-H "Authorization: Bearer [API_TOKEN]" \
-d '{
    "timesList": [ "2025-01-01T00:00:00Z" ],
    "grid": {
        "minLatitude": 25.00,
        "maxLatitude": 27.00,
        "minLongitude": -122.00,
        "maxLongitude": -120.00,
        "step": 0.5
    },
    "variables": [{ "name": "TMP", "level": "2 m above ground", "info": "", "alias": "temp" }]
}'

Sortie :

forecasted_at,forecasted_time,lat,lon,name,temp
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,27.0000,-122.0000,,290.0439
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,27.0000,-121.5000,,289.8739
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,27.0000,-121.0000,,289.7339
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,27.0000,-120.5000,,289.6239
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,27.0000,-120.0000,,289.5039
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,26.5000,-122.0000,,290.2839
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,26.5000,-121.5000,,290.1139
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,26.5000,-121.0000,,289.9739
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,26.5000,-120.5000,,289.8339
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,26.5000,-120.0000,,289.7239
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,26.0000,-122.0000,,290.5739
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,26.0000,-121.5000,,290.4239
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,26.0000,-121.0000,,290.2439
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,26.0000,-120.5000,,290.0339
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,26.0000,-120.0000,,289.8339
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,25.5000,-122.0000,,290.9539
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,25.5000,-121.5000,,290.7239
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,25.5000,-121.0000,,290.4739
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,25.5000,-120.5000,,290.2339
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,25.5000,-120.0000,,290.0439
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,25.0000,-122.0000,,291.1239
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,25.0000,-121.5000,,290.9539
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,25.0000,-121.0000,,290.7139
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,25.0000,-120.5000,,290.4539
2025-01-01T00:00:00Z,2025-01-01T00:00:00Z,25.0000,-120.0000,,290.2239
6. Filtrage par Lead Time (minLeadTime, maxLeadTime)

Le Lead Time est l’écart entre le moment où une prévision a été émise (forecasted_at) et l’heure de validité à laquelle elle s’applique (forecasted_time).

  • minLeadTime : inclut uniquement les valeurs prédites avec au moins cette avance (par exemple "12h", "45m").
  • maxLeadTime : inclut uniquement les valeurs prédites avec au plus cette avance (par exemple "3h").

Autrement dit, ces paramètres contrôlent le Lead Time minimal et maximal de chaque valeur prévisionnelle récupérée (minHorizon/maxHorizon sont d’anciens alias obsolètes qui acceptent des heures entières).

Exemple : pour voir uniquement les valeurs prédites au moins 12 heures à l’avance :

Code :

curl -X POST 'https://gribstream.com/api/v2/nbm/timeseries' \
-H "Authorization: Bearer [API_TOKEN]" \
-d '{
    "fromTime": "2025-05-01T12:00:00Z",
    "untilTime": "2025-05-01T18:00:00Z",
    "minLeadTime": "12h",
    "coordinates": [{ "lat": 40.758, "lon": -73.985, "name": "TimesSquare" }],
    "variables": [{ "name": "TMP", "level": "2 m above ground", "alias": "temp" }]
}'

Sortie :

forecasted_at,forecasted_time,lat,lon,name,temp,horizon_hours
            2025-05-01T00:00:00Z,2025-05-01T12:00:00Z,40.7580,-73.9850,TimesSquare,286.6700
            2025-05-01T01:00:00Z,2025-05-01T13:00:00Z,40.7580,-73.9850,TimesSquare,288.0300
            2025-05-01T02:00:00Z,2025-05-01T14:00:00Z,40.7580,-73.9850,TimesSquare,289.5900
            2025-05-01T03:00:00Z,2025-05-01T15:00:00Z,40.7580,-73.9850,TimesSquare,290.9600
            2025-05-01T04:00:00Z,2025-05-01T16:00:00Z,40.7580,-73.9850,TimesSquare,292.3300
            2025-05-01T05:00:00Z,2025-05-01T17:00:00Z,40.7580,-73.9850,TimesSquare,293.1900

Note : les lignes sont triées ici par forecasted_at pour faciliter la lecture du résultat.

