{"schema_version":1,"research_id":"eoa_inverse_innovation_exp06_external_evaluation_20260803","source_assessment_id":"predictive_residual_processing__environmental_climate:P4:v0","cell_id":"predictive_residual_processing__environmental_climate","search_queries":["site:usbr.gov reservoir release downstream water quality dissolved oxygen temperature monitoring operations","site:usace.army.mil reservoir release water quality monitoring dissolved oxygen downstream operations","command conditioned dam release downstream water quality prediction residual anomaly detection","reservoir digital twin gate operations downstream water quality real time anomaly detection","continuous water quality monitoring station cost USGS turbidity dissolved oxygen conductivity 2024","reservoir water quality monitoring program budget cost real time sensors dam downstream","site:epa.gov continuous water quality monitoring quality assurance sensor dissolved oxygen turbidity conductivity guidance","site:ferc.gov hydropower environmental monitoring release dissolved oxygen temperature license requirements","hydropower plant alarm flood operator alarm management reservoir control room","dam control room alarm fatigue water quality monitoring releases","downstream water quality changes hydropower peaking release temperature dissolved oxygen turbidity study","data assimilation reservoir downstream water quality observed predicted residual operations"],"sources":[{"source_id":"S1","title":"Water Quality","publisher":"U.S. Army Corps of Engineers, Great Lakes and Ohio River Division","url":"https://www.lrd.usace.army.mil/Public-Notices/Programs/Article/3650069/water-quality/","source_class":"OFFICIAL_ORGANIZATION_DATA","publication_date":"2024-01","accessed_at":"2026-08-03","claims_supported":["The Pittsburgh District water-quality team supports reservoir operations and monitors reservoir inflows, outflows, downstream regulated reaches, and relevant tributaries.","The district operates continuous real-time monitoring at more than 35 operational control points for temperature, dissolved oxygen, conductivity, turbidity, and related parameters.","The team conducts incident-response surveys for spills, fish kills, harmful algal blooms, and other events affecting water-management or natural-resource missions.","USACE water-quality teams are identifiable prospective adopters and operational authorizers for a shadow evaluation."]},{"source_id":"S2","title":"The Effects of Hydropower Releases from Lake Texoma on Downstream Water Quality","publisher":"U.S. Environmental Protection Agency HERO record for Journal of Freshwater Ecology research","url":"https://hero.epa.gov/hero/index.cfm/reference/details/reference_id/7786697","source_class":"AUTHORITATIVE_SECONDARY","publication_date":"1999","accessed_at":"2026-08-03","claims_supported":["Peer-reviewed field sampling before, during, and after hydropower generation found material release-associated changes in downstream temperature, dissolved oxygen, metals, and turbidity.","Observed turbidity rose from about 3 to 25 NTU with downstream metal increases during generation, while dissolved-oxygen patterns varied by generation state and location.","Release responses have travel-time and leading-edge structure, supporting time-conditioned rather than static thresholds."]},{"source_id":"S3","title":"HEC-ResSim Version 4.0, New Water Quality Feature","publisher":"U.S. Army Corps of Engineers Hydrologic Engineering Center","url":"https://www.hec.usace.army.mil/confluence/hecnews/summer-2026/hec-ressim-version-4-0-new-water-quality-feature","source_class":"OFFICIAL_PRODUCT_DOCUMENTATION","publication_date":"2026","accessed_at":"2026-08-03","claims_supported":["HEC-ResSim 4.0 integrates reservoir-operation, river-hydraulic, and water-quality simulation and is already implemented in multiple federally managed reservoir systems.","USBR uses the capability to predict and manage water temperatures in Central Valley Project reservoirs and rivers.","The model predicts releases, diversions, flows, and elevations, then calculates spatial water-quality transport and transformation.","Observed water-quality time series support calibration and validation, and simulated point-source or non-point-source mass injections can test coincident disturbances.","This is close prior art for command-conditioned downstream prediction, but the documentation does not establish a governed residual-only operator-attention channel."]},{"source_id":"S4","title":"HOW TO Add Real-Time Gate Operation Capability to a ResSim Model","publisher":"U.S. Army Corps of Engineers Hydrologic Engineering Center","url":"https://www.hec.usace.army.mil/confluence/ressimguides/how-to-articles/how-to-add-real-time-gate-operation-capability-to-a-ressim-model","source_class":"OFFICIAL_PRODUCT_DOCUMENTATION","publication_date":"undated","accessed_at":"2026-08-03","claims_supported":["HEC-ResSim can represent time-indexed gate settings and calculate discharge from gate opening, pool elevation, and tailwater elevation.","The Operation Support Interface supports editable gate-setting scenarios and real-time operational what-if analysis.","Gate-command ingestion and modeled actuation consequences are technically feasible, although the guide does not verify physical actuator response or implement residual attention routing."]