Handling Big Stuff – PlanetServer helps Exploring Mars in Highspeed

written by Heike Hoenig

Frozen and dusty. That is Mars. Not a very hospitable place one would expect, but Mars’ surface shows the opposite in its past: Furrows and grooves in the landscape surprisingly point to a history of flowing rivers, standing lakes and possibly even planetary-scale oceans. On Earth, water means life – so was there life on Mars? Answering such questions is a long way ahead, but the first tool to look for answers is data: Researchers all over the world spend their time exploring the red planet, analysing satellite imagery, diving in huge data amounts, exploring meticulously its topography. The group of researchers working with Dr. Angelo Pio Rossi, Professor of Earth and Planetary Science at Jacobs University Bremen, recently developed a Web GIS called PlanetServer which enables online visualization and analysis of Compact Reconnaissance Imaging Spectrometer (CRISM) hyperspectral data from Mars. Such research helps understanding the processes which once have shaped the red planet, and maybe even continue shaping it.

„The Martians“: Prof. Dr. Angelo Pio Rossi (right) and Lead Developer, PhD student Ramiro Marco Figuera (left) at Jacobs University Bremen explore the surfaces of Solar System bodies.

The data we have gave us some understanding of the surface composition which is useful to find out, for example, where ancient signs of water on Mars are, Rossi says, but this road to knowledge is a long one.

His PlanetServer science portal builds upon Solar System space missions: data collected by spacecrafts from Mars and Moon, and more in the future. „Actually, while these data are available they are not always processed and calibrated“. Typically, such a preparatory workflow includes a challenging multitude of steps: searching, downloading, and processing data to the highest level available, reduce it, and, as a last step, combine the files into seamless thematic maps or even spatio-temporal datacubes. Results show Rossi that the team is on the right way: „Our goal, namely finding a more convenient way of processing, delivering and analysing massive planetary datasets, is successful with PlanetServer.“

PlanetServer, initiated within the EC 7th Framework Programme project EarthServer, is being enhanced as part of the EC H2020 project EarthServer-2. Until now Rossi and his team have integrated 25,000 files and heading towards an overall volume of 30+ TB from the NASA Planetary Data System (PDS) and the ESA Planetary Science Archive (PSA).
PlanetServer is based on an infrastructure allowing direct high-speed analysis of massive Planetary Science datasets: Big Data Server rasdaman is pioneer and world technology leader in flexible, scalable, and multi-dimensional array analytics. By using unique adaptive data partitioning and query processing techniques, data cubes are analysed and combined in an easy-to-use manner.
„The queries are done directly in the web browser, it really cannot get much easier than this. PlanetServer looks like a website, but it is an application. Each click is like querying the database. In this way, researchers can easily identify and characterise different surface geological features on Mars.

PlanetServer web client displaying Mars data: NASA MRO CRISM hyperspectral data draped on NASA MGS MOLA topography in an oblique perspective view (vertically exaggerated). The point indicates the location of the extracted spectrum through a WCPS query.

The client, which leverages the NASA Web World Wind virtual globe toolkit, provides various options for selecting and analysing topographical areas individually. „Our technology is absolutely unique. Other services offer data access as well, but users don´t get a visualization in combination with analysing the data simultaneously.“

PlanetServer web client displaying footprints of data coverage (NASA MRO CRISM) and multiple cubes visualised in red-green-blue combination of selected hyperspectral bands among the hundreds contained in each coverage/cube.

PlanetServer allows choosing different map projections such as 3D, Mercator, Equirectangular; North and South Polar North and South UPS, North and South Gnomonic. By centering on a position and selecting a „footprint“, the planetary service provides a matured functionality for searching a place on Mars. Searching by location, image ID and coordinates, users easily find special locations such as Gale Crater, Valles Marineris, Olympus Mons, or Gusev Crater. Compositional variability can be retrieved easily, highlighted by user-defined colours. The RGB combinator allows users to select a list of bands and assigning them to a certain colour channel (RGB). PlanetServer also provides summary image products that can be also loaded on a certain channel (R, G, or B). Once the channels are selected users can run the query, and the output image will be shown on the globe. In order to analyse individual or multiple spectra PlanetServer offers plotting functionalities. The user only needs to click on the location inside the image, and the spectra will be displayed in the plot dock/widget. When analysing spectra common practice is ratioing spectra by enhancing possible absorption bands. PlanetServer allows the selection of both numerator and denominator pixels, computing ratioed spectra on the plotting widget. Spectral library items can also be loaded on the same widget and visually compared.

„There is so much functionality you can use, even the area you are going to investigate is gigantic. If we have a look on Valles Marineris - the biggest canyon of Mars - we explore a massive area: 4,000 km long, 200 km wide and up to 7 km deep. It is one of the largest canyons of the Solar System. Using PlanetServer you can immerse into a large tectonic "crack" in the Martian crust. That´s big stuff.“

CRISM cubes with the very same colour combination (RGB) highlighting the variability of surface units. The lower-resolution NASA Viking colour mosaic (courtesy USGS) provides the background imagery (served as WMS)

PlanetServer strictly commits itself to open standards. In particular, the OGC Web Coverage Processing Service (WCPS) standard adds unprecedented flexibility through its image query language: users select functionality via point-and-click, the Web browser creates queries without users even seeing it, and the server processes those queries close to the data. Users just see the final result. Queries also help scientists reconstructing exactly what they or other researchers have done: all results, in particular: all screenshots, are reproducible – it simply means re-running the queries. “Currently there is no technology which can compete with rasdaman and WCPS. The fact that you can dynamically query, download, and extract an image or a spectrum is unique“, summarizes Rossi. “What we do can be re-used and modified or extended by others, as all is publicly released and openly available, both tools and their source code”, adds Rossi.

PlanetServer aims at supporting collaborative data analysis, enabling scientists to share queries, results, over the web, both through the use of its 3d-client powered by NASA WebWorldWind and via more custom Jupyter notebooks, as well as interfacing with the Planetary Virtual Observatory of EuroPlanet VESPA (http://europlanet-vespa.eu). All of the server and client code of PlanetServer is available on GitHub (https://github.com/planetserver)

Multiple views of PlanetServer web client with different background and data layers for Mars (NASA Viking mosaic, NASA MGS MOLA color-coded topography) and the Moon (NASA LRO LROC mosaic).