Skip to main content
URBAN ECOTOPIACLIMATE INTELLIGENCE · LONDONCheck pilot fit
METHOD / CONTROLLED EVIDENCE

DECISION BEFORE DATA

A controlled route from observation to professional review.

The method begins with one site decision, not a model. The site, options, evaluator, usefulness endpoint and release authority are fixed before Earth observation is tested.

EARTH OBSERVATION · PHYSICS · EXPLAINABLE AI · REVIEW

Evidence moves. Authority does not.

A proposed scientific workflow for turning archived climate signals into reviewable site evidence. Motion reveals each analytical hand-off and the limits that travel with it.

ILLUSTRATIVE SITE · RESEARCH-INFORMED METHOD · NO MEASURED RESULT
OBSERVATION ARCHIVEHISTORIC · DATED · RIGHTS-CHECKED
  1. Surface signal
  2. Vegetation
  3. Missing or masked

LST ≠ AIR TEMPERATURE ≠ THERMAL COMFORT

PHYSICALLY INFORMED FEATURES
  1. BUILDING FORM
  2. STREET TOPOLOGY
  3. CANOPY + SURFACE
  4. SOLAR + SKY VIEW
  5. SOCIAL CONTEXT
BOUNDED COMPARISONSAME SITE · SAME SCALE · EXPLICIT INPUT CHANGE
00EO-DERIVED BASELINEHistoric screening context
01CANOPY + SHADEAnalytical scenario
02SURFACE + ROOFAnalytical scenario

MODELLED CHANGE · NOT REALISED COOLING

  1. SOURCE RECEIPTDATE · LICENCE · UNIT
  2. VALIDATION CARDSPATIAL HOLD-OUT · SCALE
  3. EXPLANATIONMODEL FEATURE ATTRIBUTION
  4. UNCERTAINTYMISSING · OUT OF DOMAIN

MODEL FEATURE ATTRIBUTION ≠ CAUSAL EXPLANATION

CLIMATE INTELLIGENCESITE EVIDENCE PACKDecision scope · sources · assumptions · limits
ACCOUNTABLE REVIEWRELEASENARROWSTOP

THE MODEL CANNOT RELEASE ITS OWN RESULT

  1. 01SOURCERights · date · unit
  2. 02REGISTERTime · grid · coverage
  3. 03DERIVEPhysical features
  4. 04MODELSpatial validation
  5. 05EXPLAINReason · uncertainty
  6. 06RELEASEHuman decision
ARCHIVE REGISTEREDDATED BASELINE + COVERAGE
What was observed, when, and at what support?

Separate historic land-surface temperature, vegetation, urban form and social context. Cloud, coverage and source rights remain visible.

01 / 04
01 / EVIDENCE CLASSES

OBSERVATION IS NOT OUTCOME

Keep every inference boundary visible.

Earth observation, derived indicators, place representations, scenarios, model attribution and professional judgement are different objects. No arrow between them is automatic.

EVIDENCE CLASSIFICATIONEach transformation changes what the evidence may support.
  1. 01
    OBSERVED EO
    Satellite acquisition

    Sensor, date, time, unit and coverage

    Not air temperature or comfort
  2. 02
    DERIVED INDICATOR
    Surface evidence

    Documented processing and spatial support

    Not a causal mechanism
  3. 03
    EMBEDDING
    Optional retrieval feature

    Validated similarity for a defined task

    Not suitability or causality
  4. 04
    MODELLED SCENARIO
    Bounded comparison

    Fixed options, inputs and assumptions

    Not realised cooling
  5. 05
    FEATURE ATTRIBUTION
    Model explanation

    Contribution under one fitted model

    Not physical explanation
  6. 06
    PROFESSIONAL JUDGEMENT
    Review and release

    Named role, date, scope and decision

    Not implementation
  7. 07
    REALISED OUTCOME
    Monitoring after delivery

    Independent observed evaluation

    Not available in this concept
02 / MATCHED EO TEST

CHANGE IS NOT IMPROVEMENT

EO is judged against a pre-registered usefulness endpoint.

The non-EO and EO arms use the same site, options, spatial support and evaluator. An independent review asks whether the added observation improves a stated decision need, such as admissible prioritisation, confidence, evidence-gap closure, monitoring design or auditability. A ranking change alone does not pass.

PROPOSED EVALUATION · SECTION A-ADoes EO change the decision?
Illustrative architectural section through a courtyard showing shade, trees, heat exposure and permeable ground
  1. 01 Roof treatment
  2. 03 Shade structure
  3. 05 Canopy cooling
  4. 06 Permeable ground
WITHOUT EOCurrent site and design evidence

Establish the comparator before adding another spatial signal.

WITH EOSame question, one extra evidence layer

Test whether priority, confidence, gaps or auditability materially change.

Architectural concept image only. No invented temperatures or cooling outcomes. The gate remains open until evaluated on an authorised site.
03 / STOP CONDITIONS

FAIL CLOSED

The method can narrow, abstain or stop.

  1. 01Source rights are not approved
  2. 02Observation scale cannot support the site question
  3. 03Coverage or quality is inadequate
  4. 04EO does not improve the pre-registered decision endpoint
  5. 05Required specialist modelling is unavailable or unvalidated
  6. 06The accountable reviewer rejects the evidence