How the index is built
The Carbon Index is generated by continuously gathering carbon-related information from public and industrial sources, organizing those observations into a common structure, and producing a traceable measure of real-world carbon activity.
Every published figure begins as an observation. CarbonWorks does not create or estimate the underlying data; it collects, qualifies, indexes, and relates measurements produced by authoritative sources.
The purpose of the index is to organize the movement of carbon through energy production, industry, transportation, infrastructure, and the environment into a single measurable system while preserving the provenance of every published value.
Continuous acquisition
CarbonWorks continuously gathers information from multiple external sources.
Each acquired document, dataset, or observation is identified, classified, indexed, and incorporated into the Carbon Index only after it has been qualified for publication.
As additional information becomes available, the index expands without changing the published history of earlier observations.
Observation windows
The observation window is the fundamental unit of the index. A window represents one measured signal, from one source, over one period of time.
Each window records the observation count, the mean measured value, the mean confidence, the confidence-weighted value, and the source records from which it was derived.
Windows remain independent. Measurements from different sources or different signal classes are not merged simply because they occur during the same period.
Signal classes
Every observation belongs to a defined signal class — coal production, petroleum production, atmospheric pressure, industrial operating state.
Each signal class is named according to both the quantity being measured and the organization that produced the data, allowing every published value to be traced back to its original series.
Confidence
Confidence is treated as part of every published measurement. Each observation window carries both the measured value and the confidence associated with that value.
Where confidence is lower, that uncertainty remains visible rather than being hidden inside a single aggregate statistic.
Confidence is therefore part of the published measurement itself.
Provenance
Every published value carries its provenance. CarbonWorks records the originating organization, the source dataset, the observation period, the processing history, the confidence, and the publication history.
Every published figure can therefore be traced back to the observations from which it was derived.
Coverage
Coverage grows continuously as new sources are qualified and incorporated.
Different sources begin at different times, update at different rates, and provide different levels of historical depth. CarbonWorks reports the coverage actually available rather than implying uniform completeness.
Design principles
- Every published value is traceable to its source. - Confidence accompanies every published measurement. - Measurements are not merged across unrelated quantities or sources. - Missing capability is reported explicitly. - Published results are reproducible from recorded provenance.
Continuous development
The Carbon Index is continuously expanded as new carbon-related information becomes available.
The objective is not simply to accumulate documents, but to build an increasingly complete, traceable representation of real-world carbon activity.
As the corpus grows, additional measurements, signal classes, visualizations, and analytical capabilities are added while preserving the provenance and structure of previously published information.