← Back to catalogue
Published

Observation / Measurement Record

vr.wm-mat-008 · wm-mat-008-observation-measurement-record

Provide a format-neutral context structure for a single observation act and the measurement record it produces, binding a feature of interest, an observable property, an observing procedure, a result value and its unit, together with time, uncertainty, quality, provenance and governance context, so that an AI agent can create, inspect, validate and operate observation records without owning the lifecycles of the sensors, samples, methods or catalogues it references.

World Models Physical world and living systems PHY.MAT.OBS

Bundle → Layer → Finding → Questions Filled

6 bundles · 13 layers · 29 findings · 115 questions

Observation Act and Identity Establishes what one observation record is: how it is identified and versioned, how it is classified, and how it binds to the procedure execution and the executing observer that produced it.

Record Identity and Kind

Identity, versioning and classification of the observation record and of the act it documents.

Observation record identity and version identity

How a single observation record is uniquely and stably identified, how the identity of the record is distinguished from the identity of the observation act, and how successive versions of the same record are identified without colliding with new observations.

  1. Which system is the master system of record for this observation, and what identifier does it issue? identity
  2. Is this identifier for the observation act or for the record that documents it, and how are the two related when one act yields several records? identity
  3. How is a corrected or reprocessed version of this record identified relative to its predecessor? provenance
  4. If no authoritative or governed identifier exists, what surrogate is minted and how is that fact recorded? authority

Observation kind and result-type classification

How an observation is typed so that an agent can predict the shape of the result and the applicable interpretation rules: the observation type or result type, the domain category, and whether the observation is direct, derived, simulated or estimated.

  1. What result type does this observation declare, and does the declared type match the actual result payload? classification
  2. Under which domain category or programme is this observation classified for discovery and rule selection? classification
  3. Was the value obtained by direct determination, by derivation from other observations, by simulation or by expert estimation? definition
  4. Which observation profile or requirements class does this record claim to satisfy? interoperability

Act and Procedure Execution

The binding between the observation record and the procedure, observer, host and deployment that executed it, together with the execution-specific parameters in force.

Procedure, observer, host and deployment binding

How the record names the observing procedure that was executed, the observer or system that executed it, the host or platform carrying that observer, and the deployment in force at the time, each by reference to the model that owns it.

  1. Which observing procedure, at which version, was executed to produce this result? relationship
  2. Which observer, sensor, analyst or computational system carried out the procedure? relationship
  3. Which deployment of that observer onto which host was in force when the act occurred? composition
  4. Were there documented deviations from the procedure during this execution, and where is the deviation record held? exception

Observation parameters and influence quantities

Named execution parameters and influence quantities that qualify the observation but are neither the measurand nor the result: settings, conditions applied, dilution or scaling factors, and any parameter needed to make the result reproducible.

  1. Which named parameters qualify this observation, and what value and unit does each carry? measurement
  2. Which influence quantities were recorded or controlled because they affect the measured value? constraint
  3. Which parameters are inherited from the series, datastream or method default rather than stated on this record? composition
Subject and Property Binding What was observed and on what: the feature of interest chain including any sample that stood in for it, and the observable property and measurand actually determined.

Feature of Interest Chain

The proximate and ultimate features of interest and the sampling chain that connects them.

Proximate and ultimate feature of interest

How the record states the feature the result is actually about and, where observation was indirect, the intermediate feature that was directly observed, so that a result taken on a sample is never mistaken for a result taken on the whole.

  1. Which real-world feature is this result ultimately a statement about? relationship
  2. Which feature was directly observed, and does it differ from the ultimate feature? relationship
  3. On what basis is the proximate feature held to represent the ultimate feature, and what limits that representativeness? evidence
  4. How is a feature reference resolved when the referenced feature is renamed, merged or split? interoperability

Sample linkage and sampling-chain reference

How an observation made on a sample records which sample, where that sample sits in a sub-sampling or derivation chain, and which preparation steps had been applied before the observation, without absorbing the sample lifecycle.

  1. Which sample or specimen was this observation performed on? relationship
  2. Where does that sample sit in the sub-sampling or derived-sample chain relative to the originally collected sample? composition
  3. Which preparation steps had been applied to the sample at the moment of observation? process
  4. Were there sample-condition findings that qualify the validity of this result? quality

Observable Property and Measurand

The property observed and the fully qualified measurand, including the conditions that make the quantity intended to be measured unambiguous.

Observable property binding

How the record names exactly one observable property by reference to a governed vocabulary, distinguishes the property from its quantity kind, and handles properties that are only definable relative to a method.

  1. Which observable property term, from which governed vocabulary and version, does this observation determine? identity
  2. What quantity kind underlies the property, and is it dimensionally consistent with the recorded unit? validation
  3. Is this property method-defined such that results from different procedures are not comparable? constraint
  4. Is the observed characteristic a quantity or a nominal property with no magnitude? classification

Measurand specification and stated conditions

The quantity intended to be measured, stated completely enough to be unambiguous: the property, the feature or matrix it is measured in, and the conditions (temperature, pressure, basis, fraction, time base) under which it is defined.