Autre exemple : pour inclure uniquement les prédictions à court terme jusqu’à 3 heures : 

curl -X POST 'https://gribstream.com/api/v2/nbm/timeseries' \
-H "Authorization: Bearer [API_TOKEN]" \
-d '{
    "fromTime": "2025-05-01T12:00:00Z",
    "untilTime": "2025-05-01T15:00:00Z",
    "maxLeadTime": "3h",
    "coordinates": [{ "lat": 40.758, "lon": -73.985 }],
    "variables": [{ "name": "TMP", "level": "2 m above ground" }]
}'

Utilisez minLeadTime et maxLeadTime ensemble pour limiter les valeurs prévisionnelles à la fenêtre de prédiction souhaitée.

Astuce : contrairement au paramètre asOf (voir l’exemple suivant), qui limite les exécutions éligibles par heure d'exécution du modèle, minLeadTime et maxLeadTime filtrent selon l’âge de la prévision par rapport à la valeur prédite.

7. Coupure par heure d'exécution du modèle avec asOf (pour le backtesting)

Pour le backtesting de modèles prédictifs, vous devez souvent exclure les cycles de modèle générés après un instant de décision historique.

Par exemple : “Quelle était la prévision pour le 1er mai, en utilisant seulement les exécutions du modèle générées au plus tard le 30 avril à 12:00 UTC ?”

asOf? est une coupure par heure d'exécution du modèle : seules les exécutions générées à ce moment ou avant sont éligibles. Ce n'est pas une heure exacte de disponibilité dans l'API en direct ; si l'heure de première disponibilité importe, appliquez une marge conservatrice fondée sur le retard habituel de publication du fournisseur pour les horizons utilisés.

Code :

curl -X POST 'https://gribstream.com/api/v2/nbm/timeseries' \
-H "Authorization: Bearer [API_TOKEN]" \
-d '{
    "fromTime": "2025-05-01T00:00:00Z",
    "untilTime": "2025-05-02T00:00:00Z",
    "asOf": "2025-05-01T12:00:00Z",
    "coordinates": [{ "lat": 40.758, "lon": -73.985 , "name": "TimesSquare"}],
    "variables": [{ "name": "TMP", "level": "2 m above ground", "info": "", "alias": "temp" }]
}'

Sortie :

forecasted_at,forecasted_time,lat,lon,name,temp
2025-04-30T23:00:00Z,2025-05-01T00:00:00Z,40.7580,-73.9850,TimesSquare,292.6500
2025-05-01T00:00:00Z,2025-05-01T01:00:00Z,40.7580,-73.9850,TimesSquare,291.6000
2025-05-01T01:00:00Z,2025-05-01T02:00:00Z,40.7580,-73.9850,TimesSquare,291.2300
2025-05-01T02:00:00Z,2025-05-01T03:00:00Z,40.7580,-73.9850,TimesSquare,290.1200
2025-05-01T03:00:00Z,2025-05-01T04:00:00Z,40.7580,-73.9850,TimesSquare,289.1600
2025-05-01T04:00:00Z,2025-05-01T05:00:00Z,40.7580,-73.9850,TimesSquare,288.4300
2025-05-01T05:00:00Z,2025-05-01T06:00:00Z,40.7580,-73.9850,TimesSquare,287.7400
2025-05-01T06:00:00Z,2025-05-01T07:00:00Z,40.7580,-73.9850,TimesSquare,286.8900
2025-05-01T07:00:00Z,2025-05-01T08:00:00Z,40.7580,-73.9850,TimesSquare,286.3500
2025-05-01T08:00:00Z,2025-05-01T09:00:00Z,40.7580,-73.9850,TimesSquare,285.7200
2025-05-01T09:00:00Z,2025-05-01T10:00:00Z,40.7580,-73.9850,TimesSquare,285.4300
2025-05-01T10:00:00Z,2025-05-01T11:00:00Z,40.7580,-73.9850,TimesSquare,285.7200
2025-05-01T11:00:00Z,2025-05-01T12:00:00Z,40.7580,-73.9850,TimesSquare,286.8000
2025-05-01T12:00:00Z,2025-05-01T13:00:00Z,40.7580,-73.9850,TimesSquare,288.1300
2025-05-01T12:00:00Z,2025-05-01T14:00:00Z,40.7580,-73.9850,TimesSquare,290.1600
2025-05-01T12:00:00Z,2025-05-01T15:00:00Z,40.7580,-73.9850,TimesSquare,291.3400
2025-05-01T12:00:00Z,2025-05-01T16:00:00Z,40.7580,-73.9850,TimesSquare,292.6500
2025-05-01T12:00:00Z,2025-05-01T17:00:00Z,40.7580,-73.9850,TimesSquare,293.0100
2025-05-01T12:00:00Z,2025-05-01T18:00:00Z,40.7580,-73.9850,TimesSquare,293.1500
2025-05-01T12:00:00Z,2025-05-01T19:00:00Z,40.7580,-73.9850,TimesSquare,292.7500
2025-05-01T12:00:00Z,2025-05-01T20:00:00Z,40.7580,-73.9850,TimesSquare,292.1500
2025-05-01T12:00:00Z,2025-05-01T21:00:00Z,40.7580,-73.9850,TimesSquare,291.3500
2025-05-01T12:00:00Z,2025-05-01T22:00:00Z,40.7580,-73.9850,TimesSquare,290.7100
2025-05-01T12:00:00Z,2025-05-01T23:00:00Z,40.7580,-73.9850,TimesSquare,289.5800

