Post on 16-Feb-2017
SmartGeo/Eiagrid portal
SmartGeo
Gestore di Risorse Condivise per Analisi di Dati e Applicazioni Ambientali
In-field optimization of seismic data acquisition by real-time subsurface
imaging using a remote GRID computing
environment.
Gestore di Risorse Condivise per Analisi di Dati e Applicazioni Ambientali
Overview
The Grida3 project
The basic concept of EIAGRID
The EIAGRID portal
Data Examples
Conclusions
Aggregating Infrastructure for Environmental Solutions
The great environmental challenges make necessary expertise from different disciplines and a strong integration of the activities, based on
Advanced information and communication technologies New problem solving paradigms
Organizations of federated entities (i.e. peripheral organs) with common interests and objectives in the management and monitoring of the environment and the territory must be in condition to set up shared knowledge-based systems and provide high value-added activities leading to
Industrial products Advanced services
tuned to implement environmental quality systems.
Real Collaborations and Virtual Organizations
Working Group 2: monitoring, planning and sustainable water resource management
Working Group 3: information systems for the analysis of integrated environmental and territorial data
Working Group 1: short term prediction of extreme events
A Grid is an infrastructure that allows the integrated and collaborative use of virtualized resources § Data servers § Computational servers § Connecting networks § Numerical applications § Information systems owned and managed by one or more organizations
The virtual organization acts as the Grid provider while each partner becomes the recipient of the Grid services
Grida3, Gestore di Risorse Condivise per Analisi di Dati e Applicazioni Ambientali (MIUR Prog. N. 1433/2006)
A problem-solving grid platform for the integration, through a computing portal, of § resources for
§ communication § computation § data storage § visualization
§ simulation software § instrumentation § human know-how in Environmental Sciences
TECHNOLOGIES
Infrastructure User Interfaces Secure access
APPLICATIONS
GIS
Meteorology
Hydrology
Site Remediation
Geophysical Imaging
Numerical applications
Grid computing somehow appears as natural extension of distributed parallel computing
Code porting and new products development are focused to the domain of advanced SW technology for the intensive use of geographically distributed resources
With Grida3 applications, no disruption of algorithmic or mathematical nature is expected, at least at the moment
Data-dominated applications
Data acquisition costs are tremendously reduced
Environmental sciences are characterized by very large volumes of heterogeneous data generated by modern recording digital apparatus
Datasets are not always usable by traditional prediction solvers (PDEs and ODEs) without a preliminary conceptualization phase
Use of clever metadata, a describer designed to infer relationships within data collections and validate new model hypothesis
Data conceptualization
Need to develop SW tools for empirical analysis in environmental sciences based on geographical information systems, Integrating qualitative and quantitative data
Grouping data by relationships that may not be clearly visible among collections of field data
Looking for evidence that would support or refute the implication of a theory
Problems consequently arise at the level of data collection, retrieval and analysis
Most of the non-trivial knowledge extraction is based on heuristics operating on distributed databanks:
Data mining discovery - Ontology-based knowledge representations
Groundwater: Modeling, Management and Planning
Groundwater application § GIS (input&output) § Pre-processing § Solver and Optimizer § Post-processing § Visualization
Application developer
Site 1
Data grid infrastructure SRB/iRODS
Compute grid infrastructure via Genius/EnginFrame
Site 2
Data integration and problem solving § WEB collaborative environment § Customizable analysis tools § Problem solving driven by physical models § Web GIS (solver output, field data, maps…) § Decision Support System (DSS)
Environmental engineer
Groundwater: Modeling, Management and Planning
Site 3
Data grid infrastructure SRB/iRODS
Compute grid infrastructure via Genius/EnginFrame
Environmental manager
Collaborative problem-solving platform as a decision support system § Interactive simulation tools based on physics § Web GIS environment for data
§ Storage § Retrieval § Rendering
§ Analysis and decision instruments for § Management § Planning § Costs evaluation
§ Editing of results and dissemination
Grida3, Shared Resources Manager for Environmental Data Analysis and Applications
The Grida3 portal aims at supporting problem solving and decision making by integrating
resources for communication computation data storage
software for simulation inversion visualization
and human know how into a grid computing platform for Environmental Sciences
TECHNOLOGIES
Infrastructure User Interfaces Secure access
APPLICATIONS
GIS Tools
Meteorology
Hydrology
Site Remediation
Geophysical Imaging EIAGRID
Service
Gestore di Risorse Condivise per Analisi di Dati e Applicazioni Ambientali
Creating a grid computing environment for in-field QC and Optimization of SR/GPR data acquisition by:
