Creating a long term trend data set from multiple data sets

Bold textCreating a long term trend data set from multiple data sets Summary: Input datasets are gathered to produce a new data set. Inputs cover a long time duration and come from multiple sources. The Producers of inputs are no longer available for consultation. Provenance of inputs are essential to have been captured for this use case to succeed. Reciprocally, for future users of the derived long-term dataset, provenance will need to be captured.

Objective/Context: Create dataset from multiple previously produced datasets to enable long-term use. Assume producers of datasets are no longer available for consultation.

Actors: Producers disconnected from consumers; consumers; archivist/curator; data distributor; science data intermediaries

Sequence of events:

identify purpose of creating long-term dataset
Write proposal
Get funding
Identify input datasets required
Search for (discover) input datasets
Negotiate rights (in case of restricted data)/obtain datasets
Consult with data intermediary
Verify that datasets are applicable for purpose
Designs algorithms
Create software
Generate long term trend dataset
Document lessons learned
Document provenance

PCCS artifacts: For each input dataset, need to decide which artifacts need to be preserved to satisfy the use case. Artifacts also need to be preserved for the long term derived dataset generated by the use case.

the data itself
processing version
source code *
Format
Size of data
parameter descriptions
content descriptions
tools and web apps needed to read/use/transform the data
reputation of source
calibration method
processing method
algorithms used *
difference from previous versions *
data version
processing history *
history of what has happened to the data since it was created *
data inputs used *
pre-cursor products used *
instrument schematics, etc. *
instrument capabilities and characteristics
calibration/validation data *
validation method *