SIParCS 2019 - Colton Grainger

Colton Grainger

Colton Grainger, University of Colorado Boulder

Towards Categorical Metadata for Unreduced Climate Observations

Recorded Talk

In this talk, I outline an initial implementation of the Research Data Archive Images Module (`rdaim`).

`rdaim` was developed to address three issues: First, the ACRE (Atmospheric Circulation Reconstructions over the Earth) initiative has produced a *decentralized ~100Tb collection of images* of historical climate documents. Second, the state-of-the-art method for reducing each ~8Mb image to a ~2Kb observation requires *15 minutes of transcription from 2 humans*. Last, because there is no common description framework for images, there is a *fragile link* between a parent image and its child observations.

Therefore, this talk focuses on how `rdaim` is able to accomplish 3 tasks: gathering images into a single repository, providing programmatic access to individual images, and establishing a common image description framework.

Mentors: Thomas Cram, Matt Mayernik, Steve Worley