},{"source_id":"S5","title":"Guidelines and Standard Procedures for Continuous Water-Quality Monitors: Station Operation, Record Computation, and Data Reporting","publisher":"U.S. Geological Survey","url":"https://www.usgs.gov/publications/guidelines-and-standard-procedures-continuous-water-quality-monitors-station-operation","source_class":"OFFICIAL_GUIDANCE","publication_date":"2006-04-28","accessed_at":"2026-08-03","claims_supported":["Continuous stations commonly measure temperature, specific conductance, dissolved oxygen, and pH and may also measure turbidity.","Reliable records require field inspection, cleaning, calibration, data evaluation, correction, computation, review, and reporting.","Raw-stream retention and independent sensor-quality procedures are necessary because model residuals alone cannot distinguish environmental change from sensor error."]},{"source_id":"S6","title":"ContDataQC: An R Package and Shiny App for Quality Control of Continuous Water Quality Sensor Data","publisher":"U.S. Environmental Protection Agency","url":"https://assessments.epa.gov/risk/document/%26deid%3D365551","source_class":"PRIMARY_RESEARCH","publication_date":"2025","accessed_at":"2026-08-03","claims_supported":["A free first-party tool already detects anomalies and erroneous values in continuous temperature, dissolved-oxygen, conductivity, turbidity, and related sensor streams.","Continuous sensors capture episodic events missed by sparse visits, but their data-management and quality-control burden can be daunting.","Generic sensor anomaly detection is established adjacent practice but is not conditioned on verified reservoir commands or downstream release-response predictions."]},{"source_id":"S7","title":"Alarm Management for Hydropower Plants","publisher":"Power Engineering","url":"https://www.power-eng.com/renewables/hydropower/alarm-management-for-hydropower-plants/","source_class":"AUTHORITATIVE_SECONDARY","publication_date":"2015-12-02","accessed_at":"2026-08-03","claims_supported":["Hydropower operators encounter alarm floods and nuisance alarms across operational, environmental, equipment, electrical, and safety systems.","A documented startup example caused operators to become desensitized to predictable vibration alarms.","The source supports the general attention problem and risk of simply raising thresholds, but does not quantify environmental-event misses during releases."]},{"source_id":"S8","title":"Guidelines for Collecting Data to Support Riverine Hydrodynamic and Water Quality Simulation Models","publisher":"U.S. Bureau of Reclamation","url":"https://www.usbr.gov/tsc/techreferences/mands/mands-pdfs/River_water_quality_manual.pdf","source_class":"OFFICIAL_GUIDANCE","publication_date":"2010-08","accessed_at":"2026-08-03","claims_supported":["Tailwater modeling requires geometry, tributary and dam flows, release temperature and quality, downstream observations, travel-time information, and meteorological data.","Release data and hourly or finer meteorological data are needed to model tailwater flow, temperature, and dissolved-oxygen dynamics.","Model development requires calibration across hydrologic conditions, project-specific sampling and QA/QC planning, and multidisciplinary cooperation.","The guidance notes that data-collection programs are sometimes abandoned for economic reasons, making cost and data completeness material feasibility risks.","Existing limited data and initial testing can support a bounded first assessment before broader model development."]}],"problem_evidence":{"support":"MODERATE","rationale":"Field research directly shows that hydropower releases can dominate short-timescale downstream temperature, dissolved-oxygen, turbidity, and related patterns, while USACE operates continuous downstream monitoring and responds to coincident incidents. Hydropower alarm-flood evidence supports a general attention constraint. No direct source quantifies how often release-related displays cause delayed or false attribution of external environmental events, so the candidate-specific prevalence and realized harm remain unmeasured.","source_ids":["S1","S2","S7"]},"stakeholder_evidence":{"support":"MODERATE","rationale":"USACE water-quality teams are identifiable adopters with responsibility for reservoir-operational monitoring and incident surveys, and USACE/USBR already deploy integrated release and water-quality modeling. This establishes capability, mission fit, and plausible authorization for a retrospective or non-operational shadow evaluation. No source expresses demand for this exact residual-attention intervention or commits a site, budget, data access, or operator