{"schema_version":1,"assessment_id":"eoa_inverse_innovation_exp03_opportunity320_20260801","source_experiment_id":"eoa_inverse_innovation_exp03_full320_20260801","cell_id":"invariant_mode_decomposition_design__earth_sciences","archetype_slug":"invariant_mode_decomposition_design","domain_slug":"earth_sciences","title":"Growth-aware modal alerting for instrumented landslide slopes","opportunity_summary":"Test whether locally estimated, sensor-traceable growing modes across hydrologic and deformation measurements can increase landslide-warning lead time at a fixed false-alert burden relative to independent thresholds and generic multivariate anomaly detection. The opportunity is limited to instrumented slopes and declared regimes, and both the coupled-precursor problem and operational advantage remain unverified hypotheses.","adopter_authorizer":"Geotechnical analysts or geological-survey hazard staff would adopt the advisory analysis; legally designated geological-hazard and emergency-management authorities would authorize its use and retain all inspection, closure, warning, and evacuation decisions.","scores":{"meaningful_impact":{"score":4,"rationale":"If the hypothesized coupled precursor exists, additional warning lead time could support investigation, access closure, or evacuation for people and infrastructure exposed to slope failure. Impact is conditional because the packet provides no evidence that such precursors are prevalent or reproducible."},"stakeholder_pull":{"score":2,"rationale":"The candidate identifies analysts, hazard authorities, emergency managers, residents, workers, and road users, but contains no adopter interviews, requests, commitments, procurement signals, or evidence that univariate-alert limitations are a recognized priority at candidate sites."},"incremental_advantage":{"score":3,"rationale":"Growth-aware modal scoring makes a testable incremental claim against independent thresholds and a PCA or generic anomaly detector: distinguish directions that grow under repeated state evolution. Whether this yields usable lead time or fewer false alerts is entirely untested."},"distinctiveness_plausibility":{"score":3,"rationale":"The combination of estimated modal growth, sensor traceability, residual checks, and explicit drift and conditioning gates is conceptually differentiated from the named rivals. Prior-art status is UNS​EARCHED, so distinctiveness beyond those packet-defined comparators is uncertain."},"technical_implementability":{"score":3,"rationale":"A retrospective implementation using synchronized standardized sensor vectors and local transition estimation is technically bounded, but adequate stationarity, conditioning, spectral separation, missing-data handling, and stable mode identity may not be achievable on heterogeneous nonlinear slopes."},"adoption_authority_feasibility":{"score":4,"rationale":"The packet clearly assigns decisions to legally designated authorities, restricts the model to advisory evidence, and provides a shadow-only route through an existing alert protocol. Actual authority participation, data agreements, workflow fit, and approval requirements are not established."},"evidence_readiness":{"score":3,"rationale":"The candidate supplies baselines, a nearest rival, temporal replay and shadow designs, metrics, falsifiers, and halt rules. Readiness is limited by unknown access to synchronized failure-relevant histories, representative seasons, sufficient hazardous events, and reliable site metadata."},"safety_net_benefit":{"score":4,"rationale":"Potential earlier corroborated detection directly supports protective decision-making, while analyst review, residual visibility, regime limits, suspension rules, and exclusion of automated public actions reduce misuse. False reassurance, false alarms, coverage inequality, and review latency remain material risks."},"scalability":{"score":2,"rationale":"The method is explicitly site- and regime-bounded and depends on existing multisensor observatories, local stationarity, site-specific fitting, sensor continuity, and repeated validation. Natural-slope heterogeneity and changing sensors make low-touch replication unlikely without further evidence."}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"50K_TO_250K","scope":"One site-bounded retrospective replay using an existing synchronized multisensor history, including data preparation, baseline and nearest-rival implementations, temporal holdouts, conditioning and drift analysis, and a documented go/no-go review.","confidence":"MODERATE","assumptions":["Usable historical rain, pore-pressure, groundwater, displacement, tilt, and event data already exist.","No new field instrumentation or hazardous slope intervention is required.","One research and geotechnical partner can obtain and interpret the data.","The evaluation remains retrospective and does not inform public decisions."]},"initial_deployment_startup":{"band_2026_usd":"50K_TO_250K","scope":"Prepare prospective shadow monitoring at one already instrumented slope, including data-pipeline integration, model monitoring, access controls, logging, analyst displays, operating limits, training, and rollback procedures.","confidence":"LOW","assumptions":["Existing sensors and telemetry are sufficiently reliable and accessible.","The modal route remains invisible outside authorized analysts.","No major procurement, telemetry replacement, or bespoke safety certification is required.","Retrospective results satisfy predeclared performance and stability thresholds."]},"operational_launch":{"band_2026_usd":"250K_TO_1M","scope":"After successful replay and shadow evidence, qualify a one-site advisory route for routine hazard operations, including independent validation, workflow and governance approval, incident exercises, documentation, security, compliance, and integration with corroboration and review.","confidence":"LOW","assumptions":["Launch does not automate warnings, closures, or evacuations.","Relevant geological-hazard and emergency-management authorities participate.","Existing monitoring infrastructure remains in place.","Site-specific validation is adequate for advisory use within one declared regime."]