  1. What is the complete statement of the quantity intended to be measured, including the conditions that define it? definition
  2. On what basis is the value expressed (wet or dry, as-received, normalised, per unit mass, volume or area)? measurement
  3. What definitional uncertainty arises because the measurand is not fully specified, and is it bounded? constraint
  4. Over what time base or integration window is the measurand defined? temporal
Result, Quantity and Uncertainty The result itself: its value and datatype, its unit and quantity binding, its absence or censoring, the uncertainty attached to it, the traceability that makes it comparable, and the interpretation or conformity statement drawn from it.

Result Representation

How the result value is represented, dimensioned and qualified when it is missing or bounded.

Result value, datatype and cardinality

How the result payload is carried for simple scalars, categorical values, vectors, components, coverages, images and time-series blocks, including how the declared datatype constrains the payload and how multi-component results relate to their parent.

  1. What is the shape of the result payload, and which datatype governs its parsing? definition
  2. If the result has components, what is each component's property, value and unit, and can the parent be interpreted without them? composition
  3. How many significant figures or what resolution does the recorded value carry, and is that a property of the value or of its rendering? measurement
  4. For a large or binary result payload, how is the payload bound to the record and its integrity verified? evidence

Unit of measure and quantity binding

How the unit of a quantitative result is recorded unambiguously by naming both the code system and the code, how series-level unit defaults are reconciled with record-level values, and why conversion is forbidden without recording it.

  1. Which unit code, from which unit code system and version, applies to the recorded value? identity
  2. If the series or datastream declares a unit and this record declares another, which prevails and how is the conflict recorded? exception
  3. Was the value converted from the unit in which it was originally determined, and where is that conversion recorded? provenance
  4. If no unit is recorded, is the result genuinely dimensionless or non-quantitative, and how is that asserted? validation

Absent, censored and bounded results

How the record expresses that no value exists, that a value is bounded rather than determined (below detection or quantitation limit, above range), or that a value was suppressed, so that absence is never encoded as zero or as an unexplained null.

  1. If no result value is present, what is the declared reason for its absence? exception
  2. Is the value censored, and if so what is the bound, its direction and the limit that produced it? measurement
  3. What rule must a consumer apply when aggregating or comparing a censored value? constraint
  4. If the value exists but is withheld, what authority withheld it and under what condition may it be released? access

Uncertainty, Traceability and Interpretation

What makes the result usable as evidence: its stated uncertainty, its metrological traceability, and any interpretation or conformity statement drawn from it.

Measurement uncertainty statement

How the uncertainty associated with the reported value is stated so that it can be used: the kind of uncertainty, its value and unit, the coverage factor and coverage probability where expanded, and the basis on which it was evaluated.

  1. What kind of uncertainty is reported, with what numeric value and unit? measurement
  2. For an expanded uncertainty, what coverage factor and coverage probability apply? measurement
  3. On what basis was the uncertainty evaluated, and which uncertainty budget or method does it come from? evidence
  4. If no uncertainty is reported, is that a declared decision and on what ground? authority

Metrological traceability declaration

How the record declares that the result is related to a stated reference through a documented unbroken chain of calibrations, by reference to the calibration and reference-material evidence in force at the time of observation.

  1. To which stated metrological reference is this result traceable? authority
  2. Which calibration of the observer or measuring system was in force when the observation was made? temporal
  3. Is any part of the traceability chain unestablished or broken, and how is that disclosed? quality
  4. Which certified reference materials or reference measurements underpin the chain for this determination? evidence

Reference ranges, interpretation and conformity statements

How a result is turned into a judgement: the reference range or specification limit it was compared against, the resulting interpretation code, and, where a statement of conformity is issued, the decision rule identifier and the risk basis behind it.

  1. Against which reference range or specification limit was this result compared, and for which population or condition does that range apply? relationship
  2. What interpretation was recorded for the result, and using which code list? classification
  3. If a statement of conformity was issued, which decision rule was applied and where is it defined? decision
  4. What level of false-accept or false-reject risk does the applied decision rule carry, and who accepted it? authority
Time, Place and Context The temporal frame that makes a value locatable in time, the resampling semantics that make it safe to aggregate, and the spatial and environmental context in which it was obtained.

Temporal Frame

The several distinct times attached to an observation and the semantics governing how the value behaves across time.

Phenomenon, result, valid and ingestion time

How the record distinguishes the time the observed phenomenon applies to, the time the result became available, the interval over which the result is asserted to be valid, and the time the record was captured or ingested, each with an explicit offset.

  1. To what instant or interval in the real world does this result apply? temporal
  2. When did the result become available, and separately when was it captured into this record store? temporal
  3. Over what period is this result asserted to remain valid for use? state
  4. What clock and offset were in force at the observing location, and how are local-time ambiguities resolved? constraint

Aggregation, interpolation and sampling regime

The semantics that determine how a value may be resampled, aggregated or compared: whether it is instantaneous, averaged, accumulated or cumulative, over what duration, and how interim values between observations may be inferred.

  1. Does this value describe an instant or an interval, and if an interval, over what duration was it aggregated? measurement
  2. What interpolation type governs inference of values between this observation and its neighbours? constraint
  3. Is the value cumulative or a reset-prone running total, and what is the accumulation start reference? state
  4. At what nominal frequency or sampling regime was the series producing this observation operating? process

Spatial and Environmental Context

Where the observation applies and under what conditions the observing system was operating.