Note : les lignes sont triées ici par forecasted_at pour faciliter la lecture du résultat.

Cet exemple utilise NOAA National Blend of Models (NBM), qui s’exécute toutes les heures. Notez que forecasted_at est toujours l’heure précédant forecasted_time, ce qui correspond à la meilleure prévision pour cette heure, sauf pour les temps postérieurs à asOf, défini à 2025-05-01T12:00:00Z. Les exécutions de modèle postérieures à cet instant sont exclues.

8. Expressions calculées (valeurs dérivées, champs calculés, formules, etc.)

GribStream permet de définir des expressions pour calculer des colonnes dérivées, des transformations et des filtres booléens dans vos requêtes API.

Les expressions utilisent le langage expr, avec des opérations numériques, logiques et sur valeurs de type string, ainsi que des fonctions mathématiques.

Pour voir un exemple de tableau de bord reposant sur une seule requête API : simulation de croissance du maïs

Fonctionnalités

  • Utiliser les colonnes du dataset et les expressions calculées précédemment comme entrées
  • Chaîner les expressions : les expressions suivantes peuvent utiliser les résultats des précédentes
  • Appliquer des filtres booléens pour inclure ou exclure des lignes
  • Utiliser les fonctions mathématiques enregistrées par GribStream (voir ci-dessous)

Exemples d’expressions

  • Vitesse du vent à partir des composantes U/V : func.Hypot(uwind, vwind)
  • Direction du vent en degrés météorologiques : int(270 - func.Atan2(vwind, uwind) * 180 / 3.14159) % 360
  • Filtre pour vent fort en surface : func.Hypot(uwind, vwind) > 10

Code :

curl -X POST 'https://gribstream.com/api/v2/hrrr/timeseries' \
-H "Content-Type: application/json" \
-H "Authorization: Bearer [API_TOKEN]" \
-d '{
    "fromTime": "2024-09-10T00:00:00Z",
    "untilTime": "2024-09-10T10:00:00Z",
    "coordinates": [{ "lat": 40.7306, "lon": -73.9352, "name": "New York City" }],
    "variables": [
    { "name": "UGRD", "level": "1000 mb", "alias": "uwind" },
    { "name": "VGRD", "level": "1000 mb", "alias": "vwind" }
    ],
    "expressions": [
    { "expression": "func.Hypot(uwind, vwind)", "alias": "wind_magnitude" },
    { "expression": "int(270 - func.Atan2(vwind, uwind) * 180 / 3.14159) % 360", "alias": "wind_direction" }
    ]
}'

