H1 2026 Evaluation
Japan Reservoir Transparency
B+78Good — Ranked #13 out of 167 countries
weight 30%
weight 20%
weight 15%
weight 13%
weight 10%
weight 8%
weight 5%
Primary source evaluated
MLIT River Disaster Prevention Information (river.go.jp) + MLIT Data Platform (mlit-data.jp) + Hydrological & Water Quality Database (www1.river.go.jp)
https://www.river.go.jp/Dimension breakdown
Data Availability
20% of total score
Japan operates one of the most operationally rich dam monitoring systems in the world. The MLIT River Disaster Prevention Information portal (river.go.jp) publishes real-time data every 10 minutes for 735+ individual dams nationwide, including reservoir water level (貯水位), storage volume (貯水量), storage rate (貯水率), inflow, and discharge. The Hydrological and Water Quality Database (www1.river.go.jp) consolidates historical hourly/daily records since FY1993 across rainfall, water level, discharge, water quality, sediment, groundwater, snow depth, and dam/weir management quantities. The NILIM ダム諸量データベース (mudam.nilim.go.jp) provides per-dam statistical archives. As of Sept 2023, these systems are progressively being federated through the MLIT Data Platform (国土交通データプラットフォーム), which by Sept 2024 linked 21 systems and 2.76M datasets including 水文水質データベース and ダム便覧. Coverage of nationally and prefecturally managed dams is comprehensive. Score is not higher because Japan's 150,000+ agricultural irrigation ponds and many small private/utility dams remain outside national portal coverage.
Technical Accessibility
15% of total score
Re-evaluated upward. Japan does have an officially documented REST API: the MLIT Data Platform User API (国土交通DPF利用者API), released September 2023, with published reference documentation, OpenAPI-style query specs, JSON responses, and a formal data interoperability standard (データ連携標準仕様書 v1.0, Sept 2024). The platform federates 21 systems including the 水文水質データベース and ダム便覧, and in November 2025 added an MCP server enabling natural-language access via AI clients. Access requires free account registration and an API key (Bearer auth). The dataset catalogue at search.ckan.jp / geospatial.jp exposes the hydrological/water-quality dataset (id 9005) with documented JSON download and API spec. Real-time kawabou dam quantities are still primarily served through CGI-bin endpoints (DspDamData.exe) returning HTML, which third-party projects (jacopen/damapi, WeatherNote/japan-dam-map) wrap to produce JSON; the paid 水防災オープンデータ提供サービス distributes structured telemetry for ~680 dam stations at 10-60 min cadence via River Information Center but is fee-based (not open). Score is not higher because (a) the DPF API documentation, registration, and most metadata remain Japanese-only; (b) real-time per-dam storage at 10-min resolution is not yet fully exposed as a stable open REST endpoint — the structured open path goes through the DPF catalog rather than a single live dam-storage API; (c) the open-data license picture is fragmented (Government Standard Terms / Public Data License v1.0 are CC BY 4.0 compatible for most MLIT data, but 水文水質DB uses a custom license).
Coverage
30% of total score
Conservative estimate — denominator includes the ~150,000 small agricultural irrigation ponds (溜池) not in MLIT live monitoring, private hydropower dams (TEPCO, KEPCO, J-POWER operational storage published only as facility metadata not live storage), and many small private/utility dams outside national portal coverage. Methodology v1.3.0 (capacity-weighted). Covered capacity through the MLIT kawabou real-time portal (735+ individual dams nationwide with 10-minute storage-volume readings) is approximately 20,000 hm³. A realistic national denominator including private hydropower operational storage and the aggregate of larger agricultural ponds reaches approximately 22,700 hm³. Score = round(100 × 20,000 / 22,700) = 88. Japan's strategic multipurpose and large hydroelectric reservoirs are very well covered; the conservative discount reflects the systematic exclusion of private utility live data and the long tail of small agricultural reservoirs.
Historical Depth
13% of total score
The NILIM ダム諸量データベース contains data from fiscal year 1993 (平成5年) onwards — approximately 33 years of historical hourly/daily records covering dam storage, inflow, outflow, water temperature, turbidity, and water quality. The broader 水文水質データベース (www1.river.go.jp), established following the 1997 River Law revision, consolidates monitoring records including dam management data going back to at least FY1993. Pre-1993 dam storage time series are not available through public national portals; researchers requiring older records must apply to individual dam management offices. Score is held at 72 (not raised) because 33 years is solid but materially shorter than US (USGS to 1880s), Spain (CEDEX 70+ years), or France (Banque Hydro 100+ years) records.