Observation location, geometry and spatial reference

How the record states the position or extent the result applies to, distinguishing the location of the observer from the location of the feature or sample, and always naming the coordinate reference system and vertical datum.

  1. To what position or spatial extent does the result apply, as distinct from where the observer stood? spatial
  2. Which coordinate reference system, vertical datum and epoch express these coordinates? interoperability
  3. What positional uncertainty attaches to the stated geometry, and how was it obtained? quality
  4. Was the location inherited from a station, site or sample record rather than measured for this observation? provenance

Operating and environmental conditions of validity

The conditions under which the observing system was operating and the environment surrounding it, which bound the applicability of any stated system capability such as accuracy, resolution or detection limit.

  1. Were the observing system's declared normal operating conditions satisfied at the time of the act? validation
  2. What environmental context surrounded the observation and could plausibly bias the result? measurement
  3. Under which validity context does the stated accuracy, resolution or detection limit hold for this observation? constraint
  4. Was the system operating in a degraded, suboptimal or survival regime, and how is that disclosed with the result? exception
Quality, Validation and Lifecycle How the record moves through its states, how it is corrected, what quality judgements have been recorded against it, and who is responsible for it.

Record Lifecycle

The states an observation record can occupy and the controlled transitions between them.

Record status and permitted transitions

The lifecycle states of an observation record from provisional capture through review to final, amended, superseded or retracted, and the rules governing which transitions are permitted, by whom and with what evidence.

  1. What is the current lifecycle status of this record, and what does that status permit a consumer to do with it? state
  2. Which transitions out of the current status are permitted, and what precondition does each require? lifecycle
  3. Who is authorised to move this record between statuses, and how is that authorisation evidenced? authority
  4. Does this record represent a real-time provisional value or a delayed-mode reviewed value, and are both retained? process

Amendment, correction and supersession

How a result that must change is handled: whether by amendment in place with a reason, by issuing a superseding record, or by retraction, always preserving the prior content and the reason for change.

  1. Which change mechanism applies to this record: amendment, supersession or retraction? lifecycle
  2. What reason is recorded for the change, and is it classified as a correction, reprocessing or reinterpretation? provenance
  3. How is the superseded content preserved and made retrievable after the change? evidence
  4. How are consumers of the superseded value informed that it has changed? interoperability

Quality Assessment

Quality judgements recorded against the observation and the evidence supporting them, recorded but never evaluated here.

Quality flags and quality-control outcomes

How quality judgements produced by external quality-control procedures are attached to the record: the flag value, the code list it comes from, which test produced it, when, and by which agent, without this model owning the tests or their execution.

  1. What quality flag is recorded against this result, and from which flag code list and version? quality
  2. Which quality-control test or test suite produced that flag, and where is the test defined? relationship
  3. When was the flag assigned and by which agent or service? provenance
  4. When several flags disagree, which prevails for consumption and how is the disagreement preserved? exception

Validation and verification evidence references

How the record points to the evidence that supports its acceptance: control results, replicate or duplicate observations, blanks, spikes, proficiency-test outcomes and review sign-offs, each held by the model that owns it.

  1. Which control, blank, spike or replicate observations support the acceptance of this result? evidence
  2. Against which acceptance criteria was this result judged, and where are those criteria defined? validation
  3. Was the record reviewed and released, by whom and on what date and time? authority
  4. Was the procedure verified as fit for purpose in this laboratory for this matrix before the observation? requirement

Provenance and Responsibility

Where the record came from, what it was derived from, and who is accountable for it.

Provenance chain and derivation

How the record states what generated it, which prior entities it used or was derived from, and how the original acquisition record is preserved so that the reported value can be traced back to what the instrument or observer actually produced.

  1. Which activity generated this record, and what did that activity use as input? provenance
  2. Which prior observations or entities was this result derived from, and by what transformation? relationship
  3. Where is the original acquisition record, and how is the reported value shown to correspond to it? evidence
  4. Has this record been reprocessed, and which processing version produced the current value? lifecycle

Ownership, attribution and responsible agents

Who owns the record, who is attributed as its producer, which organisation is the authoritative source for which part of it, and who may be contacted about it.

  1. Which organisation owns this record and is accountable for its correctness? ownership
  2. To whom is the observation attributed, and does that differ from the owner? ownership
  3. Which organisation is the authoritative source for each part of the record when several contribute? authority
  4. Who is the responsible contact for questions about this observation, and how is that contact kept current? process
Governance, Access and Interoperability How the record is protected and retired, how it is expressed for exchange without losing meaning, how its identifiers resolve across systems, and how it declares its membership in larger collections.

Access and Retention

Classification of the record for disclosure and its binding to a retention and disposition policy.

Access classification and disclosure constraints

How the record declares its sensitivity, the licence or access rights under which it may be used, and any constraint arising from personal, commercial or security sensitivity of the observation or its subject.

  1. What sensitivity classification applies to this record and to its result payload separately? security
  2. Under which access rights and licence may this observation be used and redistributed? access
  3. Does the observation identify or relate to an identifiable person, and what constraint follows? privacy
  4. Is the record embargoed or restricted for a period, and what releases it? exception

Retention, disposition and tombstoning

How long the record and its payload must be kept, which policy determines that, what disposition decision was reached, and what evidence remains after the record is destroyed.

