Alternative Method for Data Fusion of NASA CERES and A-TRAIN datasets: An Evaluation of Triplestore

The synergism of weather and research satellite sensors with different spectral, spatial, and temporal resolution have created a need to combine multi-sensor datasets. In climate research, co-location of observations in time and space can be difficult especially with the variety of satellites with different orbits, sensor viewing geometry, and data formats. A new method for data fusion of geospatial data is being evaluated. We attempt to merge and integrate data products from the NASA Clouds and the Earth's Radiant Energy System (CERES) and A-Train satellites. CERES cloud products, top-of-the-atmosphere (TOA) radiative energy fluxes, and Single Scanner Footprint (SSF) observations are mapped into Resource Description Framework (RDF) triples consisting of subject, predicate, and object that represents each observation point geographically by the 4-D dimensions: latitude, longitude, altitude and time. These triples are stored into a massive triplestore database where SPARQL queries can then be use to retrieve the data. Observations can be stored in nominal resolution without the need to average or remap. Thus, this enables model or observational atmospheric properties from both satellites and surface data to be retrieved for a given spatial and temporal conditions. Flux profiles can then be computed and validated with observational fluxes from ground sites to provide closure of radiative transfer algorithms. MODIS, CloudSat, and CALIPSO cloud properties will also be integrated into this system allowing for validation with CERES fluxes. This paper will discuss benefits and applications of using Triplestore to store and retrieve Atmospheric Radiation datasets.