participation.","source_ids":["S1","S3","S4"]},"prior_art":{"proximity":"ADJACENT_PRIOR_ART","closest_analogues":[{"name":"HEC-ResSim 4.0 integrated reservoir and water-quality modeling","similarity":"Predicts releases, downstream hydraulics, temperature, dissolved oxygen, and other constituents; stores observed data for calibration and can simulate superposed constituent injections.","remaining_difference":"The documentation does not make observed-minus-command-predicted residuals the governed routine attention channel, verify commanded versus measured actuation, or require independent raw audits and protected full-signal bypasses.","source_ids":["S3"]},{"name":"HEC-ResSim real-time gate-operation workflow","similarity":"Represents gate settings as time series and computes discharge from gate, pool, and tailwater states for operational scenarios.","remaining_difference":"It is a gate-scenario modeling workflow rather than an actuator-verified downstream environmental residual and operator-triage system.","source_ids":["S4"]},{"name":"Continuous monitor QC and anomaly detection","similarity":"Existing USGS procedures and EPA tooling manage continuous temperature, dissolved-oxygen, conductivity, and turbidity streams and detect sensor anomalies.","remaining_difference":"These practices assess raw-sensor quality or generic departures, not deviations relative to a frozen prediction of a verified reservoir action.","source_ids":["S5","S6"]},{"name":"Hydropower alarm management","similarity":"Recognizes nuisance alarms, predictable startup conditions, desensitization, and the need to preserve operator trust.","remaining_difference":"Alarm rationalization is established, but the cited practice is not a reconstructive water-quality residual architecture with model versioning, coincident-event tests, raw audit, and decompression.","source_ids":["S7"]}],"distinctive_claim_remaining":"Against an unchanged raw-threshold dashboard during controlled releases, a frozen model conditioned on both the issued command and measured actuator response can reduce routine operator-review volume by at least 30% while routing 100% of predefined protected signals, detecting and acknowledging at least 95% of scripted coincident conductivity, turbidity, and dissolved-oxygen departures, reconstructing held-out raw observations within preregistered variable-specific tolerances, and entering full-signal fallback for every injected actuation, checksum, heartbeat, travel-time, or persistent-bias fault.","confidence":"MODERATE"},"implementation_evidence":{"support":"MODERATE","rationale":"The necessary building blocks—time-indexed gate inputs, coupled release and river water-quality models, continuous multivariable sensors, observed-series calibration, sensor QC, and raw-record procedures—are established. Implementation is data- and workflow-intensive: geometry, tributaries, travel time, meteorology, actuator telemetry, calibration across regimes, model-version controls, and independent archives must align. No direct source demonstrates the complete command-to-actuator-to-water-quality residual-attention loop in production, and local cybersecurity, licensing, labor agreements, water-control manuals, and record obligations were not verified.","source_ids":["S3","S4","S5","S6","S8"]},"scores":{"meaningful_impact":{"score":4,"rationale":"Release-associated water-quality changes and incident-response responsibilities are material, and better discrimination could improve attention without weakening raw records; the frequency and severity of current misattribution are unknown.","source_ids":["S1","S2","S7"]},"stakeholder_pull":{"score":3,"rationale":"USACE and USBR have clear mission fit and use closely related monitoring and modeling, but no adopter has requested or funded the exact intervention.","source_ids":["S1","S3"]},"incremental_advantage":{"score":3,"rationale":"Actuator verification, reconstructive residual routing, independent raw audit, and governed fallback add a coherent layer beyond forecast displays, QC, and alarm management, but comparative benefit is untested.","source_ids":["S3","S4","S5","S6","S7"]},"distinctiveness_plausibility":{"score":3,"rationale":"No exact implementation was found in the bounded search, but each major technical element has close established analogues; only the integrated governance and comparator claim appears distinctive.","source_ids":["S3","S4","S6","S7"]},"technical_implementability":{"score":4,"rationale":"Official tools and guidance establish the core modeling, telemetry, validation, and QC components. Site-specific data completeness and model identifiability remain significant risks.","source_ids":["S3","S4","S5","S8"]},"adoption_authority_feasibility":{"score":4,"rationale":"A reservoir water-quality organization can plausibly authorize retrospective and shadow analysis while existing operators retain all control authority. Site-specific manuals, licenses, and data-security approvals remain