Francesco Casu [email protected] - Benvenuto su A.S.I. · PDF fileFrancesco Casu...
Transcript of Francesco Casu [email protected] - Benvenuto su A.S.I. · PDF fileFrancesco Casu...
Outline The SBAS‐DInSAR algorithm SBAS application scenarios The ESA Supersites Exploitation Platform (SSEP) project SBAS‐DInSAR in a parallel computational environment
‐ Envisat results‐ COSMO‐SkyMed results
Conclusions
The SBAS‐DInSAR algorithmExploiting the existing huge SAR data archives, we can generate Earth surfacedisplacement time‐series.
Interferograms
SBAS Application Scenario
Earthquakes
Volcanoes
Water Resources
SAR data set
Sequential SBASProcessing
(could be extremelytime consuming!!!)
Support GEO to better understand the geophysical processes causing Geohazards (earthquakes and volcanoes)
Global partnership of scientists,satellite and in‐situ data providers(multi‐sensor DInSAR, seismic, GPS ‐complete data sets)
Brings together community &relevant data
Support national authorities andpolicy makers in risk assessment andmitigation strategies for Geohazards
The Geohazard Supersites
Past, present and futureSAR Satellite Constellations
Time
swath width: ≈ 100 kmrevisit time: ≈ monthly
swath width: ≈ 40 kmrevisit time: 4 ‐ 11 days
swath width: ≈ 250 kmrevisit time: 6‐12 days
Too much data Too many users
EO Scenario
Processing is a bottleneck:Sequential Scientific algorithms need to be
updated to deal with the new big data scenario!
Too much data Too many users
Cloud platforms exploitation
“Infinite” processing power at affordable costs
ESA Supersites Exploitation Platform (SSEP)
Universities Research Centers
SMEs
Crowd sourcing
high speednetworks
SSEP in details
DataArchive
EOSoftware
ComputingResources
Output / Results
• ESA GPOD (Grid processing on demand)• Helix Nebula (Data/Cloud/Services)InSAR/PSI codes
Scietific• ESA NEST• CNR SBAS• JPL ROI_PAC• SCRIPPS
GMTSAR
Commercial• GAMMA
ESA SAR virtual archive:60000+ ERS / Envisat SAR products
• Land Motion• Coherence• Interferograms
>1<‐1cm/year
SBAS algorithm in a parallel environmentNapoli Bay area: ENVISAT processing
Time interval: 2002‐2010
Sequentialmode
Area 100x100 km
#Images 64
#Interf 195
#CPU 1
RAM/CPU
8 GB
TIME ≈ 5 days
Sequentialmode
Parallelmode
Area 100x100 km 100x100 km
#Images 64 64
#Interf 195 195
#CPU 1 16
RAM/CPU
8 GB 8 GB
TIME ≈ 5 days ≈ 0.6 days
Time interval: 2009‐2013
Sequentialmode
Parallelmode
Area 30x30 km2
#Images 255
#Pixel 10000x18000
#Interf 750
#Measur. points
768376
#CPU 1 16
RAM/CPU
32 GB 32 GB
TIME > 100 days ≈ 7 days
Napoli Bay area: COSMO‐SkyMed processingSBAS algorithm in a parallel environment
Conclusions EO data are becoming to be “Big” (if not yet). Effective and
efficient tools are needed for dealing and analysing such Bigdata
The SBAS‐DInSAR is one of these tools: it deals with a largeamount of data and can be shared in a commonenvironment for science
Contribution to SSEP ESA project to create a scientificEcosystem for putting together and making accessible past(ERS‐ENVISAT) and future (Sentinel) satellite data, scientifictools (as SBAS‐DInSAR), results, publications, …
SSEP and forthcoming ESA Thematic Exploitation Platforms(TEP) could represent an interesting “enhancement” ofGround Segment within the Big Data paradigm