1. Providing a web-browser-based user interface easily accessible from the field
2. On-the-fly processing of the seismic field data using a remote GRID environment
3. Fast optimization of data analysis and imaging parameters by parallel processing of alternative workflows
The EIAGRID Portal Main Objectives
Gestore di Risorse Condivise per Analisi di Dati e Applicazioni Ambientali
Creating a data grid environment to facilitate analysis & decision making in integrated multi-disciplinary studies by:
1. Providing a flexible and customizable data grid management architecture using iRODS
2. Georeferencing the data using Geo I n f o r m a t i o n S y s t e m ( G I S ) technologies
3. Interconnecting the different types of data by mesh-generators and data crossing techniques
The EIAGRID Portal Main Objectives
Gestore di Risorse Condivise per Analisi di Dati e Applicazioni Ambientali
Seismic reflection data acquisition
GPR data
Multi-offset GPR data: Aim: monitoring of water content and water conductivity Target depth: 0 - 5 m 2D line: length 55 m RAMAC/GPR CU II with MC4 +
4 unshielded 200 Hz antennas Number of sources: 546 Source spacing: 0.1m Number of receivers: 28 Receiver spacing: 0.2 m Maximum offset: 0.6 m
Gestore di Risorse Condivise per Analisi di Dati e Applicazioni Ambientali
SR/GPR data: Fields of application
Environmental engineering: Ø Detection of problematic solid-waste in dumping grounds
Ø Control of the topography of the impermeable basement
Ø Characterization of landslides on slopes proximal to the ground rupture
Seismic and geotechnical engineering:
Ø Evaluation of the seismic local response
Hydrogeology: Ø Identification of aquifer boundaries Ø Estimation of hydrological parameters (porosity, fluid
content, etc.)
Near-surface geophysics
• Near-surface geophysics is the use of geophysical methods to investigate small-scale features in the shallow (tens of meters) subsurface.
• It is closely related to exploration geophysics. • Methods used include seismic refraction and reflection,
gravity, magnetic, electric, and electromagnetic methods.
• These methods are used for archaeology, environmental science, forensic science, military intelligence, geotechnical investigation, and hydrogeology.
• near-surface geophysics includes the study of biogeochemical cycles.
Hydrogeophysics
• Hydrogeophysics involves use of geophysical measurements for estimating parameters and monitoring processes that are important to hydrological studies: water resources, contaminant transport, ecological and climate investigations.
• Improved characterization and monitoring using hydrogeophysical techniques can lead to improved management of our natural resources, understanding of natural systems, and remediation of contaminants.
Gestore di Risorse Condivise per Analisi di Dati e Applicazioni Ambientali
Main Steps of SR/GPR Data Processing
PREPROCESSING
GEOMETRICAL PROCESSING
WAVELET PROCESSING
IMAGING Pr
inci
pal e
labo
ratio
n st
eps
Data conversion Geometry setup Trace editing Noise atenuation CMP sorting
Velocity analysis and NMO Residual static CMP stacking
Deconvolution
Seismic migration
Seismic Records Processing Phases Subsurface Image
Input System Output
Gestore di Risorse Condivise per Analisi di Dati e Applicazioni Ambientali
Seismic reflection data processing
Seismic Records
Input System Output
Processing Phases Subsurface Image
Gestore di Risorse Condivise per Analisi di Dati e Applicazioni Ambientali
Conventional velocity analysis Final Processing Results
Gestore di Risorse Condivise per Analisi di Dati e Applicazioni Ambientali
Main Problem of SR/GPR acquisition:
Real-time processing is difficult and cost intensive
Ø Acquisition parameters such as recording time, sampling interval, source strength and receiver
spacing cannot be optimized in the field
Solution: Wireless data
transmission + remote GRID computing facilities
Gestore di Risorse Condivise per Analisi di Dati e Applicazioni Ambientali
Main Concept
25
EIAGRID Portal
1. Web-browser-based interface accessible from the field
2. Real-time processing using distant HPC resources
3. Remote collaboration and acquisition controlling
Features
26
EIAGRID Portal
1. Web-browser-based interface accessible from the field
2. On-the-fly processing using a distant HPC resources
3. Remote collaboration and acquisition controlling
Features
1. Project, data and user management
2. Simplistic toolbox for data visualization and manipulation
3. Data-driven imaging method suited for parallel computing
Components
• Riduzione dei tempi di Elaborazione
• Ottimizzazione delle fasi di acquisizione
Obiettivi:
Abbattere i costi e trasformare la Sismica a Riflessione in una tecnica di
indagine appetibile e di Routine in campo ambientale e ingegneristico
• CRS-Stack per la Sismica Superficiale • Analisi di velocità completamente automatica • Trasferimento dei dati sismici, via cellulare, dal sito al centro di calcolo; • Elaborazione in tempo reale; • trasferimento sezione sismica dal centro di calcolo al sito per ottimizzazione acquisizione
Implementazioni:
Gestore di Risorse Condivise per Analisi di Dati e Applicazioni Ambientali
Remote Grid Computing
Preprocessing and Visualization using SU: Ø Basic preprocessing steps can be applied without installing
the complex SU processing package.
