Chapter 11 — Data Acquisition, Management, and Databases¶
11.1 From reading to record¶
Raw sensor output is not information. The data chain — acquisition, validation, storage, and retrieval — determines whether measurements become defensible evidence of performance.
11.2 Reading frequency¶
Frequency is set by:
- Parameter dynamics (pore pressure and movement during a storm change fast; settlement slowly).
- Life stage (commissioning and peak filling need more; steady operation less).
- Risk category (higher consequence → more frequent, often automated).
A minimum frequency is documented in the plan; event-driven increases follow storms, earthquakes, or anomalies.
11.3 Validation and quality control¶
Before data enters the record it should be checked for:
- Plausibility (within range, no wild jumps).
- Continuity (gaps explained, not hidden).
- Calibration drift (compared to manual checks and factory cal).
- Environmental correlation (e.g., temperature effects removed).
Gaps are data too
A missing reading is a finding. Investigate and document gaps rather than silently interpolating. A pattern of gaps in a critical instrument is itself an alarm.
11.4 Databases and retrieval¶
A structured database (not scattered spreadsheets) supports:
- Time-series plots per instrument and parameter.
- Cross-plots (e.g., movement vs. pore pressure).
- Audit trail and access control.
- Long-term continuity as staff and instruments turn over.
11.5 Preservation and handover¶
Monitoring records must outlast any individual or contract. Archive raw and processed data, installation records, and review notes in a form that survives closure and ownership transfer. The record is the facility's memory.
11.6 Key takeaways¶
- Data chain quality determines whether readings become evidence.
- Set frequency by parameter, life stage, and risk.
- Validate plausibility, continuity, and calibration; document gaps.
- Archive for the long term, through closure and ownership change.