},"annual_recurring":{"band_2026_usd":"50K_TO_250K","scope":"Operate and govern one validated site, including data-quality monitoring, analyst review, drift and residual surveillance, sensor-change assessment, periodic replay, controlled refitting, audit logs, and performance reporting.","confidence":"LOW","assumptions":["Existing observatory operations cover sensor hardware and routine field maintenance.","Material sensor or regime changes trigger suspension rather than automatic adaptation.","Human review is available within the existing hazard program.","Expansion to additional slopes is excluded."]}},"research_burden":"HIGH","earliest_credible_horizon":"12_TO_36_MONTHS","pipeline_gates":{"recognizable_externally_supportable_problem":{"status":"UNCERTAIN","reason":"The packet gives a coherent domain problem and falsifier, but explicitly labels the coupled hidden-mode failure a hypothesis and provides no external evidence that it occurs with decision-relevant frequency at representative slopes."},"identifiable_adopter_or_authorizer":{"status":"YES","reason":"Geotechnical analysts and geological-survey hazard staff are identifiable users, while legally designated geological-hazard and emergency-management authorities are explicitly assigned decision authority."},"distinct_testable_incremental_claim":{"status":"YES","reason":"The candidate claims greater warning lead time at a fixed false-alert burden from growth-aware modal scores, with direct comparisons to independent thresholds and a PCA or generic multivariate anomaly detector."},"bounded_next_evidence_step":{"status":"YES","reason":"A retrospective, single-site, temporal holdout replay is bounded and safe and can compare the proposal with both named comparators using predeclared performance, reconstruction, conditioning, separation, and drift criteria."},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"The first step is retrospective and any subsequent shadow alerts are analyst-only; automated protective actions and hazardous interventions are excluded, authorities retain decisions, and explicit halt and rollback conditions are supplied."},"implementation_cost_scope_and_range":{"status":"YES","reason":"The packet bounds initial implementation to an existing instrumented slope, local model fitting, retrospective evaluation, and analyst-only shadow integration, allowing broad resource bands despite uncertainty about data quality and local compliance effort."}},"blocking_evidence":["No evidence establishes that hazardous accelerations are preceded by reproducible coupled multivariate patterns across relevant seasons or locations.","No held-out result shows improvement in warning lead time at a fixed false-alert burden over the strongest single-variable or rainfall-rule baseline.","No comparison shows that growth-aware modal scoring improves on PCA or a generic multivariate anomaly detector.","The availability, synchronization, event labeling, missingness, and regime coverage of suitable historical site data are unknown.","Modal conditioning, spectral separation, reconstruction quality, and identity stability under storms, maintenance, sensor drift, and seasonal change are untested.","No adopter or authority evidence establishes workflow demand, acceptable false-alert burden, review capacity, or approval requirements.","Prior art is unsearched, so distinctiveness from existing landslide state-space, modal, and multivariate early-warning methods is unresolved."],"next_evidence_step":"With one observatory and its authorized data holder, run a retrospective replay on a predeclared historical period using forward-in-time holdouts. Compare the proposed growth-aware modal route against the existing independent thresholds, rate and rainfall rules, the strongest single observable, and a PCA or generic multivariate anomaly detector. Predeclare lead time at a fixed false-alert burden, event and benign-period coverage, reconstruction residual, conditioning, spectral-gap, missingness, and mode-stability thresholds. Stop if the single-variable baseline matches performance, the modal method adds no held-out advantage, or stability and availability gates fail; do not expose outputs to operational decision-makers.","research_questions":["Do hazardous accelerations exhibit coupled precursors that recur across forward-held-out seasons, events, and monitored slope locations?","Does growth-aware scoring improve lead time at the same false-alert burden over the strongest single-variable, rainfall-rule, and rate-of-change baselines?","Does it add decision-relevant performance beyond PCA or a generic multivariate anomaly detector?","How sensitive are modes and alerts to non-normality, near-degeneracy, missing observations, storm forcing, maintenance, and sensor drift?","How often do conditioning, spectral-gap, reconstruction, drift, and regime gates make the method unavailable when review is needed?","Are residuals and sensor loadings sufficiently interpretable to reveal omitted behavior and avoid false mechanistic claims?","Which sites or slope areas receive weaker protection because of unequal sensor coverage?","What false-alert burden, corroboration latency, evidence display, and suspension behavior will analysts and authorities accept?","Does existing prior art already contain the claimed growth-aware modal lever or the same governance bundle?"],"recommendation":"PARTNERED_RESEARCH","uncertainty_constraints":["Closed-book assessment: no external evidence was available for problem prevalence, stakeholder demand, prior art, realized performance, market size, or exact cost.","The candidate is a HYPOTHESIS at instrumented slopes, not evidence of operational effectiveness.","All performance claims are site-, sensor-, season-, and regime-dependent until representative validation exists.","Cost bands assume existing instrumentation and usable data; new sensors, telemetry remediation, extensive compliance work, or multisite evaluation could move costs upward.","The proposed modes are statistical local approximations and must not be treated as verified geologic mechanisms.","The 12-to-36-month horizon assumes timely partner access to historical data and enough prospective shadow duration to encounter representative conditions."],"closed_book_prior_art_boundary":"The sealed packet marks prior art as UNSEARCHED and supports neither novelty nor lack of novelty. The assessment distinguishes the proposal only from its stated baseline and nearest rival; external research must compare it with existing landslide state-space, modal, multivariate-precursor, anomaly-detection, and operational early-warning approaches before any distinctiveness claim."}