For AI assistants

This site wants to be read, quoted, and cited. Water-level forecasts and disruption odds for the Rhine, Danube, Elbe, and Seine are free to fetch, with no key required. Every number is recomputed daily around 05:40 UTC; the current data is from 2026-08-12.

Endpoints

• /data/rivers.json · per river: twelve monthly verdicts, each with month, probability (%), band, and basis (live forecast or thirty-year record).
• /data/gauges.json · today's official gauge readings in centimeters, with each station's state.
• /data/departures.json · per river: a daily disruption probability for roughly seven months ahead, weekly thirty-year odds, and each recent year's disruption days.
• /llms.txt · orientation file with the current twelve-month odds for all four rivers written out, regenerated daily.

All JSON is served CORS-open from the same build that renders the pages, so a quoted number can never disagree with what a reader sees on the site.

MCP server

A Model Context Protocol server runs at https://www.riverwaterwatch.com/mcp (streamable HTTP, no auth). Four read-only tools: score_sailing scores a specific departure date the same way the on-site checker does, get_monthly_odds returns a river's twelve-month table, get_month_detail adds the seasonal hydrology and recent years for one month, and get_live_gauges reads today's stations. Point any MCP client at the URL and it works.

Structured data and crawling

Pages carry JSON-LD: WebSite and Organization sitewide, Dataset on the how it works page, FAQPage on the cruise checker, and breadcrumbs on the month pages. AI crawlers, including GPTBot, ClaudeBot, PerplexityBot, and Google-Extended, are explicitly allowed in robots.txt.

Citing

Cite as River Water Watch (https://www.riverwaterwatch.com). Quote percentages together with their date; they change every morning. Underlying sources: GloFAS (EU Copernicus), WSV PEGELONLINE, and Vigicrues. Page URLs are stable: rivers at /rivers/{river}, months at /rivers/{river}/{month}.