specification: API Commons Rate Limits
specificationVersion: '0.1'
schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/RateLimits
provider: Saildrone
providerId: saildrone
created: '2026-05-24'
modified: '2026-05-24'
reconciled: false
tags:
  - Maritime
  - Ocean Data
  - USV
description: >
  Saildrone has not published rate-limit numbers for the Public Mission API (BETA)
  on developer-mission.saildrone.com. Observed behavior from open source clients
  (e.g. nwfsc-fram/SaildroneTS) shows pagination via `limit` (default 1000 rows)
  and `interval` (minutes between samples) parameters rather than per-key request
  budgets. Bearer tokens issued by POST /v1/auth are short-lived and clients are
  expected to refresh them. Documented values below are derived from client code
  and Saildrone's stated "schema-driven, AI/ML-ready" design rather than from a
  published rate-limit page.
sources:
  - https://developer-mission.saildrone.com/api-docs
  - https://github.com/nwfsc-fram/SaildroneTS
  - https://www.saildrone.com/data-delivery/api
policies:
  - id: saildrone-mission-api-pagination
    name: Mission API pagination window
    scope: per-request
    surface: https://developer-mission.saildrone.com/v1/timeseries/{mission}
    type: pagination
    metric: records
    limit: 1000
    timeFrame: per-request
    description: >
      The /v1/timeseries/{mission} endpoint paginates by `limit` rows per
      response, defaulting to 1000 in widely-used reference clients. Clients
      re-window using the latest gps_time and call again to advance.
    headers: []
    overage: continue-paging
  - id: saildrone-mission-api-interval
    name: Mission API sampling interval
    scope: per-request
    surface: https://developer-mission.saildrone.com/v1/timeseries/{mission}
    type: thinning
    metric: minutes
    limit: 1
    timeFrame: per-sample
    description: >
      Clients can pass `interval` (minutes) to thin the time series at the
      server side; default of 1 minute matches Saildrone's native 1-minute
      averaged METOC cadence.
    headers: []
    overage: server-thins
  - id: saildrone-mission-api-auth-token-ttl
    name: Bearer token TTL
    scope: per-token
    surface: https://developer-mission.saildrone.com/v1/auth
    type: token-lifetime
    metric: tokens
    limit: 1
    timeFrame: short-lived
    description: >
      Bearer tokens are short-lived; observed Saildrone reference clients
      re-authenticate before each batch run. Saildrone has not published a fixed
      TTL value.
    headers: []
    overage: re-authenticate
  - id: saildrone-open-data-pmel-erddap
    name: NOAA PMEL ERDDAP rate limits
    scope: per-ip
    surface: https://data.pmel.noaa.gov/generic/erddap/
    type: shared-data-portal
    metric: requests
    limit: -1
    timeFrame: per-day
    description: >
      Saildrone open data hosted on NOAA PMEL ERDDAP follows ERDDAP's own
      fair-use rate limits and per-IP concurrency caps, which are not
      Saildrone-controlled. Heavy users should follow ERDDAP best practices.
    headers: []
    overage: throttled-by-erddap