Texas Water Watch

Data access

The same source-linked information, for readers and software.

Download 45 source documents from 20 issuing districts, covering 298 physical source pages with selected source or scheduled dates from 2012-03-21 through 2026-08-21. The district coverage file includes all 98 statewide directory identities; 78 have no source documents in this collection. These are document counts, not independently established meetings held.

Source-linked research summary. Verify important details against the official district record. Historical records with incomplete district coverage. Recorded decisions do not establish current permit status or that conditions were fulfilled.

Browse Meeting intelligence → · Entities and timelines →

Downloads

Individual meeting JSON is linked from every meeting page and index entry. Plain GET requests need no account, cookies or JavaScript.

Separate collections: 182 Facts, 60 Board Actions, statewide district directory and document coverage, and official-source navigation catalog. Counts refer to different kinds of records and must not be added together. The directory provides document counts, roles and official archive links; individual document payloads are not included in this site download.

Available meeting records

Read and cite a record

Start with the catalog, check district coverage, then use the meeting index to retrieve an individual record. A district with zero meeting records is a coverage gap, not evidence of no activity.

Stable identity: Record keys and local item IDs are retained from R3 and Batch 02. R3 local identifiers are reversibly scoped to their record; Batch 02 identifiers already include the record scope. Each object has a citation URL and JSON pointer. Similar names across records do not establish identity.
Content profiles: record_format selects r3 (nested observations and procedural actions) or batch02 (water_objects, episodes, quantities, ordered tables and unavailable_references). Both share source identity, discovery URLs, dates and evidence links. The schema defines both profiles; no empty R3 arrays are invented for Batch 02.
Tables and row navigation: Each table preserves columns, ordered rows, stage and notes. Row navigation provides a stable fragment and JSON pointer into rows. Evidence remains table-level when the supplied source gives no row-specific page. Null, zero, negative numbers, ranges and lists remain distinct. Human tables use the same values as JSON.
Values and units: Observation values preserve nested numbers, original unit labels, time bases, source text and associations. Field names such as requested_production_af_per_year_as_repeated retain their stated role and denominator. Unknown denominators stay unknown. Never sum repeated allocation rows or convert inconsistent source calculations into an adopted quantity.
Decisions and conditions: Observation type, procedural action, result and epistemic state distinguish a request, recommendation, approval, interpretation and unresolved issue. A condition is not evidence it was fulfilled.
Dates: Meeting/source dates, approval dates and routing dates have distinct roles. Extraction, last source check and publication dates remain null when unknown. A scheduled date does not prove an event occurred. Build time is not a government-source observation.
Evidence: Evidence includes the official source URL, SHA-256 identity and physical page. Batch 02 uses PDF_PAGE_REFERENCE: no exact quote, page-image hash or uniquely assigned reading method is claimed. Its reading-method descriptions remain reported provenance. Native quotations, explicitly shortened excerpts and visual-source descriptions are different fields. Image hashes identify evidence; source images and PDFs are not republished.
Mentions and candidate leads: The entity-mentions endpoint provides dated occurrences, source-local roles/stages, object links and page evidence. Lookup groups only normalize case, spaces and punctuation; permit/well identifiers also carry their issuing district and identifier type. They do not establish legal identity. Research leads preserve possible_linkage_only and identity_merged=false; resolved older extraction-index references are navigation candidates, not corroboration. Timeline dates use explicit item dates where supplied, otherwise clearly labeled document context.
Provenance and limitations: Records derive from the supplied R3 and Batch 02 extractions, with selected source-page checks by a coding agent. No independent human review is documented; R3’s original false review flags remain false and Batch 02’s unspecified flags remain null. Its reported extraction date is September 17, 2026; missing R3 extraction dates remain null. Source completeness, unreadable pages, unavailable attachments and uncertain relationships remain in the records. This is a publicly released research collection. Publication does not change the original review flags or validate historical assertions. The record_state field retains the original derivative status independently of website delivery.

Cite both the record’s canonical_url plus the object fragment in citation_url, and the official source URL with its physical page. Texas Water Sources is the research publisher; the district is the official source issuer.

Updates and usage

Schema version 1.0.0 on the /v1/ path defines explicit r3 and batch02 profiles. Consumers must inspect record_format. Incompatible changes require a new major path. Record content versions change when their public content changes. Stable IDs do not change when a title or date is corrected. Index versions support incremental retrieval.

The actual deployment time and source commit are recorded in the first cumulative release record. A build time is not a source-check time. The existing site terms apply. No separate dataset reuse license is currently specified; this download does not grant a new Creative Commons or CC0 license or claim ownership of government records. Dataset reuse terms remain a publisher decision.

Plain HTTP example

The repository includes a small reader that starts only from the catalog, discovers coverage and returns cross-batch mentions, profile-aware items, typed table rows and candidate leads with citations:

node scripts/read-water-watch.mjs https://texaswatersources.com/data/water-watch/v1/index.json "Aqua Texas"

For a local build, replace the catalog URL with your local server’s catalog URL. The reader remaps catalog links to that origin while preserving production citation URLs. Try BLF, PGCD-000838, MWCZ-1B, RC6447 or --coverage-gap. Server-side readers do not require browser CORS. Browser cross-origin access depends on the static host’s headers.