Systematizing the Record of Earth's Shapes and Colors: A Framework for Data and Metadata Models

Recent standards have provided a wide range of choices for representing large and complex arrays of environmental remote sensing data. The standards offer many features for the data arrays themselves and descriptive metadata. Current sensors for applications such as multispectral and conical microwave imaging, or limb and vertical sounding produce complex data sets that can take good advantage of data standard features. To accommodate data set complexity, producers, users and archivists create profiles for specific data sets within a standard. Depending on the data architect's view of the system, these profiles can represent the same underlying data set in any of several different ways.

This paper introduces a consistent framework within which data profiles can be created. The work is motivated by the large number and wide variety of raw, sensor, and environmental data sets to be produced by the NPOESS polar environmental satellites. The goal is first to make clear and precise the relationship between digital measurements, corrected estimates, and modeled inferences; and second to make clear and precise the relationship between primary independent variables (such as time), associate independent variables (such as observed location), and dependent variables (the measurements).

The benefits should include more easily understood data sets, clarity in defining data and metadata structures, reduced duplication in software development effort, and consistent metadata application—especially within a large project. Scientists, sensor engineers, and data processors will all be able to grasp the relationships. Without attempting to force a single structure on all data sets, the framework may provide a simplifying way to view complex data sets, and make the features of powerful data and metadata standards more accessible.