Sortie :

forecasted_at,forecasted_time,lat,lon,name,uwind,vwind,wind_direction,wind_magnitude
2024-09-10T02:00:00Z,2024-09-10T03:00:00Z,40.7306,-73.9352,New York City,7.1213,1.7680,256,7.3375
2024-09-10T01:00:00Z,2024-09-10T02:00:00Z,40.7306,-73.9352,New York City,8.1815,1.6363,258,8.3435
2024-09-10T00:00:00Z,2024-09-10T01:00:00Z,40.7306,-73.9352,New York City,8.0572,2.9389,249,8.5765
2024-09-10T05:00:00Z,2024-09-10T06:00:00Z,40.7306,-73.9352,New York City,6.8368,0.1001,269,6.8375
2024-09-10T04:00:00Z,2024-09-10T05:00:00Z,40.7306,-73.9352,New York City,6.4087,3.0921,244,7.1157
2024-09-10T03:00:00Z,2024-09-10T04:00:00Z,40.7306,-73.9352,New York City,6.4841,2.5385,248,6.9633
2024-09-09T23:00:00Z,2024-09-10T00:00:00Z,40.7306,-73.9352,New York City,6.1795,3.9539,237,7.3362
2024-09-10T08:00:00Z,2024-09-10T09:00:00Z,40.7306,-73.9352,New York City,5.9424,-0.6889,276,5.9822
2024-09-10T06:00:00Z,2024-09-10T07:00:00Z,40.7306,-73.9352,New York City,7.3567,1.5594,258,7.5201
2024-09-10T07:00:00Z,2024-09-10T08:00:00Z,40.7306,-73.9352,New York City,6.5438,0.7486,263,6.5864
9. Filtrer ou rechercher des conditions avec des expressions calculées

GribStream vous permet de définir des expressions pour filtrer dans l’espace et dans le temps, en ne retournant que les données qui respectent la condition.

Les expressions de filtre suivent la même syntaxe que les expressions calculées, mais elles doivent évaluer à true ou false.

Pour voir un exemple de tableau de bord reposant sur une seule requête API : chasseurs d’orages

Code :

curl -X POST 'https://gribstream.com/api/v2/gfs/timeseries' \
-H "Content-Type: application/json" \
-H "Authorization: Bearer [API_TOKEN]" \
-d '{
    "fromTime": "2024-09-10T00:00:00Z",
    "untilTime": "2024-09-10T10:00:00Z",
    "grid": {
            "minLatitude": 24.52,
            "maxLatitude": 49.38,
            "minLongitude": -124.77,
            "maxLongitude": -66.93,
            "step": 0.5
        },
        "variables": [
            {"name": "CAPE", "level": "180-0 mb above ground", "info": "", "alias": "cape"},
            {"name": "CIN", "level": "180-0 mb above ground", "info": "", "alias": "cin"}
        ],
        "expressions":[
            { "expression": "cape + cin", "alias": "storm_severity"}
        ],
        "filter":{"expression": "storm_severity >= 1200"}
}'

Sortie :