unchecked.","source_ids":["S1","S3"]},"evidence_readiness":{"score":3,"rationale":"Historical releases and continuous records make a bounded replay plausible, but synchronized command, measured-actuator, tributary, meteorological, and downstream records may be proprietary, incomplete, or insufficiently aligned.","source_ids":["S1","S5","S8"]},"safety_net_benefit":{"score":4,"rationale":"Keeping the raw dashboard authoritative and testing explicit bypass and fallback paths could preserve context while evaluating attention savings. Safety benefit remains prospective until fault-injection and shadow results exist.","source_ids":["S5","S7"]},"scalability":{"score":3,"rationale":"The architecture and official modeling stack are reusable, but every dam requires local geometry, travel-time, operating-regime, sensor, ecological-limit, and governance calibration.","source_ids":["S3","S8"]}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"50K_TO_250K","scope":"Preregister and execute one historical-release replay using existing command, actuator, upstream, meteorological, and downstream records; freeze one model; create scripted coincident-event and fault tests; compare with the unchanged raw-threshold baseline; and conduct operator review.","confidence":"LOW","assumptions":["Existing synchronized data and a usable hydraulic or water-quality model are available.","Scope is limited to one release, two to four downstream stations, and five variables.","Approximately 8–20 person-weeks cover data engineering, modeling, test construction, environmental review, and operator evaluation.","No new field instrumentation, permit amendment, or production control-system integration is included."],"source_ids":["S3","S6","S8"]},"initial_deployment_startup":{"band_2026_usd":"250K_TO_1M","scope":"Prepare and observe one scheduled release in non-operational shadow mode, including read-only historian interfaces, versioned predictions, residual display, independent archive checks, fault injection, procedures, training, and evaluation.","confidence":"LOW","assumptions":["Existing certified sensors, telemetry, and raw dashboards remain authoritative.","Work is read-only and does not modify gate logic, alarms, emergency procedures, or mandatory records.","Cybersecurity review and limited interface development are required.","Field sampling validates transition and high-uncertainty windows."],"source_ids":["S1","S4","S5","S8"]},"operational_launch":{"band_2026_usd":"1M_TO_5M","scope":"Production-harden a multi-station residual-attention service with redundant ingestion, actuator verification, model and station configuration management, audit storage, fallback, observability, incident routing, validation across seasons and release regimes, cybersecurity, SOPs, and training.","confidence":"LOW","assumptions":["One reservoir system with several downstream stations is included.","Existing SCADA and monitoring infrastructure can expose read-only authenticated data.","At least one full seasonal calibration and validation cycle is required.","Major new civil works, sensor-network replacement, and automatic gate control are excluded."],"source_ids":["S1","S3","S4","S5","S8"]},"annual_recurring":{"band_2026_usd":"250K_TO_1M","scope":"Operate data pipelines and audit archives; clean and calibrate sensors; review residuals and incidents; maintain models and thresholds; test fallback; conduct field validation; and complete annual governance and cybersecurity review.","confidence":"LOW","assumptions":["Recurring work requires fractional coverage from operations, environmental science, modeling, data engineering, field technicians, and compliance staff.","Sensor replacement and telemetry repairs are routine rather than system-wide.","Model updates remain offline, versioned, and separately approved.","The range is a resource-equivalent estimate, not a vendor quotation or site budget."],"source_ids":["S5","S6","S8"]}},"verified_pipeline_gates":{"externally_supported_problem":{"status":"YES","reason":"Release-driven downstream water-quality variation and hydropower nuisance-alarm problems are externally visible, though the candidate-specific misattribution rate is unknown.","source_ids":["S1","S2","S7"]},"externally_credible_adopter_or_authorizer":{"status":"YES","reason":"USACE water-quality teams monitor reservoir operational control points, support water-management operations, and investigate environmental incidents; USACE and USBR already use related modeling systems.","source_ids":["S1","S3"]},"distinct_testable_incremental_claim":{"status":"YES","reason":"The claim compares an actuator-verified residual-attention layer against the unchanged raw-threshold dashboard using notification burden, detection, protected routing, reconstruction, attribution, and fallback metrics.","source_ids":["S3","S4","S7"]},"bounded_next_evidence_step":{"status":"YES","reason":"One historical replay followed conditionally by one shadow release is bounded by release, station set, variables, frozen model, untouched holdout, comparator, scripted