{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp03_full320_20260801","cell_id":"invariant_mode_decomposition_design__marine_science","trajectory_id":"R","attempt_index":0,"archetype_slug":"invariant_mode_decomposition_design","domain_slug":"marine_science","decision":"CANDIDATE","problem_id":"coastal_hypoxia_hidden_coupled_precursor","causal_lever_id":"unstable_biogeochemical_mode_targeting","proposal":{"problem":"Coastal hypoxia management often reacts only after dissolved oxygen crosses a threshold, while coupled changes in stratification, temperature, nutrients, plankton, respiration, and circulation may already be moving toward oxygen depletion. The independently recognizable problem is late detection and poorly targeted mitigation of seasonal hypoxia because station-level variables and spatial averages conceal the combinations that precede it. Whether a sufficiently stable precursor mode exists is a HYPOTHESIS.","actors_substrate":["coastal water body and its coupled physical-biogeochemical state","monitoring agencies and observing systems","watershed and coastal managers","marine organisms, fisheries, aquaculture, and coastal communities"],"observable_state":"Repeated multivariate observations show apparently tolerable mean oxygen while a particular combination of stronger stratification, warmer bottom water, organic-matter loading, respiration, and weak ventilation grows across successive intervals; forecast residuals and modal drift are observable.","consequence":"Managers receive insufficient lead time, misattribute risk to single variables, or impose broad controls that miss the combinations and locations driving oxygen loss.","affected_objective":"Increase validated warning lead time and improve targeting of reversible monitoring or nutrient-management responses without suppressing residual uncertainty.","structural_mapping":[{"archetype_element":"coupled transformation","domain_realization":"A locally fitted transition operator maps the coastal physical-biogeochemical state between monitoring intervals.","claim_kind":"HYPOTHESIS"},{"archetype_element":"invariant directions","domain_realization":"Approximately persistent combinations of oxygen, stratification, temperature, nutrients, biomass, respiration, and exchange serve as candidate modes.","claim_kind":"HYPOTHESIS"},{"archetype_element":"scalar modal response","domain_realization":"Estimated gains distinguish decaying, persistent, oscillatory, and growing combinations within a declared seasonal regime.","claim_kind":"INFERENCE"},{"archetype_element":"hidden unstable direction","domain_realization":"A growing oxygen-depletion precursor can be weak in any single coordinate yet consequential in combination.","claim_kind":"HYPOTHESIS"},{"archetype_element":"intervention in modal coordinates","domain_realization":"Sensitivity analysis maps feasible actions and added observations onto the modes they can change or resolve.","claim_kind":"INFERENCE"},{"archetype_element":"residual and drift governance","domain_realization":"Out-of-sample reconstruction error, basis rotation, and spectral-gap loss determine when warnings are withheld or the model is re-estimated.","claim_kind":"INFERENCE"}],"component_map":[{"component":"Transformation Scope","status":"adapted","domain_realization":"One embayment or shelf sector, a fixed sampling interval, and a season-specific transition model."},{"component":"State-Vector Definition","status":"adapted","domain_realization":"Quality-controlled oxygen, temperature, salinity/stratification, nutrients, chlorophyll or biomass, respiration proxies, currents, and exchange variables."},{"component":"Invariant Mode Basis","status":"adapted","domain_realization":"Eigenmodes of the fitted local transition operator, traceable to measured variables."},{"component":"Modal Gain Spectrum","status":"direct","domain_realization":"Estimated eigenvalues with uncertainty across resamples."},{"component":"Dominant Mode Selection Rule","status":"adapted","domain_realization":"Retain modes only when out-of-sample fidelity and hypoxia-relevant sensitivity exceed preregistered thresholds."},{"component":"Stable/Unstable Mode Partition","status":"direct","domain_realization":"Classify modes by gain relative to the sampling-interval stability boundary."},{"component":"Modal Intervention Map","status":"adapted","domain_realization":"Map added sampling, nutrient-load scenarios, and feasible aeration or mixing options to predicted modal changes."},{"component":"Reconstruction Residual Check","status":"direct","domain_realization":"Measure held-out oxygen and state reconstruction error and inspect structured residuals."