forecasted_at,forecasted_time,lat,lon,name,cape,cin,offset_hours,storm_severity
2024-09-10T00:00:00Z,2024-09-10T01:00:00Z,29.0200,-70.7700,,1337.0000,-1.6324,-5713.0000,1335.3676
2024-09-10T00:00:00Z,2024-09-10T01:00:00Z,29.0200,-70.2700,,1204.0000,-1.6324,-5713.0000,1202.3676
2024-09-10T00:00:00Z,2024-09-10T01:00:00Z,28.5200,-71.7700,,1432.0000,-0.6324,-5713.0000,1431.3676
2024-09-10T00:00:00Z,2024-09-10T01:00:00Z,28.5200,-71.2700,,1384.0000,-1.6324,-5713.0000,1382.3676
2024-09-10T00:00:00Z,2024-09-10T01:00:00Z,28.5200,-70.7700,,1355.0000,-0.6324,-5713.0000,1354.3676
2024-09-10T00:00:00Z,2024-09-10T01:00:00Z,28.5200,-70.2700,,1388.0000,-1.6324,-5713.0000,1386.3676
2024-09-10T00:00:00Z,2024-09-10T01:00:00Z,28.0200,-112.2700,,1599.0000,-148.6324,-5713.0000,1450.3676
2024-09-10T00:00:00Z,2024-09-10T01:00:00Z,28.0200,-111.2700,,1424.0000,-104.6324,-5713.0000,1319.3676
2024-09-10T00:00:00Z,2024-09-10T01:00:00Z,28.0200,-75.2700,,1328.0000,-0.6324,-5713.0000,1327.3676
2024-09-10T00:00:00Z,2024-09-10T01:00:00Z,28.0200,-74.7700,,1242.0000,-1.6324,-5713.0000,1240.3676
2024-09-10T00:00:00Z,2024-09-10T01:00:00Z,28.0200,-73.2700,,1327.0000,0.3676,-5713.0000,1327.3676
2024-09-10T00:00:00Z,2024-09-10T01:00:00Z,28.0200,-72.7700,,1471.0000,0.3676,-5713.0000,1471.3676
...
10. Récupérer les exécutions de modèle

Tous les exemples précédents utilisent l’endpoint timeseries (anciennement history). Son objectif est de retourner la meilleure prévision possible pour chaque temps. La meilleure prévision est toujours la plus récente, autrement dit celle avec le plus faible Lead Time. Ici, le Lead Time est l’écart entre l’heure d’exécution du modèle (forecasted_at) et l’heure de validité (forecasted_time). Les paramètres minLeadTime, maxLeadTime et asOf limitent les exécutions de modèle prises en compte, mais une seule exécution est choisie pour chaque valeur prévisionnelle.

Pour récupérer toutes les exécutions de chaque valeur, il faut utiliser l’endpoint runs (anciennement forecasts).

POST https://gribstream.com/api/v2/[MODEL]/runs

Cet endpoint choisit la plage temporelle avec les paramètres forecastedFrom et forecastedUntil afin d’indiquer clairement que les données sont sélectionnées par heure d’exécution du modèle, le même horodatage retourné comme forecasted_at. Sur /runs, timesList énumère aussi des heures d’exécution, ce qui est utile pour cibler des cycles précis.

Tous les autres paramètres restent disponibles et identiques, sauf asOf, qui n’est pas disponible. Vous pouvez donc filtrer avec minLeadTime et maxLeadTime, définir des expressions avec des valeurs calculées, filtrer sur ces valeurs, etc. (minHorizon/maxHorizon sont d’anciens alias obsolètes qui acceptent des heures entières.)

Code :

curl -X POST 'https://gribstream.com/api/v2/nbm/runs' \
-H "Authorization: Bearer [API_TOKEN]" \
-d '{
    "forecastedFrom": "2024-09-10T00:00:00Z",
    "forecastedUntil": "2024-09-10T03:00:00Z",
    "minLeadTime": "1h",
    "maxLeadTime": "3h",
    "coordinates": [
        { "lat": 47.6, "lon": -122.33 }
    ],
    "variables": [
        { "name": "TMP", "level": "2 m above ground", "info": "", "alias": "temp" },
        { "name": "WIND", "level": "10 m above ground", "info": "", "alias": "wind_speed" },
        { "name": "DPT", "level": "2 m above ground", "info": "", "alias": "dew_point" }
    ]
    }'

Sortie :

