Console catalog · Coastal response plan
Use the checklist to connect a diagnostic question to the evidence it needs and a practical next step. Open Response plan on a console, then choose the public or operator view. Coastal has a dedicated response-plan tab.
Descriptive decision support. These are adaptations of research questions. A numerical proxy does not establish the original CRT estimator, a causal network effect or a validated operating threshold. Unknown means the evidence cannot assess the dimension; it does not mean safe.
Public steps emphasize official information, preparedness and clear escalation contacts. Operator steps emphasize evidence verification, handoffs, documented criteria and review within the organization's own systems. The checklist does not authorize dispatch, electrical switching, evacuation, clinical treatment or trading.
The Do next preview names the first action for your current selection. Choose View actions to see the ordered steps and switch between public and operator responsibilities. Every card gives a place or source record, why it was prioritized, when to review it, the next step, an escalation trigger and recipient, and a recheck condition.
For rivers, current source categories and still-valid point forecasts guide verification. For the grid, comparisons use the same report hour. NHS long-wait counts are compared within the same reporting week; annual health service gaps require matching years and populations. Older reports prioritize planning reviews. Reference tools prioritize source checks and qualified review without reading your personal inputs.
Current official warning · check now means read the actual valid notice and follow its instructions. Next check is evidence verification or preparation; Planning review uses dated context; Follow-up maintains source and scope awareness. Timing is an interpretive planning prompt unless attributed to an official notice. It does not create an operational deadline. Expired evidence removes its dependent action, and changed evidence or priorities reset local review notes.
Open Dimension-by-dimension follow-up for the full ten-question chain. Unknown inputs remain visible even when other evidence supports a specific action. Public-data priorities never replace current official instructions or an operator’s approved systems.
Hover over a parameter name or focus its ? button for a short explanation. On a phone, tap ?; press Escape or tap elsewhere to close. Each console also explains the actual measure used in that view.
| Dimension | Question to investigate | What a proxy does not establish |
|---|---|---|
| 1. Population Balance | Do the comparison groups carry comparable exposure, population or service weights? | Equal counts do not imply equal populations or resources. |
| 2. Wasserstein Distance | How do comparable empirical distributions differ, in stated units? | A distribution gap does not identify its cause or an intervention benefit. |
| 3. Density Symmetry | Are the stated geographic or statistical densities comparable? | Point counts alone do not measure spatial density balance. |
| 4. Euler Characteristic | Does the specified structure split into distinct components? | A graph component count is an adaptation, not a full topological estimator. |
| 5. Persistence Gap | Do spatial clusters or gaps persist as the connection scale changes? Does this console instead use a declared threshold-sensitivity band? | A chosen band is not a confidence interval or a safety clearance. |
| 6. Spatial Mixing | Does the conclusion change when the declared spatial or analytical scale changes? | Threshold sensitivity is not automatically a fitted spatial-mixing statistic. |
| 7. Variational Free Energy | How well does a stated model describe dependence, allowing for model complexity? | The CRT reference uses a BIC fit-versus-complexity proxy under this inherited name. A link-count adaptation is not a fitted variational-free-energy bound. |
| 8. Contamination AUC | How does exposure accumulate over a declared time or scenario ladder? | An exposure-screen AUC is not trial contamination, harm or loss probability. |
| 9. Frontier Risk | How sensitive is a defined boundary to the chosen buffer depth? | Graph hops do not establish travel, onset or restoration time. |
| 10. Mobility Network Leakage | Which verified flows or labeled model links cross the comparison boundary? | Estimated associations and proximity links do not establish actual movement or causal transmission. |
Choose View impacts and sources to see the quantities the selected public records can support. Each item gives its scope, source period, method and evidence links. Observed is a current reported quantity; forecast is a still-applicable official forecast; context includes dated reports and reference quantities; scenario is a planning exercise. A source model estimate also carries a modeled badge.
These distinctions matter: a river stage statement is conditional on a threshold, a PAGER exposure estimate is not a confirmed loss, grid demand is not unavailable supply, and annual health statistics are not predictions for an individual. Population-based counts require the right denominator and the same year. Missing, stale, undated or unmatched inputs produce Not assessable, with the input gap shown.
The ten-parameter checklist uses Review for a condition needing investigation and Context · condition not met when its declared test is not met. Neither label certifies safety. Source records and calculations change with the console selection; local review notes reset when that evidence changes.
Measured identifies public observations or reported quantities, with dates and sources. Modeled identifies calculations or estimates and their assumptions. Interpretive identifies review criteria and actions. Demo identifies synthetic practice values. An old report stays historical after it is downloaded today.
