Making Sense of 2.5 Million Surface Reflectance Spectra of Mercury from MESSENGER

Kurzfassung

The Mercury Atmospheric and Surface and Composition Spectrometer (MASCS) on the MErcury Surface, Space ENvironment, GEochemistry, and Ranging (MESSENGER) spacecraft has mapped the surface of Mercury on a global basis during its one-year primary orbital mission and the first third of its extended mission, producing ~2.5 million spectra from March 2011 to July 2012. The primary challenge to analyzing this dataset is to cope with its large size. In earlier studies of MASCS data, we combined several approaches, ranging from principal component analysis (PCA) to unsupervised cluster analysis and regridding to global and local fixed grids. Each of those techniques provided insight into spectral variations for different volumes of data, but each was quickly overcome by the growing dataset. The most recent version of our data analysis procedure uses PostgreSQL, a type of database management that controls the creation, integrity, maintenance, and use of a database. It embeds a high-level query language, which greatly simplifies database organization as well as retrieval and presentation of database information. We set up a data pipeline to update automatically the MASCS data, read them from the NASA Planetary Data System format, regrid the data to a common grid length, and store all information in the database. All data are then readily available to any authorized user in our network. We are working on a library to access the data directly from within our analysis software, and some preliminary functions have been implemented. As an example, the calculation of a parameter representing the database takes 2 s even for the full dataset of 2.5 million entries. It is thus straightforward to create and analyze rapidly the data, as for example the distribution of normalized radiance at a fixed wavelength. The new methodology provides facilities for controlling data access, enforcing data integrity, managing concurrency control, and recovering the database after a failure and restoring it from backup files, as well as maintaining database security. As an example, a simple query on the volume of data results in 2,476,048 spectra in 700 ms. Moreover, we use PostGIS to add support for geographic objects in a geographic information system (GIS) and to extend the database language with functions to create and manipulate geographic objects. A typical application is the definition of a large number of regions of interest (ROIs) and the search for all data points falling within each ROI. This facility may be used to extract spectral signatures specific to user-defined geological units in a few seconds and to explore the properties of the data from the different ROIs allowing quick analysis of the spectral characteristics of Mercury. We have successfully tested remote access to the database with a GIS visualization system, and we have created data visualization products that layer camera data and real-time-queried MASCS data.