Update Frequency
10% of total score
The river.go.jp real-time portal updates dam parameters (storage level, inflow, discharge, storage rate) every 10 minutes, confirmed both by the official glossary documentation ('10分ごとに観測されるダム諸量を提供') and by independent implementations (jacopen/damapi, WeatherNote/japan-dam-map). The paid telemetry service updates dam stations every 10-60 minutes, crisis-management water-level gauges every 2-10 minutes, and XRAIN rainfall every minute. Historical/archived data is available at hourly resolution via the 水文水質DB. A score of 90 rather than 100 reflects that not all 735+ dams update identically (some JWA realtime pages are weekday-only), and the DPF API exposes the federated catalog rather than streaming live 10-min ticks directly.
Methodological Transparency
8% of total score
Dam capacity and technical specifications are extensively documented. The 川の防災情報 glossary defines storage-rate calculations precisely, distinguishing 貯水率(利水容量)(beneficial-use capacity) from 貯水率(有効容量)(effective capacity). The damnet.or.jp ダム便覧 publishes per-dam technical specs including height, capacity, type, purpose, completion year, and operator. The National Land Numerical Information W01 GIS dataset publishes the same in GML/Shapefile. The MLIT Data Platform published a formal 'データ連携標準仕様書' (Data Interoperability Standard Spec) v1.0 in September 2024, defining schema, identifiers, and cross-system linking for federated datasets including dam/river data. Individual MLIT regional bureau dam-management offices publish detailed technical data sheets. Bumped from 70 to 75 reflecting the new formal interoperability standard. Score is not higher because explicit measurement uncertainty, sensor-calibration protocols, and a unified machine-readable national station-metadata catalog are still not centrally published.
Language and Usability
5% of total score
Japan's dam-data ecosystem remains overwhelmingly Japanese-only. The river.go.jp real-time portal has only a minimal English stub at /e/. The www1.river.go.jp historical database, mudam.nilim.go.jp dam database, MLIT Data Platform (mlit-data.jp), DPF API documentation, the CKAN catalog (search.ckan.jp), and the National Land Numerical Information GIS downloads are all Japanese-only. The damnet.or.jp ダム便覧 maintains an English index linking to English fact sheets for major dams. JWA publishes English profiles for its 27 multipurpose dams but no live data in English. MLIT publishes English brochures on river administration but not the data interfaces themselves. A non-Japanese-speaking researcher cannot effectively navigate the national portals or register for the DPF API without language assistance or automated translation. Marginal bump from 28 to 30 reflects the broader English presence of JWA dam pages and the multiple English summary PDFs from MLIT, but the operational language barrier is essentially unchanged.
Evaluator notes
Re-evaluation conducted 2026-05-29 raises Japan's RTI score from 67.3 (B-) to approximately 73.0 (B). The key correction: the previous evaluation stated Japan has 'no officially documented stable REST API' and gave technical_accessibility a 42. This was materially wrong as of the evaluation window. The MLIT Data Platform User API (国土交通DPF利用者API) has been live since September 2023, with formal reference documentation, OpenAPI-style query schemas, JSON responses, account/API-key authentication, and a published Data Interoperability Standard v1.0 (Sept 2024). By Sept 2024 the platform federated 21 systems and 2.76 million datasets including the 水文水質データベース and ダム便覧, and in November 2025 MLIT released a Model Context Protocol (MCP) server allowing natural-language queries against the platform — a notable open-data innovation. Technical accessibility is therefore raised from 42 to 62. The second correction is upward on data_availability (82→85) and methodological_transparency (70→75): the federated DPF catalog plus the new formal data-interoperability spec materially strengthen both. Coverage is bumped 68→70 reflecting confirmed ~680 telemetry dam stations in the (paid) River Information Center feed and 735+ dams in the free kawabou portal, against a national large-dam universe of ~3,100. Update_frequency is held at 90 (10-min real-time, confirmed). Historical_depth held at 72 (FY1993+, solid but shorter than US/EU peers). Language_usability nudged 28→30 — still the dominant weakness; the DPF API and its docs are Japanese-only, and registering for an API key without Japanese capability is impractical. The revised composite of 73.0 (B) places Japan below Korea (77.0, B+) and Australia (78.6, B+) but closes the gap meaningfully. Korea's WAMIS edge comes from a single unified bilingual portal with documented English-language Open API for dam hydrologic data; Australia's BOM Water Data Online edge comes from a single WISKI-API endpoint with daily-updated water-storage time series for ~5,000 stations under a clear CC BY license. Japan's monitoring infrastructure is arguably more operationally sophisticated than either (10-min cadence, 735+ dams, 33-year archive), but is held back by (a) fragmented access paths across river.go.jp, www1.river.go.jp, mudam.nilim.go.jp, the paid River Information Center feed, and the newer DPF API; (b) the absence of a single open REST endpoint streaming real-time per-dam storage volumes; and (c) the near-total Japanese-only interface that excludes non-Japanese researchers. Japan should reach Korea/Australia territory once the DPF API exposes the kawabou real-time dam quantities as a first-class JSON endpoint with English documentation.
Evaluated by Jaime Delgado · 2026-05-29 · Methodology v1.3.0