Software, Databases and Data Quality Management
What this domain tests
This domain tests practical data judgement: selecting representative datasets, documenting metadata, managing versions, and using software without hiding modelling choices.
Core rules and concepts
- Common LCI databases include ecoinvent, Agribalyse, the US LCI database, ELCD, and official Environmental Footprint datasets. Each has different geography, system model, and licensing.
- Common software includes openLCA, SimaPro, GaBi/Sphera, and Umberto. Software automates calculation but does not validate representativeness.
- Data quality indicators cover temporal, geographical, and technological representativeness, precision, completeness, consistency, and reproducibility.
- Pedigree matrices convert qualitative data-quality judgements into uncertainty factors; they do not replace source checks.
- Version control is essential: database version, method version, software version, and any modified factors must be recorded.
Worked example
A study models electricity for a factory in Germany in 2022. A global average dataset is easy to use but fails geographical and technological representativeness. A German low-voltage market dataset for 2022, including imports, is better. The modeller records the database version, voltage level, year, and whether guarantees of origin were modelled. If the database is updated later, the result may shift even with no process change, so the version is part of the result's traceability.
Traps that cost marks
- Using a global dataset for a site-specific process without justification.
- Mixing datasets from different database versions in one model.
- Assuming a free database is automatically representative.
- Reporting a software name instead of documenting the dataset and method.
- Treating pedigree scores as measured uncertainty.