Materials for working group on integrating heterogeneous data across scales
- have data + meta-data about some quantity
- sometimes direct measurement
- sometimes indirect measurement + transformation / interpretation process
- generally has mixed regions of overlap
- data have different "resolution" + "uncertainty"
- conventional spatial / temporal issues
- also precision generally: variation, bias, complex distribution of results relative "finer" scale measurement
- sometimes well characterized in meta data, sometimes no provenance
- want to create reference mesh from this data
- automatable process
- propagated uncertainty
- synthesized finer scale where none available
- using mesh:
- forecasting
- back-propagation of subsequent modeling reults to re-evaluate data
- identify experiments / observations with maximal returns
- 1D "space": all measures are located on a line + no time
- observations with x + measurement source, and particular z measurements
- measurement sources have distribution shapes / lineal area descriptions (possibly as functions of x / z?)
- consumer supplies grid + desired sample size + whole grid model prior
- process returns meshes of particular z values