Measured evidence. Defined limits.
Engineering before claims.
Credible optimisation begins with understanding where a model agrees with measurements—and where uncertainty remains. AridMetrics is under active technical validation, with performance claims dependent on sufficient supporting evidence.
Measured-data validation in progress
Uncertainty · scope · method version
- 01Measurements
Identify the source and its limitations.
- 02Model definition
Record inputs, assumptions and version.
- 03Independent review
Separate supported findings from uncertainty.
A reviewable method
Make the comparison understandable and repeatable.
Validation is more than agreement with a single output. A useful comparison explains what was observed, how the model represents it and which limitations remain. These principles define the intended standard of evidence, not a certification of completed work.
- 01Measured-data comparison
- Compare quantities that represent the same physical observation.
- 02Transparent assumptions
- Document conventions, exclusions, uncertainty and relevant simplifications.
- 03Reproducible analysis
- Preserve inputs, processing choices and evaluation definitions.
- 04Versioned methods
- Keep each conclusion associated with the model that produced it.
What we examine
Validation scope
Relevant comparisons may examine irradiance behaviour, tracker orientation, shading, energy-model response or scenario sensitivity. The available measurements determine which question can be answered; one observed quantity does not automatically qualify the entire model chain.
- Physical correspondence
- Establish what an instrument measures, its placement, timestamp meaning and relationship to the modelled surface or state.
- Evidence coverage
- Consider calibration, uncertainty, operating conditions and missing information. Agreement in one regime should not silently become a claim about unobserved conditions.
Why versions matter
Model versioning
Results from different energy-model versions should not be silently treated as directly comparable. A change to core physics, assumptions or processing can alter a scenario's response and the ranking of alternatives—even when the scenario names stay the same.
A defensible comparison therefore identifies the model version and preserves its evaluation basis. Earlier conclusions remain associated with their original method. A revised method calls for an explicit new assessment; it does not retroactively validate an earlier optimisation result or make every historical result equivalent.
Current status
Measured-data validation is ongoing.
AridMetrics is under active technical development and validation. We publish performance claims only when the supporting evidence is sufficient for the specific claim and its stated scope. This website does not report a completed field-validation result, guaranteed energy gain or certified controller performance.
Where evidence is incomplete, the uncertainty remains part of the conclusion. A software illustration or optimisation objective cannot replace a measurement. Public descriptions of the approach explain how the work is assessed; they are not substitutes for a validated result.
Future technical material
Future technical notes
Technical notes will be published as validation progresses. Future material may describe model assumptions, comparison methods, uncertainty and the scope of an evidence-backed conclusion. It should make clear what was evaluated and what remains outside that evaluation.
No publication, validation summary or case study is announced here before it exists. For now, the modelling approach provides the public technical context, while this page states the evidence principles and the ongoing status of the work.