A missing graph remains missing. A console may have valid measurements for one dimension and insufficient data for another. Unsupported dimensions retain their required-input list. Operator “review needed” flags are self-reported notes; they do not change measured evidence or convert unknown to safe.
Coastal opens on the real public-station map, and its response checklist defaults to the City-owned public terrain dataset. Street tiles load without a visitor API key; a bundled generalized Natural Earth coastline remains available offline. Map colors show County planning-zone classifications, not current inundation. The Synthetic exercise tab retains the fictional network and its distinct labels. The public-station tab contains a dated official GIS register and exact County parcel-zone intersections. A station with no intersecting evacuation parcel remains unknown; only an explicit source classification is shown as non-evacuation. Use the County's official address lookup and current emergency instructions.
USGS terrain estimates describe bare-earth elevation at a published point, with acquisition vintage and NAVD88 provenance. They do not survey a pump, electrical panel, flood barrier or pipe invert. Comparing terrain with a water plane is a screening exercise; it does not establish hydraulic inundation, failure or an evacuation order.
Projected impacts open in automatic public-data mode. The remaining NOAA forecast or a selected scenario level screens public terrain. Distinct official basin IDs select a union of Census blocks; each block population is counted once. These dated geographic estimates describe basin footprints, not customers, flooded residents or actual service loss. Published average annual treatment flows remain whole-facility reference values. Their equivalent volume over a real forecast-sample span is not an outage or spill forecast.
Sources, measurement years, formulas and coverage are shown. Public context refreshes automatically from a weekly validated publication, retaining last-good data if a check fails. Optional local scenarios expose population, duration, generation, repair and funding assumptions for quantities public data cannot establish. The separate synthetic network supports the calibrated ten-metric exercise; its fictional station IDs and links are not mapped onto real stations.
The dashboard update guide explains each console’s public inputs, update behavior and remaining local or modeled quantities. River plans use source-provided stage impact statements; earthquake plans use official PAGER population exposure; NHS and macro reports update through validated public packages. Country health statistics retain their individual observation years. An automatic data fetch fills supported calculations and next steps; it does not supply missing operational topology, causal effects or private records.
These calculations fill a dimension only when its source coverage and comparison requirements are met. A supported method can still be unavailable for a particular country, year, river or event. Each result retains its observation period, source, units and required inputs.
| Console and dimensions | Calculation and interpretation |
|---|---|
| Country health: 1 and 3 | Same-year population balance and people-per-land-area symmetry between the declared income groups, using exactly the countries included in the comparison. These national aggregates do not measure local service access or catchment equity. |
| Air pollution: 8 | Five consecutive annual population-weighted PM2.5 estimates integrated above the WHO annual guideline of 5 µg/m³. Historical concentration exposure in µg·year/m³ is not attributable illness or an intervention benefit. |
| NHS Scotland: 4 and 10 | Quarterly residence-to-treatment board stays establish positive care-use connections and the reported share treated outside the residence board. Missing cells remain unknown; repeated stays are not unique people, transfers or current bed capacity. |
| Resilience Telemetry: 3 | Total variation between return distributions in shared pooled-quantile bins, using at least 20 common-date observations. Ties merge bins. This is a probability-density comparison, not spatial density or portfolio exposure. |
| Rivers: 2, 5 and 8 | Single-gauge distribution changes between complete observed hourly windows, persistence above the official action threshold and integrated exceedance over the action-to-major ladder. Units and datums remain gauge-specific; these are hydraulic-observation summaries, not property flooding or flood duration estimates. |
| Grid: 2 and 6 | Demand-distribution changes between complete 24-hour windows and sensitivity of the demand-to-recent-peak ratio to complete lookback windows. A historical demand peak is not rated capacity, reserve margin or dispatch guidance. |
| Earthquakes: 1 | Country population-share composition of USGS PAGER modeled MMI7+ exposure compared with its MMI4+ baseline. The exposure groups are nested; zero higher exposure cannot define a composition. Neither group establishes observed losses. |
| Respiratory and overdose surveillance: 6 | Directional sensitivity between a national series and jurisdictions with matching definitions and reporting periods. Available-jurisdiction coverage is stated; missing states do not imply an all-clear. |
Variational free energy remains unavailable without a specified probability model and its energy/entropy terms. Public data also cannot supply private service weights, verified pipes, current staffing, individual bills or personal portfolio holdings.
Checkboxes record local planning review, not completed operations. Notes reset when the evidence or scenario changes. JSON export and printing run locally; the checklist does not transmit these notes or read private patient, debt, medication, wearable or portfolio inputs. Keep sensitive operational information in your approved systems.
Research: Contamination-resilient CRT design · Directed asymmetry and contagion · Manuscript → code → demo map.
Maintainers: Integration contract and supported-input rules.