Hyperspectral atmospheric data, such as that generated by aircraft and satellite-mounted high spectral resolution infrared sounders and imagers, presents many challenges in its processing, storage and retrieval. Unlike many other scientific datasets, atmospheric data is always changing in time and successive observations of the same geographical region simply add to the volume of new data. An added complication is that instruments may generate point spectra (such as up-looking ground-based stations), swaths (cross-track scanning aircraft and polar orbiting satellite instruments) and two dimensional maps (geosynchronous satellite instruments) of various temporal, spatial and spectral resolutions.

Design studies for metadata schema are presented with intent to manage these complexities, and to fulfill additional requirements, such as those imposed by reprocessing existing observational data with constantly improved algorithm pipelines. We also consider the challenges presented by queries for spatially and temporally collocated datasets from multiple instruments, and by the desire to separate the metadata design from details of underlying storage methodologies.