  1. Which retention policy governs this record, and what retention period does it impose? retention
  2. Do the record, its result payload and its original acquisition record carry different retention periods? retention
  3. Is the record under legal hold or another suspension of disposition, and who imposed it? authority
  4. What tombstone remains after disposition, and what may it disclose? lifecycle

Interoperability and Membership

Alignment to external observation standards, resolution of identifiers across systems, and membership in collections and series.

Standard alignment and encoding binding

Which external observation standards this record is aligned to, which elements map cleanly, where mappings are lossy, and how the record is projected into a named exchange encoding without the encoding becoming the semantics.

  1. Which external observation standards and versions is this record aligned to, and is the alignment tested or asserted? interoperability
  2. Which elements are lost, coerced or fabricated when projecting into each target encoding? constraint
  3. Where do two aligned standards disagree about the same concept, and which is authoritative for this record? decision
  4. Which encoding and profile was used for a given exchanged representation of this record? provenance

Identifier resolution and cross-system correlation

How identifiers on and around this record resolve to the things they name, how the same observation is correlated across systems that assign their own keys, and how duplicate or conflicting records are detected and reconciled.

  1. How does a consumer resolve each identifier on this record to the entity it names? interoperability
  2. How is the same observation correlated when two systems have each assigned their own key? relationship
  3. What constitutes a duplicate observation, and how is a duplicate reconciled rather than silently merged? validation
  4. What happens when a referenced entity has been superseded, merged or destroyed since the observation was made? exception

Collection, datastream and series membership

How a single observation declares its membership in the collections, datastreams, series or composite observations that contain it, and which properties it inherits from them, without absorbing the containing aggregate's identity or completeness accounting.

  1. Which collection, datastream, series or dataset does this observation belong to? composition
  2. Which properties does this observation inherit from its container, and which does it override? composition
  3. Is this observation part of a composite or grouped observation, and can it be interpreted on its own? relationship
  4. If extracted from its container, does the record still carry everything needed to interpret the value? validation

Classifiers Filled

Family
World Models
Category
Physical world and living systems
Entry kind
event
Navigation path
NAV.PHY.MAT.OBS
Domain
PHY.MAT.OBS
Industry
Cross-industry
Tags
observationmeasurementrecordphy.mat.obs

What it is Filled

The subject is the observation event and its resulting record: the act of determining a value for an observable property of a feature of interest by applying an observing procedure, plus everything needed to interpret, trust, correlate and retire that single record. OMS/ISO 19156 and SOSA/SSN both model an Observation as an execution (an act), so this model is an event, not the sensor, the sample, the property definition or the dataset. It carries references, bindings and observation-specific parameters for those neighbours; it does not reproduce their identity, lifecycle or operational machinery. Storage and interface (JSON, XML, RDF, Markdown, Git, MCP, MongoDB, relational LIMS tables) are projections of this semantics, never part of it.

In scope

  • Identity, versioning and classification of one observation record and of the observation act it documents
  • Binding to exactly one observable property and one observing procedure, and to the observer, host and deployment that executed it, by reference
  • Proximate and ultimate feature of interest, and the reference to the sample or sampling chain when observation is indirect
  • Result representation: value, datatype, cardinality, complex or coverage payloads, and the unit of measure or quantity-kind binding
  • Absent, censored, below-detection-limit, out-of-range and non-numeric results, with the reason they are absent or bounded
  • Measurement uncertainty statements (standard, expanded, coverage factor and coverage probability) recorded against the result
  • References to metrological traceability chains and calibration evidence held by other models
  • Reference ranges, interpretation codes and recorded statements of conformity together with the identifier of the decision rule applied
  • Temporal frame: phenomenon time, result time, valid time and ingestion or observation-capture time, each with explicit offsets
  • Aggregation duration, interpolation type, cumulative flags and sampling regime that determine how the value may be resampled
  • Observation geometry, observed area and the environmental or operating conditions in force during the act
  • Record lifecycle status, amendment, correction, supersession and retraction semantics for this record only
  • Recorded quality flags and quality-control outcomes produced elsewhere, plus references to the evidence supporting them
  • Provenance chain, derivation from prior observations, responsible agents, ownership and attribution
  • Access classification, retention binding, and identifier resolution and cross-system correlation
  • Alignment declarations and encoding bindings to external observation standards, and membership in collections, datastreams or series

Out of scope

  • The sensor, instrument or observing system as a managed asset, including its own registration, maintenance, firmware and decommissioning lifecycle
  • Calibration and traceability records as governed documents: this model references a calibration certificate but does not issue, renew, revoke or verify one
  • The sample or specimen as an entity, including collection, preparation, custody transfer, storage and disposal lifecycle
  • Definition and governance of observable properties, quantity kinds and controlled vocabularies, including term proposal, deprecation and mapping maintenance
  • Unit-of-measure registry maintenance, unit conversion tables and dimensional-analysis services
  • The observing procedure or test method as a controlled specification, including method development, validation studies and version approval
  • Execution of quality-control tests, rule evaluation, thresholding or automated flagging: this model records the outcome and its provenance, never the evaluator
  • Audit-trail capture, tamper-evidence infrastructure and electronic-signature enforcement, which belong to the electronic-records model
  • Physical execution of deletion, purge or anonymisation; this model records the disposition decision and its policy binding only
  • Dataset catalogue publication, licensing negotiation, DOI minting and distribution management for collections of observations
  • Downstream analytics, model assimilation, forecasting and statistical inference beyond recording that a result was derived
  • Sensor tasking, actuation and control commands
  • Laboratory quality-management-system scope, accreditation status and personnel competence records
  • Transport-level streaming guarantees, message ordering, delivery semantics and API pagination

