Skip to content

Observation and measurement

An interpretable observation needs more than a number: what entity or feature was studied, what property was targeted, how the result was obtained, and which support and context it represents. Different community models organize those elements differently.

Compare models without declaring them equivalent

OBOE organizes observations, measurements, entities, characteristics, standards and protocols. SOSA/SSN models observations, sensing, sampling, procedures, results and features of interest. These are useful structures with different commitments. A field called observation in two systems does not automatically denote the same class or event structure.

BCO addresses biological collection and sampling activity; OBI describes investigations and experimental procedures. Darwin Core and Humboldt supply exchange terms and inventory context. Exchange compliance and ontology-model compatibility remain separate questions.

Compose a variable's meaning

I-ADOPT provides a way to describe a property, object of interest, matrix, context and constraints. A concentration in water needs a different matrix description from concentration in tissue. Domain resources can identify the chemical, material and biological context. The framework does not replace all units, instruments, methods or geographical metadata; those need complementary resources.

Preserve the source model

Module 1 records declared concepts and source paths, with explicit evidence kinds. The SemanticEvidence contract permits partial records. Missing methods or ontology identifiers stay missing. Parser-derived normalized metadata is classified separately from a direct native annotation. Model-proposed hints remain MODEL_JUDGMENT.

Module 2 compares definitions, entity types, methods, units and support, then proposes a reviewed relationship. Module 3 sees the resulting harmonized observation contract, not an assumption that all observation models are interchangeable.

Glossary ยท Pipeline implementation

An observation is not necessarily a measurement

A measurement quantifies a property under a method. Observation models may also represent categorical observations and sampling context. Read the distinctions.