},{"component":"Mode Drift Monitor","status":"direct","domain_realization":"Track mode rotation, reordering, and gain changes as weather and circulation regimes shift."},{"component":"Interpretation Scope Contract","status":"direct","domain_realization":"Label modes as local predictive constructs, not automatically causal ecological entities."},{"component":"Mode-Coupling Register","status":"direct","domain_realization":"Record near-degeneracy, non-normality, and sensitivity cross-effects."},{"component":"Local Linearization Window","status":"adapted","domain_realization":"Specify season, forcing range, spatial footprint, and disturbance magnitude represented by the fit."},{"component":"Spectral Gap Threshold","status":"direct","domain_realization":"Require separation between retained and discarded modes sufficient for stable resampled identification."}],"mechanism_dispositions":[{"slug":"eigendecomposition_workflow","disposition":"selected_load_bearing","contribution_type":"CORE_CAUSAL","adaptation_or_rejection":"Decompose the explicitly fitted local transition matrix; report conditioning and complex modes.","counterfactual_removal":"No invariant gains or stability partition would exist, eliminating the proposed lever."},{"slug":"modal_sensitivity_sweep","disposition":"selected_load_bearing","contribution_type":"CORE_CAUSAL","adaptation_or_rejection":"Perturb candidate actions and modal coordinates locally to rank hypoxia-relevant leverage and cross-effects.","counterfactual_removal":"Modes could predict risk but could not support targeted action selection."},{"slug":"modal_stability_analysis","disposition":"selected_load_bearing","contribution_type":"CORE_CAUSAL","adaptation_or_rejection":"Classify growth or decay only within the declared seasonal linearization window.","counterfactual_removal":"The model could not distinguish an amplifying precursor from harmless variance."},{"slug":"mode_shape_testing","disposition":"incompatible","contribution_type":"NONE","adaptation_or_rejection":"Controlled excitation of an open coastal ecosystem is infeasible and unsafe; routine observations cannot guarantee excitation of every mode.","counterfactual_removal":"No change because operator fitting and held-out observations provide the bounded identification route."},{"slug":"network_spectral_centrality_analysis","disposition":"considered_rejected","contribution_type":"NONE","adaptation_or_rejection":"Node importance is not the target; the state transition, not a connectivity ranking, defines the problem.","counterfactual_removal":"No material change to detection or intervention logic."},{"slug":"power_iteration_probe","disposition":"considered_rejected","contribution_type":"NONE","adaptation_or_rejection":"Dominant-only recovery could miss a lower-gain but oxygen-critical mode and is unnecessary for the bounded matrix.","counterfactual_removal":"No change; full decomposition remains available."},{"slug":"principal_component_analysis","disposition":"considered_rejected","contribution_type":"NONE","adaptation_or_rejection":"Variance-ranked components do not establish temporal growth or hypoxia relevance.","counterfactual_removal":"No change to the transition-based causal chain."},{"slug":"reduced_order_model","disposition":"selected_supporting","contribution_type":"OPERATIONAL","adaptation_or_rejection":"Run retained modes as a local scenario surrogate, never outside its validation window.","counterfactual_removal":"The core diagnosis survives, but rapid action-scenario comparison becomes impractical."},{"slug":"residual_reconstruction_test","disposition":"selected_load_bearing","contribution_type":"SAFETY_GUARDRAIL","adaptation_or_rejection":"Use spatially and temporally held-out reconstruction, including structured residual inspection.","counterfactual_removal":"Discarded dynamics could remain invisible, making compression unsafe."},{"slug":"singular_value_decomposition","disposition":"selected_supporting","contribution_type":"OPERATIONAL","adaptation_or_rejection":"Regularize transition estimation and diagnose rank or non-normal amplification; do not call singular vectors invariant temporal modes.","counterfactual_removal":"Estimation becomes less robust, though eigendecomposition remains conceptually possible."},{"slug":"spectral_decomposition_report","disposition":"selected_supporting","contribution_type":"SAFETY_GUARDRAIL","adaptation_or_rejection":"Publish variable loadings, uncertainty, couplings, residuals, and prohibited interpretations.","counterfactual_removal":"Operational users could mistake local association for causal ecological structure."