Why it exists Filled

Provide a format-neutral context structure for a single observation act and the measurement record it produces, binding a feature of interest, an observable property, an observing procedure, a result value and its unit, together with time, uncertainty, quality, provenance and governance context, so that an AI agent can create, inspect, validate and operate observation records without owning the lifecycles of the sensors, samples, methods or catalogues it references.

Distinguishing features Filled

  • Binds a feature of interest, observed property, procedure, result value and unit for one observation act.
  • Unlike the sensor asset or calibration record, it references them.
  • Unlike a sample record, it describes the measurement, not the material.
  • Corrections are additive; prior values are retained.

What robots and AI may and may not do Filled

Must not

  • Publish a value without unit, measurand or time.
  • Overwrite values or quality flags.
  • Compute quality evaluation as if it were the external service.
  • Claim conformance without evidence.
  • Mix units without conversion records.

Only with a human decision

  • Accepting corrections to values used in regulatory or clinical decisions.
  • Releasing observations of identifiable persons.

May

  • Register an observation with measurand, unit and time.
  • Record externally produced quality flags.
  • Supersede an observation while retaining the old value.
  • Project records to exchange encodings.

Moral aspects Filled

  • Measurements of people can be health or personal data.
  • Wrong measurements can lead to unsafe decisions.

Who is affected

  • Observed persons and communities
  • Data users
  • Measurement providers

Owners Filled

Steward

The adopting Dimension must nominate a single master system of record for observation identifiers and declare it before any record is registered; where several producing systems exist, the mapping from producing system to master identifier must be published, not inferred.

Roles

Observation Record Steward
Own the completeness and correctness of records within a programme, including identity resolution and binding of mandatory referents.; Approve amendments and supersessions and ensure change reasons are recorded.; Resolve duplicate and cross-system correlation questions without silently merging content.
Metrology Authority
Approve the uncertainty statements, coverage parameters and traceability declarations carried on records.; Confirm that calibrations and reference materials in force at the act are correctly referenced and that chain breaks are disclosed.; Rule on whether a decision rule and its risk basis may be applied to a class of results.
Data Quality Officer
Define which quality-control test suites apply to a programme and ensure their flag code lists are declared with versions.; Maintain the flag precedence rule and adjudicate disagreements escalated from records.; Verify that flags recorded on records carry test, agent and instant attribution.
Dimension Registrar
Declare the master system of record, the surrogate namespace and the pinned versions of external registries.; Maintain the lifecycle status code list and its permitted transitions.; Approve model version changes and publish migration statements for breaking changes.
Access and Retention Custodian
Bind records to sensitivity classifications, licences and retention policies.; Register and lift holds, and authorise disposition decisions.; Ensure tombstone content is correct and that disposed records no longer expose withheld fields.

Links to other meta-models Filled

references

  • Observable Property / Quantity Kind Registry (candidate sibling model; not yet registered) - Resolve the single observable property term and its underlying quantity kind that the observation determines. This model carries the pinned term reference, vocabulary version and the observation-specific method-dependence flag only; term definition, dimensioning, deprecation and mapping maintenance remain in the registry.
  • Observing Procedure / Test Method Specification (candidate sibling model; not yet registered) - Pin the procedure and version executed, plus execution-specific parameters and any recorded deviation pointer. Method authoring, validation studies, approval and withdrawal are owned by the procedure model and are not reproduced here.
  • Feature of Interest entity models (site, station, material, subject; candidate siblings) - Name the ultimate feature the result is about and, where indirect, the proximate feature. Feature identity, geometry history, renaming, merging and lifecycle stay with the entity model; this model records the reference and the representativeness basis.
  • Unit of Measure Registry (UCUM, SI, QUDT bindings; candidate sibling model) - Bind the result unit by naming both code system and code, and record any conversion applied. Unit definitions, conversion multipliers, offsets and dimensional vectors are owned by the registry; the link is required only when the result is a quantity.
  • Sample / Specimen Record (candidate sibling model; not yet registered) - Reference the sample observed, its position in the sub-sampling chain and the preparation state effective at the observation instant. Sample collection, preparation execution, custody and disposal lifecycles remain outside this model.
  • Observer / Instrument Asset and Deployment (candidate sibling model) - Reference the observer, host and deployment in force at the act, and the declared system capability with its validity context. Asset registration, maintenance, firmware and deployment scheduling are owned by the asset model.
  • Calibration and Metrological Traceability Record (candidate sibling model) - Record which calibration and reference materials were in force and whether the traceability chain is unbroken. Calibration issuance, validity determination, renewal and revocation belong to the calibration model; no certificate content is copied here.
  • Quality-Control Procedure and Evaluation Service (candidate sibling model) - Carry the flag value, the identifier of the test that produced it and the assigning agent. Rule authoring, threshold configuration, execution, re-evaluation and enforcement are owned entirely by the quality-control model; this model records outcomes only.
  • Electronic Records and Audit Trail (candidate sibling model) - Provide linkage from a record and its status transitions to the audit entries and signatures that independently attest to them. Audit capture, tamper-evidence, signature binding and audit query remain wholly outside this model.
  • Retention and Disposition Policy of the adopting Dimension - Bind every record, payload and original acquisition record to a governing retention policy and record the resulting disposition decision and tombstone. Policy authoring, legal-hold determination and physical destruction are executed under the adopting Dimension's policy, not here.
  • Observation Collection / Datastream / Time Series aggregate (candidate sibling model) - Declare membership and record the effective inherited defaults such as unit, observation type and interpolation type. The aggregate owns its own identity, extent, completeness accounting and publication lifecycle.

