Data & Forecast Disclaimer

Last updated: May 29, 2026

1 About This Service

This site aggregates real-time, recent, and archival water-temperature data from federal, state, tribal, university, watershed-council, and citizen-science sources across the United States, then publishes 7-day statistical forecasts at each monitoring location using publicly available gridded weather as model input.

The service is operated by Advanced Field Technologies (Oregon, USA) and is intended for fisheries biologists, watershed managers, anglers, river guides, agency staff, researchers, and the public as a quick-reference and screening tool.

2 Data Sources & Attribution

Underlying observations are pulled from the public agencies, research institutions, and citizen-science platforms listed below. Each contributor retains copyright and other rights in its data, subject to the open-data license attached to that dataset (USGS public-domain, CC BY 4.0, ODbL, and others). Agency names, station IDs, dataset titles, and trademarks are the property of their owners.

Attribution: if you publish, redistribute, or build on top of any specific dataset shown on this site, please carry forward the originating contributor's name and license, and consult the licence terms attached to the dataset on the contributor's portal. We do not relicense or warrant the data.

3 Data Quality and Limitations

3.1 Provisional and unvalidated

All real-time data delivered through this service is provisional — it has not undergone the QA review the originating agencies and contributors perform before publishing finalized records. Sensors can drift, ice over, fail, or be temporarily out of service. Telemetered values occasionally include noise, gaps, or erroneous readings. The data on this site is presented as received from the upstream APIs, with only minimal automated filtering (physically impossible values are rejected).

3.2 Citizen-science observations

A share of this site's coverage comes from community-collected data (watershed councils, tribal monitoring programs, and university citizen-science partnerships, including data flowing through the EPA Water Quality Portal). These collectors follow standardized protocols and use calibrated equipment, but the data has typically not gone through the same QA pipeline as government telemetry. Treat citizen-science readings with additional caution and consult the originating dataset's metadata for collector identity, methods, and equipment.

3.3 No independent validation

This service does not independently validate, calibrate, or audit the instruments used at any station. Accuracy depends entirely on the maintenance schedule and quality controls of the operating agency or volunteer organization. We do not certify any reading or forecast as accurate to any particular tolerance.

3.4 Coverage gaps

Not every monitoring station reports water temperature continuously. Some report only flow or stage; others report seasonally; some are flagged "active" but have current sensor outages. Where a station has no current observation, the map shows today's forecasted mean with a "FORECAST" label so users can distinguish observed from predicted values.

4 Forecast Methodology and Limitations

4.1 Method

For stations with sufficient recent observations, water-temperature forecasts use a Ridge regression model trained on each station's past 14–365+ days of observed water temperature, with the NOAA NWS gridded weather forecast as the primary forward input. For stations without enough observed data — including most CO-DWR, NEON, WQP, NERRS, and KISTERS sites — the system falls back to a weather-driven equilibrium model that estimates water temperature from local air temperature, dewpoint, and humidity without using any observed water-temp history. The forecast method used for each station is shown on its detail page.

4.2 Forecast accuracy varies

Ridge models with extensive training data (R² > 0.9, sample size > 200 days) typically predict within ±1 °C for the next 24-hour period and ±2 °C for the 7-day horizon under typical conditions. Equilibrium-mode forecasts are substantially less accurate — often ±3 °C or worse — particularly during rapid-onset weather events, snowmelt pulses, dam releases, or in spring-fed and groundwater-dominated streams. Each forecast row displays a confidence range; treat the upper and lower bounds as a rough guide, not a guarantee.

4.3 Known limitations

• Models do not account for local management actions (dam releases, hatchery operations, irrigation withdrawals, fire-related diversions).
• Models do not account for unusual sensor placement, recent relocation, or biofouling.
• Lake, reservoir, and estuarine surface temperatures are particularly hard to forecast and may carry larger errors than free-flowing river/stream sites.
• During extreme heat events, low-flow periods, or atmospheric inversions, models may underestimate peak temperatures.
• Precomputed forecasts on this site update once per 24 hours (nightly at 02:39 Pacific). They do not respond intra-day to rapidly changing weather.

5 Not for Safety-Critical Decisions

Including but not limited to:

• Recreational decisions involving cold-water hypothermia risk (whitewater paddling, swimming, cold-water immersion).
• Flood, debris-flow, or wildfire-response operations.
• Fish-kill mitigation, hatchery operations, or fisheries regulatory closures — consult agency biologists and the official agency feed.
• Compliance with NPDES permits, TMDLs, ESA / SARA consultation, or any other regulatory program.
• Litigation, expert testimony, or evidentiary submissions.

For any of the above, consult the originating agency or contributor directly for finalized, quality-controlled records.

6 User Responsibilities

By using this service you agree to:

• Cross-check any decision-relevant reading or forecast against the originating agency's published data.
• Independently verify the accuracy of any data, forecast, or result before relying on it.
• Respect each contributor's licence and attribution requirements when redistributing data.
• Not present values from this service as "official" or "validated" without acknowledging this disclaimer.
• Report obvious data errors or sensor problems to the operating agency or contributor, not to us — we do not maintain the underlying instruments or volunteer programs.

7 No Warranty · Limitation of Liability

See also our Terms of Service for the binding liability cap.

8 Acknowledgments

We acknowledge and thank the contributing data providers whose continued collection and publication make this service possible:

• United States: U.S. Geological Survey, NOAA (NWS, NDBC, CO-OPS, NERRS), U.S. EPA Water Quality Portal partners, USFS NorWeST, NEON, Oregon Water Resources Department, Washington Department of Ecology, California Department of Water Resources (CDEC), Colorado Division of Water Resources, Bureau of Reclamation (HydroMet PN), U.S. Army Corps of Engineers (CWMS), Columbia Basin Research (DART), and the NSF LTER and NPS Water Resources Division networks.
• Tribal monitoring: Confederated Tribes of the Umatilla Indian Reservation (CTUIR) and other partner Tribes whose thermograph data flows through the Water Quality Portal.
• Weather forcing: NOAA National Weather Service and Open-Meteo.

The reliability of this service rests entirely on their work.

9 Updates to This Disclaimer

We may revise this disclaimer to reflect changes in data sources, forecast methodology, or applicable law. Material updates will be announced by email to active subscribers. Continued use of the service after an update constitutes acceptance.

10 Contact

Questions, corrections, or contributor concerns: info@advancedfield.tech.

See also: Privacy Policy · Terms of Service