Ø Data-driven CRS imaging technology---state-of-the-art in oil exploration---enables highly automated data processing.
Imaging and RSC using CRS technology:
ü GRID deployment using high performance computing facilities provides the necessary computing power.
Parallel processing of different Processing workflows: Ø Cumbersome sequential optimization of processing
workflow and processing parameters speeds up drastically.
Gestore di Risorse Condivise per Analisi di Dati e Applicazioni Ambientali
Seismic reflection data processing
Seismic Records
Input System Output
Processing Phases Subsurface Image
ZERO-OFFSET AND CMP METHODS
• The simplest type of acquisition would be to use a single coincident source and receiver pair and profile the earth along a line
• Such an experiment would be called a zero-offset experiment because there is no offset distance between source and receiver
• The resulting seismic data will be single-fold because there will only be a single trace per sub-surface position.
• The zero-offset concept is an important one and the method might be used in practise if noise could be ignored.
ZERO-OFFSET AND CMP METHODS
• In order to overcome the noise problem and to estimate earth velocity, the method of acquisition most commonly used is the Common-Mid-Point (CMP) method.
• The general idea is to acquire a series of traces (gather) which reflect from the same common subsurface mid-point.
• The traces are then summed (stacked) so that superior signal-to-noise ratio to that of the single-fold stack results.
• The fold of the stack is determined by the number of traces in the CMP gather.
ZERO-OFFSET AND CMP METHODS
• Traces resulting from a single six-fold CMP gather depicting reflections from a single flat interface
• The reflection from the flat interface produces a curved series of arrivals on the seismic traces since it takes longer to travel to the far offsets than the near offsets.
• This hyperbolic curve (red line) is called the Normal Moveout curve or NMO and is related to travel time, offset and velocity
• Before stacking the NMO curve must be corrected such that the seismic event lines up on the gather. Normal Moveout Correction.
• The results are shown in the central portion of the figure. The moveout corrected traces are then stacked, to produce the 6-fold stack trace, which simulates the zero-offset response but with increased signal-to-noise ratio.
• The CMP gather provides information about seismic velocity of propagation
CMP in practice
• CMP acquisition is accomplished by firing the source into many receivers simultaneously
• (a) a shot gather where a single shot (red) is fired into six receivers (green). A receiver is also co-located with the shot to produce a zero-offset trace. By moving the source position an appropriate multiple of the receiver spacing CMP gathers can be constructed by re-ordering the shot traces (this process is called sorting).
• (b) shows the original shot and second shot (traces in red). In this case, the shot has moved up a distance equal to the receiver spacing. The CMP spacing is equal to half the receiver spacing.
• (c) shows how the fold of the CMP gathers is starting to build up after six shots have been fired. At the beginning of the line the fold builds up to it's maximum of three. The fold stays at the maximum until the end of the line is reached where the fold decreases.
EFFECT OF DIPPING HORIZONS
• The CMP method holds for multiple layers and the data can be moved out and stacked to produce three reflections.
• Where dip is present it is clear that the CMP method is breaking down since the traces do not all reflect from the same mid-point location. Processing techniques such as DMO and Migration are required to accurately process CMP data acquired from dipping strata.
Common-Reflection-Surface stack
• (generalized) stacking velocity analysis – search for stacking operator fitting best actual reflection event – based on coherence analysis
• data-driven stacking with CRP trajectories – fitting a space curve to a traveltime surface highly ambiguous, hardly applicable
• solution: • consider entire reflector segments
– i. e., consider neighboring CRPs – i. e., consider local curvature of reflector – fitting spatial operator to traveltime surface – three stacking parameters
Overlap of CMP illuminations
Overlap of CMP illuminations
Overlap of CMP illuminations
Overlap of CMP illuminations
Overlap of CMP illuminations
Overlap of CMP illuminations
Basic idea
Observations: – conventional stack implicitly relies on reflector continuity (this also applies to NMO + DMO correction) – based on normal rays for offset zero – we have band-limited data
• Fresnel zone concept Consequences: If conventional stack works
– there are neighboring reflection points – they physically contribute to the wavefield at a considered CMP
Why shouldn’t we incorporate these neighboring reflection points?