aligned

  • Dataset Catalogue publication model aligned to W3C DCAT 3 - Align record-level access rights, licence and versioning vocabulary with catalogue-level DCAT terms so that observations aggregate cleanly into published datasets. Catalogue entries, distributions, data services and publication workflow are not modelled here.
  • OGC Abstract Specification Topic 20 / ISO 19156:2023 Observations, Measurements and Samples - Declare structural alignment to the OMS conceptual schema for Observation, ObservableProperty, Observer, ObservingProcedure, Deployment and Sample. Alignment is asserted, not certified; conformance requires testing against the published requirements classes.
  • W3C/OGC SOSA/SSN Semantic Sensor Network Ontology - Declare vocabulary alignment for Observation, Result, FeatureOfInterest, Procedure, Sample and the system-capability and operating-conditions terms, enabling graph projection without adopting SSN's actuation surface.
  • W3C PROV-O provenance vocabulary - Align the generation, usage, derivation and attribution statements on the record with the PROV starting-point and qualified terms so provenance is portable. Provenance stores, inference and validation services remain external.
  • HL7 FHIR R5 Observation resource - Align status, absence reason, interpretation, reference range, component and member semantics with a widely deployed independent observation model, and record where the mappings are lossy. Clinical workflow, encounter and ordering semantics are excluded.

extends

  • Domain observation profiles (clinical, environmental, geotechnical, materials-testing; candidate descendants) - Domain profiles specialise this model by constraining code lists, adding domain-mandatory parameters and tightening cardinalities. They must not restate generic identity, versioning, temporal, provenance or retention machinery, which this model already owns.

neighbor

  • Sample / Specimen Record - OMS and SOSA separate the Sample (a feature used to represent another feature) from the Observation (the act). This model carries the reference from the observation to its proximate feature of interest and, where that is a sample, to the sample identifier and the ultimate feature it stands for. Sample identity, sampling act, preparation steps, sub-sampling relationships and custody remain owned by the sample model.
  • Observable Property / Quantity Kind Registry - OMS requires an observation to name exactly one ObservableProperty, and QUDT separates QuantityKind from Unit. This model holds the reference and any observation-specific constraint or qualifier on the measurand; it does not define the property term, its dimension, its synonyms or its deprecation policy.
  • Unit of Measure Registry (UCUM / SI / QUDT) - UCUM and QUDT are governed unit code systems. This model records which code system and code express the result, and any observation-level scaling; it never redefines units, conversion multipliers or dimensional vectors, and it must not silently convert values.
  • Observing Procedure / Test Method Specification - SOSA defines Procedure as a workflow, protocol, plan, algorithm or computational method, and OMS makes the ObservingProcedure a first-class feature. This model records which procedure version was executed and the execution-specific parameters; method authoring, validation and approval belong to the method model.
  • Observer / Instrument Asset and Deployment - OMS models Observer, Host and Deployment (with deploymentReason and deploymentTime) as separate features. This model references the observer and the deployment in force at the time of the act; asset lifecycle, platform management and deployment scheduling stay outside.
  • Metrological Traceability / Calibration Certificate - VIM defines metrological traceability as a documented unbroken chain of calibrations. This model records the traceability declaration and pointers to the calibration evidence in force; it does not model the calibration act, the reference standard hierarchy or accreditation validity.
  • Quality-Control Procedure and Evaluation Service - QARTOD publishes QC test procedures and flag conventions; SensorThings exposes resultQuality as a recorded property. This model stores the resulting flag, the identifier of the test and the identifier of the evaluating agent. Rule authoring, threshold configuration, execution and re-evaluation are owned by the quality-control model and never by this one.
  • Electronic Records / Audit Trail - MHRA GxP data-integrity expectations place audit trail, tamper-evidence and signature control on the records system. This model requires that a record be linkable to its audit entries and preserves the original acquisition record as an artifact, but it does not define, generate, secure or query the audit trail itself.
  • Dataset / Catalogue Publication (DCAT) - DCAT governs Dataset, DatasetSeries, Distribution and DataService, with licensing, access rights and versioning at dataset level. This model records membership pointers from an observation to the collection, datastream or series that contains it; catalogue-level metadata, distribution management and publication workflow stay in the catalogue model.
  • Time Series / Observation Collection Aggregate - SensorThings binds unitOfMeasurement and observationType at Datastream level, and the OGC Timeseries Profile defines series-level defaults such as interpolationType. Series-level defaults are referenced here so a single record remains interpretable in isolation, but the series aggregate owns its own identity, extent, default resolution and completeness accounting.
  • Retention and Disposition Policy - This model records which retention policy applies, the disposition decision and the tombstone it leaves behind. Policy authoring, legal-hold determination and the physical execution of destruction are owned by the adopting Dimension's records policy.

