Imagerie | INSTITUT DE PHYSIQUE DU GLOBE DE PARIS

Twitter

Aller au compte twitter

Équipes de recherche

Géochimie des enveloppes externes

  Imagerie

 

• PRoSPEr
GEE uses the new computing platform at IPGP called PRoSPEr (Platform of RemOte Sensing for Planetary & Earth suRface). Contrary to the high-performance computing services at IPGP S-CAPAD, these calculations demand high memory accessibility, GPU acceleration and high-capacity storage with a fast data accessibility. Furthermore, the platform is being used for prototyping and development.  Along with the classical geomatics usage, a machine-learning approach is being considered.

The current facility is composed of :

2 Dell Precision class workstation with dual Xeon
Nvidia GPU for GPGPU (Tesla)
>128 Gb of Mem. per board (node)
Running Ubuntu Linux 16.04
2 HP Z class workstation with dual Xeon
Nvidia GPU for GPGPU (Titan)
>128 Gb of Mem. per board (node)
+8 Tb of disk (SSD)
1 under windows
1 under Ubuntu Linux
1 Dell poweredge server class with dual Xeon
10Tb HDD
128 Gb of Memory
Running CentOS 7


The five units, defined as nodes are connected through Gigabit Ethernet network and make altogether Beowulf like cluster. Homogenization will be carried out through 2018 with the additional nodes, among those which will be dedicated to the InSight mission (InSight Geospatial Service).

 

• Drone

UAV/Drone is used by team members for digital surface modeling as well as radiative transfer measurements under a wide range of conditions. This activity is also combined with classical remote-sensings acquisition such planes and satellites. Multiwavelength projects are under investigation.

Examples of cloud resulting from drone pictures:

 

Forest liter at La Sapine watershed (Mont Lozère, France) (A. Lucas)