forecasted_at,forecasted_time,lat,lon,name,dew_point,temp,wind_speed
2024-09-10T00:00:00Z,2024-09-10T01:00:00Z,47.6000,-122.3300,,288.2500,294.2700,1.6000
2024-09-10T00:00:00Z,2024-09-10T02:00:00Z,47.6000,-122.3300,,288.4000,293.4800,1.6000
2024-09-10T00:00:00Z,2024-09-10T03:00:00Z,47.6000,-122.3300,,288.2600,293.0000,1.6000
2024-09-10T01:00:00Z,2024-09-10T02:00:00Z,47.6000,-122.3300,,288.3500,293.4900,1.6000
2024-09-10T01:00:00Z,2024-09-10T03:00:00Z,47.6000,-122.3300,,288.1800,292.9800,1.6000
2024-09-10T01:00:00Z,2024-09-10T04:00:00Z,47.6000,-122.3300,,288.1900,292.2100,1.2000
2024-09-10T02:00:00Z,2024-09-10T03:00:00Z,47.6000,-122.3300,,288.1600,292.9500,1.6000
2024-09-10T02:00:00Z,2024-09-10T04:00:00Z,47.6000,-122.3300,,288.3800,292.1800,1.2000
2024-09-10T02:00:00Z,2024-09-10T05:00:00Z,47.6000,-122.3300,,287.9300,291.2800,1.2000
2024-09-10T03:00:00Z,2024-09-10T04:00:00Z,47.6000,-122.3300,,288.3300,292.0300,1.2000
2024-09-10T03:00:00Z,2024-09-10T05:00:00Z,47.6000,-122.3300,,287.9400,291.4900,1.2000
2024-09-10T03:00:00Z,2024-09-10T06:00:00Z,47.6000,-122.3300,,287.8000,290.8100,1.2000

Note : les lignes sont triées ici par forecasted_at pour faciliter la lecture du résultat. Notez que chaque exécution du modèle renvoie les horizons 1 à 3.

11. Consulter les prévisions d’ensemble et choisir des membres

Certains modèles dans GribStream, comme NOAA Global Ensemble Forecast System (GEFS Atmos) et ECMWF Integrated Forecasting System - Ensemble Forecast (IFS Enfo), sont des ensembles. Chaque exécution contient 30 à 50 membres d’ensemble qui partent de conditions initiales légèrement différentes. Consulter tous les membres permet de quantifier l’incertitude (spread), de construire des probabilités (par exemple “80 % de chance de pluie > 25 mm”) et de détecter des extrêmes peu probables qu’une exécution déterministe unique pourrait manquer.

Vous pouvez choisir un ou plusieurs membres d’ensemble avec le champ members. S’il n’est pas fourni, la valeur par défaut retourne uniquement le membre 0, c’est-à-dire la prévision de contrôle.

Code :

curl -X POST 'https://gribstream.com/api/v2/gefsatmos/timeseries' \
-H "Content-Type: application/json" \
-H "Authorization: Bearer [API_TOKEN]" \
-d '{
    "fromTime": "2021-02-24T00:00:00Z",
    "untilTime": "2021-02-24T12:00:00Z",
    "minLeadTime": "0h",
    "maxLeadTime": "3h",
    "coordinates": [
        { "lat": 40.75, "lon": -73.98, "name": "home"}
    ],
    "variables": [
        { "name": "TMP", "level": "2 m above ground", "info": "", "alias": "temp" }
    ],
    "members": [1,2,3]
}'

Sortie :

forecasted_at,forecasted_time,lat,lon,name,member,temp
2021-02-24T00:00:00Z,2021-02-24T00:00:00Z,40.7500,-73.9800,home,1,277.0675
2021-02-24T00:00:00Z,2021-02-24T00:00:00Z,40.7500,-73.9800,home,2,276.9957
2021-02-24T00:00:00Z,2021-02-24T00:00:00Z,40.7500,-73.9800,home,3,277.1360
2021-02-24T00:00:00Z,2021-02-24T03:00:00Z,40.7500,-73.9800,home,1,275.1206
2021-02-24T00:00:00Z,2021-02-24T03:00:00Z,40.7500,-73.9800,home,2,275.0662
2021-02-24T00:00:00Z,2021-02-24T03:00:00Z,40.7500,-73.9800,home,3,275.7054
2021-02-24T06:00:00Z,2021-02-24T06:00:00Z,40.7500,-73.9800,home,1,275.9707
2021-02-24T06:00:00Z,2021-02-24T06:00:00Z,40.7500,-73.9800,home,2,275.3141
2021-02-24T06:00:00Z,2021-02-24T06:00:00Z,40.7500,-73.9800,home,3,275.7535
2021-02-24T06:00:00Z,2021-02-24T09:00:00Z,40.7500,-73.9800,home,1,273.9249
2021-02-24T06:00:00Z,2021-02-24T09:00:00Z,40.7500,-73.9800,home,2,273.8208
2021-02-24T06:00:00Z,2021-02-24T09:00:00Z,40.7500,-73.9800,home,3,274.3771

Note : les lignes sont triées ici par forecasted_at pour faciliter la lecture du résultat. Notez que chaque exécution du modèle renvoie les horizons 1 à 3.

