China Reservoir Transparency

The Ministry of Water Resources (MWR) publishes an annual China Water Resources Bulletin (中国水资源公报) since 1998 with aggregate year-end storage figures for large and medium reservoirs, freely accessible on szy.mwr.gov.cn. However, individual reservoir-level current storage is not systematically published on a public portal. The Yangtze River Water Resources Commission (cjh.com.cn) posts near-real-time water levels for the Three Gorges system, but individual-reservoir current storage volume across the national estate is absent from any single public-facing source. The MWR National Hydrological Information Centre (xxfb.mwr.cn) hosts daily reservoir data used by researchers, but public access is inconsistent and the site is frequently unreachable from outside China. Operational data at the basin commission level (YRCC, CWRC) is substantially restricted — researchers explicitly state they have no right to re-share it.

No public REST API exists for reservoir storage data. An MWR hydrological API was launched in July 2021 covering 1,996 monitoring stations with twice-daily updates, but it is distributed exclusively through the Alibaba Cloud commercial marketplace (market.aliyun.com), requiring a paid subscription and account registration — it is not a free open API. Annual bulletins are published as PDFs in Chinese only. The MWR data portal (mwr.gov.cn/sj/) is Chinese-only and provides structured access only via static documents. The xxfb.mwr.cn site provides tabular data but lacks a documented API and is frequently unreachable. Scraping is the de facto method used by third-party monitors (e.g., yngtz.xyz pulling from cjh.com.cn). No open-format bulk data download (CSV, JSON) is available for the national reservoir estate.

Coverage methodology v1.2.0 uses reservoirs with storage capacity >10 hm³ as the denominator. China's official inventory recognises ~753 large-type reservoirs (>100 hm³) plus ~3,896 medium-type reservoirs (~10-100 hm³), giving n_total_above_10hm³ ≈ 4,000-4,650. The Res-CN academic dataset compiled 3,254 reservoirs covering 73.2% of national storage capacity, consistent with this range. n_covered by public individual-reservoir storage feeds is extremely small: the Three Gorges system on cjh.com.cn provides near-real-time water levels for ~10 stations on the Yangtze cascade; YRCC, CWRC and a few basin commissions publish water levels for perhaps a dozen more headline dams; the rest of the national reservoir estate has no individual public storage data. Conservative estimate of unique reservoirs with any public per-reservoir storage data ≈ 30-50. Linear coverage = round(100 × 40 / 4500) = 1. Score is set at 8 (rather than 1) reflecting that the MWR annual aggregate Water Resources Bulletin does provide national-level coverage of the 4,649 large+medium reservoirs as a single yearly figure — partial transparency at the aggregate level even though per-reservoir public coverage is near-zero. Academic datasets (Res-CN: 3,254 reservoirs = 73.2% of capacity; ESSD 2022: 338 reservoirs = 50% of capacity) demonstrate the data exists in government archives, but these were assembled by researchers with special access, not via public portals.

Annual Water Resources Bulletins have been published since 1998, providing ~28 years of aggregate national reservoir storage data in PDF form. However, individual-reservoir machine-readable historical time series are not publicly available. Researchers building Res-CN obtained daily water level records from government sources covering 2015–2021 (with a gap in 2018–2019) for 93 validation reservoirs, but these are not directly accessible by the public. The ESSD 2022 dataset covers satellite-derived storage for 338 reservoirs back to 2010 — a 12-year machine-readable series — but this relies on remote sensing, not government ground data. WMO-standard hydrological yearbooks (中国水文年鉴) contain decades of station records but are effectively paywalled or restricted to registered institutional users within China.

For the handful of major river system stations (e.g., Three Gorges at cjh.com.cn), water level readings are published hourly or twice-daily and are publicly scraped by third-party monitors. The MWR commercial API (Alibaba Cloud) offers twice-daily updates for 1,996 stations since July 2021. For the national reservoir estate, however, the publicly accessible update is annual: the Water Resources Bulletin covers only year-end aggregate totals. The xxfb.mwr.cn site appears to provide more frequent data but is not reliably accessible and has no documented public update cadence. The gap between internal government monitoring (real-time SCADA networks across ~18,000 rainfall stations) and what is actually released publicly is very large.

China's hydrological measurement standards are documented and have been standardized since the 2011 Code for Compilation and Publication of Hydrological Yearbook, applied nationally. The MWR reports an error rate of 1/100,000 and a national quality score of 96.8 in hydrological data compilation. The MWR annual bulletin describes its methodology in general terms (Thiessen polygon method, isoline method, data from ~18,000 rainfall stations). However, reservoir-specific measurement protocols — how storage volume is derived from water level, rating curve documentation, uncertainty estimates — are not publicly documented at the individual facility level. The hydrological yearbooks contain fuller methodology but are not accessible without institutional subscription.

All primary government reservoir data portals (mwr.gov.cn, szy.mwr.gov.cn, xxfb.mwr.cn, cjh.com.cn, yrcc.gov.cn) are exclusively in Simplified Chinese. The MWR publishes an English-language website (mwr.gov.cn/english) with summary news and some statistical bulletins, but these are high-level aggregates without individual reservoir data. An English-language Statistical Bulletin on China Water Activities (covering 2017) is available as a PDF from the MWR English portal, but it contains only national aggregate tables — not individual reservoir storage. No machine-readable English-language portal or API for reservoir storage data exists. Third-party international aggregators (MacroMicro, Probe International) repackage Three Gorges water level data in English but cover only a single facility.