GRID Technology

St, 2008-02-20 15:57, Josef Šobra
STATUS: Finished

The PECS-GRID was a two year project, started in July 2005, run by the Czech company Iguassu Software Systems, a.s., in close collaboration with the G-POD team in ESRIN. The main objective of the PECS-GRID project was to acquire relevant knowledge and demonstrate effective use of GRID technology for Earth Observation (EO) Synthetic Aperture Radar (SAR) data processing.

The period of PECS-GRID project was divided in two phases:

- The first phase (July 2005 – December 2005) had three objectives: i) to familiarization with the state-of-art of the GRID technology; ii) preliminary selection of the SW tools to be ported to GRID; iii) preparation of the plan of the second production phase.

- The second (production) phase (January 2006 – July 2007) of the project is dedicated to the actual development of SW prototypes. The development of the SW prototypes was finished as scheduled by July 2007. After the stop of the development ISS has continued to provide support for the SW delivered to ESA.

Based on the experience gathered during the initial phase, in agreement with the G-POD team, the production phase consisted of porting of miscellaneous SAR data handling SW packages to ESA G-POD (Grid Processing On-Demand, former GODIS - GRID-On-Demand Infrastructure and Services) and development of related high-level SAR-based functionality that was made available as services to users through the G-POD web portal. The project focused on processing of SAR images from ESA ERS and ENVISAT missions, although JAXA ALOS data were also considered (but they were not the main focus of the project).

The prepared SW package is primarily based on available existing SW (open source, public domain, or ESA in-house) although, often in case of missing functionality or unsatisfactory performance certain data processing tools were implemented as well. The final collection of SW tools scripts and auxiliary data used for processing is packed and distributed as a single package named PECS-GRID Toolbox (PGT). Although the PGT is primarily designed to be used for SAR processing on G-POD platform it does not directly depend on any specific G-POD functionality and can be run on any generic IA-32 GNU/Linux computer (this feature was extensively used for off-Grid testing) and, possibly, in another distributed computing environment.

Two prototype SAR data processing services have been prepared by the PECS-GRID project, in-line with the two major benefits brought by Grid technology to EO data processing, i.e., i) the fast and seamless access to a large bulk of data and ii) its on-demand processing.
 
1) The first service is the Large Scale Mapping (parallelized mosaicking) service which is able to produce fast high-resolution ortho-rectified ASAR mosaics, re-project/re-sample existing mosaics and combine different mosaics to RGB images. This service prototype demonstrated the ability of Grid-based system to process large bulk of EO SAR data in substantially shorter time due to the fast download of the data from on-line storage and distributed processing load over many computers. E.g., a radar mosaic of Europe, which was highlighted on the ESA portal as Image of the Week on 16 March 2007 [ESA web portal], took just few hours for the G-POD team to produce and publish.
Envisat ASAR mosaic of Europe: Preview of a 3 arc seconds (~90m) pixel size Envisat ASAR mosaic. The G¬POD mosaicking service produced this mosaic using 143 ASAR WSM strip line products acquired between January and May 2006 using grid 15 nodes in time of 18 hours. The ASAR WSM productsEnvisat ASAR mosaic of Europe: Preview of a 3 arc seconds (~90m) pixel size Envisat ASAR mosaic. The G-POD mosaicking service produced this mosaic using 143 ASAR WSM strip line products acquired between January and May 2006 using grid 15 nodes in time of 18 hours. The ASAR WSM products were automatically ortho-rectified using the SRTM3 and GTOPO30 (above 60gd. north) DEMs. The seam removal was not applied in this case. The outlines of the political boundaries were overlaid manually for purpose of the presentation.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2) The second service, Multi-Temporal Analysis (change detection), allows the selection of a single “crisis” image and single or multiple historical data for desired area of interest. A multi-temporal RGB composition highlighting the differences between the products acquired before and after the crisis event are produced based on the selected calibrated, co-registered products. This service can be a useful tool reducing the delay between the accessing and processing of the satellite data especially in case of emergency mapping demanded by natural disasters where the time is more of the essence than in other areas of the Earth observation.
Multi-temporal composition: Floods in Bihar, India in July 2007. Multi-temporal composition produced by the G-POD service displayed as a GoogleEarth™ overlay. The image was generated using pair of ASAR WSM images from August 7th, 2007 (crisis) and July 3rd, 2007 (archive). The colouMulti-temporal composition: Floods in Bihar, India in July 2007. Multi-temporal composition produced by the G-POD service displayed as a GoogleEarth™ overlay. The image was generated using pair of ASAR WSM images from August 7th, 2007 (crisis) and July 3rd, 2007 (archive). 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Project Duration: 2005 - 2007
Approved Budget: 187 450 €
Principal Investigator: Martin Pačes
Address: Iguassu Software Systems a.s.
Evropska 61, 160 00 Praha 6

Logo

Česká kosmická kancelář, o.p.s.Prvního pluku 17
186 00, Praha 8
tel. a fax:
+420 224 918 288

info czechspace.cz