Start with the shock
Represent where and when a climate hazard occurs, as an externally declared episode or observed weather evidence.
A computational research project
Climate Health Resilience Simulator
Risk maps show where. CHRS tests what happens next. Heat can raise health need, but that does not reveal who will seek care, who can reach a service, or whether capacity is left when they arrive. The model follows each step as its own mechanism — from exposure through need, seeking, access and facility choice to finite capacity and served, delayed or unmet care.
Current stage: the simulation framework is software-verified. A controlled synthetic benchmark makes every mechanism inspectable. A Barcelona structural implementation now incorporates selected official city inputs for population, geography, facilities and context — while behavioural and service mechanisms remain uncalibrated, and historical validation has not begun.
Climate hazard→
Exposure→
Vulnerability→
Care need→
Seeking→
Access→
Facility choice→
Finite capacity→
Served · delayed · unmet
The question behind CHRS
What happens between exposure and actual care?
Maps can show heat, vulnerability and service locations, but not what happens between them. A person may need care and never seek it. They may seek care but fail to reach a suitable facility. They may arrive and still wait because capacity has run out. CHRS keeps those steps separate so each one can be examined — and challenged — on its own.
Ask who develops need
Generate care need before any decision about seeking, access or allocation — so the denominator never depends on how the service system responds.
Follow what people do next
Need does not automatically become healthcare use. Recognition, seeking, mobility, access and facility choice are separate steps.
Then test the service system
Facilities have finite effective capacity. Queues, disruption and rerouting produce served, delayed or unmet care.
The research path
Explore. Describe. Explain. Then prescribe.
CHRS separates the research process into four stages. The project is currently strongest in the first three: exploring what belongs in the model, describing what it produces, and explaining which mechanisms cause it. Prescription waits until empirical calibration and historical validation exist.
Exploration
What variables, pathways and assumptions matter enough to include? Synthetic experiments probe the model and show where its behaviour is sensitive, surprising or incomplete.
Description
What is happening, where, and to whom? CHRS describes exposure, care need, access, queues, service receipt and subgroup differences without treating them as the same thing.
Explanation
Which mechanisms produce the pattern? Scenario contrasts isolate pathways such as heat exposure, transport access, opening hours, mobile services and infrastructure disruption.
Prescription
This is the destination, not the current claim. Responsible prescriptive use needs calibrated parameters, withheld validation data and a defensible link between model outputs and real decisions.
Below is the causal chain implemented in the current model.
Climate hazard
An externally defined heat episode or observed weather evidence.
Spatial exposure
Place- and time-specific environmental exposure by zone.
Population vulnerability
Heterogeneous susceptibility, chronic burden and social context.
Care need
Routine, urgent or emergency need generated before service constraints.
Care seeking + access
Recognition, behaviour, mobility, accessibility and facility selection.
Finite service capacity
Effective capacity, queues, disruption, allocation and rerouting.
Outcomes
Served, delayed or unmet care; service-pressure and equity outcomes.
Adaptation strategies can intervene at multiple points in the chain: reducing exposure, improving accessibility, protecting capacity or targeting support to populations at higher risk.
Where the project stands
Three layers, three different levels of confidence.
CHRS now rests on three layers that should not be confused with one another. The engineering layer is verified. The synthetic benchmark is controlled and inspectable. The Barcelona implementation incorporates real city structure — but structure is not calibration, and calibration is not validation.
Verified computational framework
Deterministic named random streams, a frozen daily event order, demand-ledger reconciliation and strict output artifacts are covered by the repository test suite. Engineering verification supports claims about implementation consistency — nothing more.
Controlled synthetic benchmark
A 1,000-resident population runs through seven scenario contrasts, thirty paired replications each. It exists precisely because it isolates model mechanics from city-specific evidence.
Barcelona structural implementation
Official population margins, geography and crosswalks, facility identity, type and location, area-level socioeconomic context and AEMET weather evidence are incorporated for selected structural components. Need, seeking, capacity, queue and heat-response mechanisms remain assumed.
Software verified
Barcelona structural implementation available
Empirical calibration pending
Historical validation pending
What this means in practice: CHRS can be inspected, reproduced and questioned today. It cannot yet estimate real care need, unmet care or intervention effectiveness anywhere — including Barcelona.