},{"slug":"spectral_gap_monitor","disposition":"selected_load_bearing","contribution_type":"SAFETY_GUARDRAIL","adaptation_or_rejection":"Monitor retained/discarded separation and resampled mode rotation during deployment.","counterfactual_removal":"The system could continue issuing warnings after its reduced basis loses identifiability."}],"causal_chain":["Coupled coastal variables evolve through an approximately stable local transition during a bounded season.","Decomposition exposes candidate combinations whose gains indicate persistence or growth.","Held-out reconstruction and resampling reject modes that are unstable artifacts or omit consequential structure.","Hypoxia-relevant sensitivity separates actionable modes from merely large modes.","Validated growth in an oxygen-depletion mode triggers earlier intensified observation and human review of reversible responses.","Residual growth, gap closure, or basis drift suspends the modal warning and forces re-estimation."],"baseline":"Ordinary practice is threshold-based surveillance of dissolved oxygen, chlorophyll, temperature, or station averages, followed by response after a coordinate crosses an alarm level.","nearest_rival":"A multivariate state-space or machine-learning hypoxia forecast can provide lead time without invariant-mode interpretation; it is superior if it predicts reliably while modal gains are unstable or add no actionable discrimination.","authority_safety":{"affected_parties":["marine organisms and habitats","fishers and aquaculture operators","coastal and watershed communities","dischargers and nutrient-source operators","monitoring and resource-management staff"],"decision_authority":"The responsible coastal or environmental agency retains warning and management authority; the model supplies advisory evidence only.","authorized_first_step":"Run a preregistered shadow pilot on one region using archived seasons plus one prospective season; compare warning lead time, false alarms, held-out residuals, and mode stability against threshold surveillance without changing permits or access.","excluded_actions":["automatic discharge restrictions","automatic fishery or beach closures","ecosystem-scale experimental forcing","claims that a statistical mode is a causal process without independent evidence","deployment outside the validated season or spatial footprint"],"halt_rollback":"Suspend alerts and revert to ordinary threshold surveillance if held-out oxygen error exceeds tolerance, residuals become structured, the spectral gap falls below threshold, mode identity becomes unstable across resamples, or false alarms exceed the preregistered limit."}},"negative_tests":{"strongest_counterevidence":"Historical and prospective data may show that oxygen thresholds or a conventional forecast equal or outperform the modal system in lead time and calibration, while estimated modes rotate markedly across seasons.","analogy_break":"Open coastal waters are advective, nonlinear, externally forced, and incompletely observed; an eigenmode of a local fitted operator may be a sampling artifact rather than a persistent ocean process.","failure_condition":"The approach fails if no repeatable local transition window or identifiable gap exists, missing variables dominate oxygen dynamics, or feasible actions cannot measurably influence the selected modes.","problem_falsifier":"The diagnosed problem is falsified if harmful oxygen declines in the target region are already detected with adequate lead time by single-variable thresholds and no coupled precursor adds reproducible predictive information.","intervention_falsifier":"The intervention is falsified if, under held-out and prospective evaluation, selected modal growth neither improves calibrated warning lead time nor distinguishes effective observation or management choices from the baseline and nearest rival.","risks":["false alarms causing unnecessary economic or operational response","missed hypoxia from unobserved or nonlinear dynamics","confusing predictive modes with causal mechanisms","spatial inequity from monitoring gaps","model drift during extreme weather or circulation change","management gaming of model inputs or thresholds"]},"null_rationale":null,"classification":{"candidate_kind":"MECHANISM_COMPOSITION","prior_art_status":"UNSEARCHED","evidence_maturity":"HYPOTHESIS"},"revision_change_log":{"revision_kind":"ORIGINAL","prior_problem_id":null,"prior_causal_lever_id":null,"problem_changed":false,"causal_lever_changed":false,"conceptual_changes":[],"operational_changes":[],"repairs_addressed":[]},"confidence":0.79,"generator_notes":"Closed-book structural inference from the supplied packet. The existence, stability, predictive value, and controllability of a coastal hypoxia precursor mode are hypotheses requiring the bounded shadow pilot."}