What else AI and robots need to interact with it Filled

Identity and identifiers required Filled

  • Authoritative master-system identifier issued by the declared system of record for the observation, such as a laboratory information management system result key, an observing service observation key or a clinical result identifier; this identifier takes precedence whenever it exists and must be carried verbatim.
  • Governed global identifier or IRI issued by a normative registry or namespace, such as a resolvable observation IRI under a governed namespace or a persistent identifier assigned by a recognised data authority, used when no master-system identifier exists.
  • UUID or ULID minted by the adopting Dimension, used only when neither of the above exists, recorded with an explicit surrogate marker and its minting namespace so it can be superseded additively once an authoritative identifier appears.
  • A phenomenon time, result time, sample number, file name, run sequence or any date value is never an identifier; such values may only participate as qualifying components of an explicitly scoped composite key alongside a real identifier.

Direct properties not applicable Not applicable

Not applicable

An observation record holds measured values as data; the observed object carries the physical properties.

Recognition optional Filled

  • An observation has a feature of interest, an observed property, a procedure, a result with unit and a phenomenon time.
  • Often confused with a sensor, a sample, a calibration certificate and a derived statistic.

Capabilities and actions required Filled

  • Register observation record: Create a new observation record, resolve its identity under the model identity priority, and establish its initial lifecycle status and ownership.
  • Bind measurand and observation context: Attach the mandatory referents that make the record interpretable: exactly one observable property, exactly one observing procedure, the ultimate feature of interest, and where applicable the proximate feature, sample, observer and deployment.
  • Record result value and unit binding: Attach the result payload, its datatype and, for quantities, its unit code and code system, or record a declared absence or censoring instead of a value.
  • Assert temporal frame: Set and validate the several distinct times on the record, ensuring each carries an explicit offset and that event time is not conflated with capture time.
  • Record externally produced quality outcome: Attach a quality flag, its code list, the test that produced it, the assigning agent and the assignment instant to the record. This function records an outcome computed elsewhere and never evaluates a rule.
  • Amend or supersede observation record: Apply a controlled change to a released record by issuing a new version or a superseding record, preserving prior content and recording the mechanism, reason and authority.
  • Assess record interpretability: Evaluate whether a record carries everything required to interpret its value in isolation, including mandatory referents, unit binding, temporal offsets, aggregation semantics and inherited container properties.
  • Project record to an exchange encoding: Produce a concrete representation of the record in a named encoding and profile, recording the encoding, the record version projected, the producing agent and any element-level loss.
  • Record retention and disposition decision: Bind the record to its governing retention policy, register any hold, and record the disposition decision and resulting tombstone. Execution of destruction is performed by the storage platform under the adopting Dimension's records policy.

Hazards and failure modes required Filled

  • Unsafe decisions from values without uncertainty or unit.
  • Data loss through overwrites.
  • Privacy loss from personal measurements.

Standards and interfaces required Filled

  • ISO 19156 Observations, measurements and samples.
  • OGC SensorThings API.
  • W3C SOSA/SSN.
  • UCUM unit codes.
  • HL7 FHIR Observation.

Context of use required Filled

  • MHRA GxP data-integrity expectations are UK-specific and, for GLP facilities, the OECD advisory document takes precedence; equivalent obligations elsewhere differ in detail and the adopting Dimension must bind the applicable regime rather than assuming this one.
  • ILAC guidance applies where the laboratory operates under an accreditation body signatory to the ILAC arrangement; observations produced outside accreditation carry no such obligation and their conformity statements have different standing.
  • WIGOS metadata expectations apply to WMO member observing programmes; they are used here only as evidence that observational metadata is categorised and multi-authority, not as a universal requirement.
  • QARTOD conventions originate in US ocean observing and are not international standards; other regions use different flag schemes.
  • Personal-data implications of observations about identifiable people depend entirely on the applicable data-protection regime, which is not assumed; the lawful-basis determination is placed outside this model.
  • UCUM is dominant in health informatics while SI symbols and QUDT dominate in engineering and physical sciences; the default unit code system must be declared per Dimension rather than assumed.
  • Coordinate reference frames may be dynamic in some jurisdictions, which is why a coordinate epoch is modelled; deployments in static-frame regions may leave it unset.