The first city implementation
Barcelona gives the model a real skeleton.
Observed structure, assumed mechanisms
Structural pilotThe Barcelona implementation incorporates selected official observed inputs for city structure and context. Behavioural, capacity, queue and heat-response mechanisms remain uncalibrated. That distinction is the whole point of the pilot: it shows exactly which parts of the pathway rest on evidence, and which still rest on assumption.
Population
Official population-register margins and a deterministic sampled simulation population.
Geography
Official district and area boundaries with harmonised crosswalks.
Facilities
Observed facility identity, type and location. Capacity is not observed.
Socioeconomic context
Area-level income context used for stratification, not individual behaviour.
Vulnerability context
A historical spatial proxy — not current individual vulnerability.
Weather evidence
Observed AEMET station records inform heat-event evidence and episode parameterisation.
Still assumed or uncalibrated: baseline care need, care seeking,
operational capacity, queue behaviour, heat-response parameters, the daily synthetic
climate inside the engine, and geometric travel times. Static GTFS data are available
locally but not yet integrated into the production travel model.
What has been tested
What the checks establish, and what still needs evidence.
The code runs reproducibly and controlled runs activate the intended mechanisms. That supports the implementation. It does not support claims about any real city — those require empirical targets, calibration and withheld historical validation.
1
Implementation correctnessDeterministic execution, event ordering, capacity bounds, queue integrity and ledger reconciliation pass the repository test suite.
2
Mechanism activationEach declared scenario mechanism is checked against configuration and against mechanism events in generated outputs.
3
Health-need responsePending epidemiological evidence able to anchor routine, urgent and emergency need separately.
4
Care seekingPending conditional need-to-use evidence; attendance alone cannot identify seeking.
5
Service capacityPending operational evidence — staffing, hours, throughput — kept strictly separate from utilisation.
6
Historical validationPending a leakage-locked split with withheld outcome years and pre-registered acceptance criteria.
Who this could become useful to
A model for questions maps cannot answer.
CHRS is designed for counterfactual questions: what changes if access improves, service hours are extended, transport fails, or several pressures arrive at once? The model is meant to show why an outcome changed, not simply produce another map.
Public-health teams
Examine how climate-related health need may vary across populations and places.
Health-system planners
Stress-test service continuity, queues and finite capacity under disruption.
Climate adaptation teams
Compare downstream consequences of exposure, accessibility and capacity interventions.
Universities & researchers
Use an auditable framework for mechanism tests, sensitivity analysis and empirical validation.
Cities & municipalities
Connect vulnerability and service access to transparent spatial priorities.
National & multilateral programmes
Test how infrastructure or service disruption propagates through population-health pathways.
What comes next
The path from mechanism to evidence.
Stage 01
Core simulation frameworkFrozen daily event order, named random streams, ledger accounting.
Complete · verifiedStage 02
Controlled synthetic benchmarkn=1,000, seven scenarios, thirty paired replications per scenario.
AvailableStage 03
Barcelona structural implementationPopulation, geography, facilities and context incorporated; behavioural and capacity mechanisms still assumed.
Ready with conditionsStage 04
External evidence acquisitionUtilisation, need, seeking and operational-capacity targets; two formal aggregate requests drafted.
In progressStage 05
Empirical calibrationBlocked pending target evidence: need anchors, daily utilisation, operational capacity.
Not readyStage 06
Historical validationWithheld 2024–2025 outcomes under a leakage-locked split; acceptance criteria to be frozen before any fit.
Not readyStage 07
Policy-oriented intervention evaluationPrescriptive comparisons only after calibration and validation succeed.
FutureStage 08
Cross-city transferabilityWhether structure and mechanisms transfer beyond the first city.
FutureCollaborate
Help challenge the model.
CHRS is an independent research project. The most useful help right now is specific: pointing to epidemiological need evidence, conditional care-seeking surveys, daily or weekly service-use data, or operational capacity records for Barcelona; reviewing the calibration and identifiability reasoning; or testing reproducibility. Experience in climate health, healthcare use, service resilience or spatial adaptation is especially welcome.