faults, and explicit falsifiers.","source_ids":["S3","S6","S8"]},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"The proposed evidence step is retrospective or read-only shadow evaluation; raw monitoring and existing procedures remain authoritative, no gate action is automated, and protected signals are tested rather than suppressed. Local approvals still must be obtained before the shadow release.","source_ids":["S1","S5"]},"credible_cost_scope_and_range":{"status":"UNCERTAIN","reason":"The four scopes are explicit and broad bands reflect the substantial data, calibration, field, QA/QC, and multidisciplinary work described by official guidance, but no site inventory, labor plan, vendor quote, or local cybersecurity estimate was available.","source_ids":["S5","S8"]}},"next_evidence_step":"With one reservoir partner, preregister a retrospective replay of one controlled release and reserve a contiguous interval as an untouched holdout. Freeze the command-conditioned model, actuator data, station set, horizons, uncertainty assumptions, protected classes, reconstruction tolerances, and fallback rules. Superpose blinded conductivity, turbidity, and dissolved-oxygen departures and inject command-actuator mismatch, sensor dropout, checksum conflict, travel-time shift, and persistent-bias faults. Compare against the unchanged raw dashboard with fixed thresholds using notification count and review minutes, event detection and acknowledgement, false attribution, protected-signal routing, reconstruction error, audit disagreement, fallback correctness, and total labor. Falsify the claim if notification burden falls by less than 30%, any protected signal is suppressed, scripted-event detection is below 95%, any declared fault fails to trigger fallback, held-out reconstruction exceeds tolerance, or total burden is not lower. Only after all replay criteria pass should one scheduled release be observed in read-only shadow mode.","blocking_evidence":["No site has committed synchronized command, measured-actuator, tributary, meteorological, raw sensor, alarm, and acknowledgement records.","The prevalence and consequences of current release-period alarm burden or external-event misattribution have not been measured.","No held-out evidence shows that command-conditioned residuals remain identifiable across travel-time variation, seasons, channel change, stratification, or unmodeled inflows.","No empirical result establishes coincident-event recall, protected-signal routing, reconstruction fidelity, fallback reliability, or operator time savings.","Local water-control manuals, hydropower license conditions, water-quality certificates, records rules, cybersecurity controls, and labor procedures were not reviewed.","Cost bands lack a site inventory, staffing plan, integration assessment, and quotations."],"research_disposition":"PARTNERED_RESEARCH_PROGRAM","world_novelty_boundary":"The search establishes adjacent official reservoir-operation and water-quality prediction, real-time gate modeling, continuous-sensor QC, downstream release-impact research, and hydropower alarm management. It does not establish or refute worldwide novelty of the integrated actuator-verified residual-attention architecture. Patentability, freedom to operate, market size, world novelty, and realized impact remain unmeasured.","arm":"COMPLETE_PROPOSAL_PORTFOLIO","candidate_version":0,"controller_recommendation":{"action":"STOP_EMPIRICAL_RESEARCH_NEEDED","repairable":false,"material_progress_observed":false,"progress_targets":["Secure a named reservoir partner and written authorization for historical-data access and a read-only shadow release.","Verify completeness and synchronization of command, measured-actuator, upstream, tributary, meteorological, downstream raw-sensor, alarm, and acknowledgement records.","Preregister the frozen model, untouched holdout, raw-dashboard comparator, protected classes, tolerances, burden measures, and falsification thresholds.","Demonstrate blinded coincident-event detection, actuator-discrepancy visibility, missingness handling, reconstruction fidelity, and deterministic fallback under scripted faults.","Measure operator notification volume, review time, acknowledgement time, false attribution, audit disagreement, and total labor against the raw baseline.","Produce a site-specific authority review, cybersecurity integration assessment, staffing plan, instrumentation inventory, and cost quotations."],"reason":"Bounded web research supports the problem, a credible adopter class, technical building blocks, and adjacent prior art, but it cannot determine the proposal's incremental operating benefit or safety performance. The decisive evidence requires proprietary synchronized records, operator participation, fault-injection replay, and a live read-only shadow observation. Under the required controller rule, that fieldwork and live-testing dependency requires STOP_EMPIRICAL_RESEARCH_NEEDED with repairable set to false."},"proposal_index":4}