Big Data and Data Models for Climate System Energetics

Abstract:

Multi-decade satellite missions, such as the NASA CERES mission designed to place observational constraints on the distribution of reflected solar radiation and emitted thermal radiation, present a significant challenge both in the analysis of heterogeneous Big Data and in data continuity. The NASA CERES EBAF dataset is a part of a broader effort to increase the usability of satellite observational data for the climate modeling community. Issues of accessibility, consistency, and reproducibility are paramount. Here we describe the transformation of CERES measurements from source to high level data products intended for direct use by the climate community. At each stage we examine data storage and processing patterns, metadata and potential challenges in reproducibility.

The spatial distribution of net energy uptake and transport in the climate system, and its evolution over interannual and decadal time scales, is fundamental to the development of Earth system models. The workflow begins with the CERES footprint radiance seen by a polar orbiter, to the conversion of radiance to radiometric fluxes based on scene identification from MODIS and VIIRS imagery, followed by diurnal interpolation through the use of geostationary satellite imagery and eventually to the creation of high level gridded data products, the ultimate being the Energy Balanced and Filled flux product for direct comparison to climate models.

Based on this CERES case study we try to anticipate future questions the may arise in the context of these massive satellite data collections, and what new data models may facilitate future data analysis.