Sources Filled

  1. Observations, Measurements, and Samples (OMS) - Open Geospatial Consortium
  2. OMS Abstract Specification feature catalogue input (opengeospatial/om-swg) - Open Geospatial Consortium (OMS Standards Working Group)
  3. Semantic Sensor Network Ontology (SOSA/SSN), 2023 Edition - W3C / Open Geospatial Consortium
  4. OGC SensorThings API Part 1: Sensing, Version 1.1 (OGC 18-088) - Open Geospatial Consortium
  5. OGC Timeseries Profile of Observations and Measurements (OGC 15-043r3) - Open Geospatial Consortium
  6. PROV-O: The PROV Ontology - W3C
  7. RFC 3339: Date and Time on the Internet: Timestamps - Internet Engineering Task Force
  8. JCGM Publications: Guides in Metrology (JCGM 100:2008 GUM; JCGM 200:2012 VIM; JCGM 106:2012; JCGM GUM-6:2020) - Bureau International des Poids et Mesures / Joint Committee for Guides in Metrology
  9. The Unified Code for Units of Measure (UCUM) Specification - Regenstrief Institute, Inc. and the UCUM Organization
  10. QUDT Ontologies (Quantities, Units, Dimensions and Data Types) - QUDT.org
  11. HL7 FHIR Observation resource - Health Level Seven International
  12. ILAC Guidance Series (ILAC-G8:09/2019 Guidelines on Decision Rules and Statements of Conformity; ILAC-G17:01/2021 ILAC Guidelines for Measurement Uncertainty in Testing) - International Laboratory Accreditation Cooperation
  13. 'GXP' Data Integrity Guidance and Definitions - Medicines and Healthcare products Regulatory Agency (UK)
  14. Data Catalog Vocabulary (DCAT) - Version 3 - W3C
  15. QARTOD - Quality Assurance / Quality Control of Real Time Oceanographic Data - NOAA U.S. Integrated Ocean Observing System (IOOS)
  16. WIGOS Metadata Guidelines for Data Providers (wmo-im/et-acdm wiki, referencing WMO-No. 1192 and schemas.wmo.int/wmdr) - World Meteorological Organization (Expert Team on Metadata, Information Management)

Open questions

  • Obtain and read the published OGC 20-082r4 / ISO 19156:2023 requirements classes directly, then re-derive every required element and multiplicity in this model against them and retire the 2021 draft feature catalogue as the cardinality basis.
  • Split the composite source identifiers into one id per publication (JCGM 100, 101, 102, 106, 200, GUM-1, GUM-6; ILAC-G8 and ILAC-G17 separately), obtain parseable texts of GUM and VIM, and re-attach clause-level citations to the uncertainty, measurand and traceability findings.
  • Design an uncertainty, traceability and conformity function for the next revision so that the whole layer is reachable through the function set, together with functions for identifier resolution and duplicate reconciliation, provenance and derivation binding, and access classification, embargo and hold registration.
  • Verify WMO-No. 1192 WIGOS Metadata Standard and the QARTOD flag manual from their publications rather than a wiki page and a programme landing page, and re-assess the findings that currently lean on SRC-015 and SRC-016 as one of only three or four refs.
  • Refer the registry placement to the Dimension Registrar: NAV.PHY.MAT.OBS, the PHY.MAT.OBS domain tags, the WM-MAT identifier prefix and the 0.00 robotics factor all understate a model whose evidence spans geospatial, clinical, oceanographic and meteorological practice and whose SOSA grounding is sensor-centric.
  • Resolve whether measurement uncertainty is taken into account in the conformity decision (guard-band treatment under ILAC-G8) needs its own question, since the current interp-decision-rule and interp-risk-basis pair records the rule and the accepted risk but not how uncertainty entered the judgement.
  • Populate the relationship contract with the eleven asserted neighbours, assigning link types and stating which side owns each referent, so the ownership boundary becomes machine-checkable rather than prose.
  • The published normative text of OGC 20-082r4 / ISO 19156:2023 was not read directly: the OGC HTML exceeded the fetch size limit and the ISO catalogue page returned HTTP 403. Class names and property multiplicities are grounded in OGC working feature-catalogue material explicitly marked draft and dated 2021-08-20, and every cardinality claim in this model must be re-verified against the published requirements classes before the model leaves candidate status.
  • WMO-No. 1192, the WIGOS Metadata Standard, was verified only through a WMO Information Management team wiki page and not through the publication itself; the ten metadata categories are therefore treated as indicative and not cited as normative structure.
  • The JCGM 200 (VIM) and JCGM 100 (GUM) full texts could not be parsed from their PDFs; metrology terms are grounded in the BIPM JCGM publications index and in ILAC guidance that restates the requirements, so individual clause numbers are not cited.
  • The QARTOD flag manual PDF could not be parsed; flag semantics are grounded in the IOOS project page and treated as a community convention, not an international standard.
  • Domain-specific observation profiles are not modelled: clinical coding systems, INSPIRE profiles, geoscience and geotechnical observation profiles, and remote-sensing product conventions are all left to EXTEND descendants.
  • Sensor tasking and actuation, present in SOSA and in SensorThings Part 2, are excluded entirely; a model that must both observe and command will need a sibling actuation model.
  • Uncertainty propagation methods, including Monte Carlo propagation and multivariate output quantities, are referenced but not modelled; only the resulting statement is carried.
  • Streaming delivery guarantees, event ordering at transport level, back-pressure and API pagination are excluded as interface concerns.
  • Cost, effort and instrument-time accounting for an observation are not modelled.
  • Multi-language labelling and localisation of codes and statements are not addressed.

Machine files

Provenance

world-models research · reviewable-draft

Built from: models/wm-mat-008-observation-measurement-record/spec.yaml, ver-cy/world-models/card-supplements/wm-mat-008-observation-measurement-record.json