12. Avancé : découvrir datasets et sélecteurs depuis le catalogue

Si vous connaissez déjà le dataset et les sélecteurs voulus, sautez cette section et allez directement à /timeseries ou /runs. C’est toujours la manière normale d’utiliser GribStream.

Le catalogue sert aux cas plus avancés : construire des sélecteurs de datasets, des tableaux de bord multimodèles, des outils internes ou des applications qui doivent découvrir les sélecteurs au lieu de les écrire en dur.

Étape 1 : lister les datasets

curl 'https://gribstream.com/api/v2/catalog/datasets'

Cela retourne les métadonnées publiques de chaque dataset : code, nom complet, fournisseur, tags, cadence d’exécution, informations d’archive, liens vers les sources et liens vers la page du modèle et les endpoints API correspondants.

[
  {
    "code": "gfs",
    "name": "GFS",
    "full_name": "Global Forecast System",
    "provider": "NOAA",
    "models_page_url": "/models/gfs",
    "api_timeseries_url": "/api/v2/gfs/timeseries",
    "api_runs_url": "/api/v2/gfs/runs"
  }
]

Étape 2 : inspecter les paramètres d’un dataset

curl 'https://gribstream.com/api/v2/catalog/datasets/ifsoper/parameters'

Chaque résumé de paramètre inclut le code court exact fourni par la source, le nom complet, les unités, le résumé et l’indication d’éventuelles variations de sélecteur. Les codes de paramètres sont sensibles à la casse ; si un dataset expose 100u, réutilisez exactement 100u tel qu’il est retourné.

[
  {
    "short_name": "100u",
    "full_name": "100 metre U wind component",
    "units": "m s-1",
    "variation_count": 1
  }
]

Pour inspecter un paramètre en détail, avec toutes ses variations et des sélecteurs prêts à copier :

curl 'https://gribstream.com/api/v2/catalog/datasets/ifsoper/parameters/100u'
{
  "short_name": "100u",
  "full_name": "100 metre U wind component",
  "variations": [
    {
      "selector": { "name": "100u", "level": "sfc", "info": "" },
      "selector_literal": "{\"name\":\"100u\",\"level\":\"sfc\",\"info\":\"\"}"
    }
  ]
}

Étape 3 : résoudre un paramètre partagé entre datasets

C’est le parcours de catalogue le plus avancé. Il est utile quand votre application veut un concept partagé, comme la vitesse du vent à 10 m, mais que différents datasets l’exposent avec des sélecteurs ou des expressions différents.

curl 'https://gribstream.com/api/v2/catalog/shared-parameters/wind_speed_10m?dataset=ifsoper&alias=wind'

La réponse inclut un fragment resolved_request lisible par machine que vous pouvez copier dans le payload d’une requête de prévision pour ce dataset.

{
  "code": "wind_speed_10m",
  "resolved_dataset": "ifsoper",
  "resolved_supported": true,
  "resolved_request": {
    "variables": [
      { "name": "10u", "level": "sfc", "alias": "u_component", "hidden": true },
      { "name": "10v", "level": "sfc", "alias": "v_component", "hidden": true }
    ],
    "expressions": [
      { "expression": "func.Hypot(u_component, v_component)", "alias": "wind" }
    ]
  }
}

Une fois que vous avez des sélecteurs ou un resolved_request, utilisez-les dans les mêmes endpoints /timeseries et /runs présentés plus haut dans ce guide.

Pour le schéma complet du catalogue et d’autres exemples, consultez Endpoints, OpenAPI et la spécification brute dans /docs/openapi.yaml.

Contactez-nous pour les scénarios non couverts par ce guide : info@gribstream.com.