Just pondering the changed concept, eg where series was a summary table, it's now the central table.
This is a good opportunity to do a bit of redesign.
This is good:
1) provenance. We can focus on observed property, and the processing. eg Dissolved oxygen, and not voltage
ObservedProperty provenance
- Dissolved oxygen, volts, raw, Source:Some Instrument,
- dissolved oxygen, %, derived from series, dissolved oxygen, volts, raw, Source: using calibration XXXX
DataLoad Provenance
- data load records can be attached to series, and not individual data values.
Processing Provenance
- processing can be attached to series.
2) Allows us to trim a lot of junk from DataValues
- Remove "derived from" since it is part of series information. A series needs to have the info to recompute the series.
- move qualitycontrollevelid to series
- move method, source to series
- move time support, from variables to series
- move offset from data value, and associate with series. (dwv, nov 2009, we need both)
- Move sampleMedium, DataType from variable to series
- move sample tracking into a separate table. Not all values have lab sample codes, associate results to a lab sample, and not vice versa
- move qualifiers to a separate table. Tag values with multiple qualifiers.
Leaves DataValues with Time, DataValue, value accuracy, censorcode
3) Work better for dealing with data value aggregation (day, month,year), aggregation becomes a new series, not a new variable with a new series, with provenance recorded by a processing step.
ODM Queries