CRS stack
• Features inherited from conventional stack: – normal ray concept – assumption of reflector continuity – analytical traveltime approximation (2nd order) – coherence analysis yields stacking parameters – stack yields simulated zero-offset section
• Additional features: – incorporates neighboring CMP gathers – yields additional stacking parameters – increases the coverage – improves reflector continuity and S/N ratio
From CMP to CRS stacking:
Figure taken from Perroud and Tygel 2005. NMO velocity analysis for the CMP at position x = 10 m.
45
Conventional CMP-by-CMP velocity analysis:
Fig.: Mann et al. 2007
CRS-STACK 3D Lo stacking consente di comprimere i dati, con aumento S/N,
simulando una acquisizione zero-offset (source-receiver)
L’idea del CRS STACK e’ descrivere la propagazione d’onda mediante una geometria locale (ottica parassiale): raggi + fronti d’onda paraboloide-ellittico
Massimizzazione di una funzione peso: Semblance ( )hHKHhmHKHmmw T
zyNIPzyTT
zyNzyTT ++⎟⎟
⎠
⎞⎜⎜⎝
⎛+=
0
0
2
00
2 22vt
vtthyp
2 parameters ( emergence angle & azimuth )
3 Normal Wavefront parameters ( KN,11; KN,12 ; KN22 )
3 NIP Wavefront parameters ( KNIP,11; KNIP,12 ; KNIP22 )
Definizione del problema
• Problema di ottimizzazione: – Ricerca di 8 parametri per il fit di un’ipersuperficie in
uno spazio pentadimensinale
( )
( )
( )
( )
∑ ∑
∑ ∑+
−=ʹ′ =ʹ′
+
−=ʹ′ =ʹ′⎭⎬⎫
⎩⎨⎧
=2
2
2
2
1
2,
2
1,
Ni
Nii
i
Ni
Nii
i
t
tt
M
iti
t
tt
M
iti
f
fSC
SC = SCmax
Legame fra Semblance e parametri sono i dati sismici
Ricerca velocità NMO
Metodi data-driven: trovare la velocità con algoritmi di ottimizzazione di funzioni di coerenza come la semblance
Basta limitare lo spazio di ricerca: due possibilità
2
222
20nmo
hyp vh
tt += ( )
( )
( )
( )
! !
! !+
"=# =#
+
"=# =#$%&
'()
=2
2
2
2
1
2,
2
1,
Ni
Nii
i
Ni
Nii
i
t
tt
M
iti
t
tt
M
iti
f
fSC
Limiti assoluti uguali per ogni punto della linea d’acquisizione time min max 0.00 1500.0 1900.0 0.10 1540.0 2200.0 0.20 1550.0 2450.0 0.30 1575.0 2800.0 0.50 1600.0 3300.0 0.70 1600.0 3800.0
Limiti relativi ad una mappa di velocità data con un valori diversi per ogni punto time min max 0.00 -30.0 +10.0 0.10 -45.0 +25.0 0.20 -45.0 +30.0 0.30 -10.0 +50.0 0.50 -20.0 +85.0 0.70 -80.0 +100.0
• Possibiltà di ricerche ricorsive
Strategia adottata
Step III Determination of RNIP parameters
From VNMO,min , VNMO,max and γv (Step I)
and ψ,θ (Step II), determination of
RNIP,min, RNIP,max and γNIP
3D ZO Stack one five-parametric
search for ψ, θ, γN,
RN,min and RN,max
Step II
Automatic 3D CMP Stack
one three-parametric search for VNMO,min , VNMO,max and γv.
Step I
3D ZO CRS Stack
Stack using the eight paramenters within the projectet Fresnel Zones using the complete CRS
operator
Step IV
Separare le ricerche dei parametri utilizzando sottodomini dei dati
Soluzione possibile grazie
alla formulazione di ipotesi plausibili
( )hHKHhmHKHmmw TzyNIPzy
TTzyNzy
TT ++⎟⎟⎠
⎞⎜⎜⎝
⎛+=
0
0
2
00
2 22vt
vtthyp
Possibilità di ripetere più volte ogni singolo step
Fig.: Mann et al. 2007
Data-driven stacking parameter determination
Each stacking parameter triple within a given search range defines a hypothetical second-order reflection response. The optimum parameter triple maximizes the coherence between this prediction and the actually measured data.
53
Stacking parameter search:
Pragma'c search: 3 x 1 parameter line search in specific
gathers (Mann et al. 1999) + Cloud = Real-‐5me imaging
One step search:
1 x 3 parameter surface search in prestack data (Garabito et al. 2001)
+ Cloud = High-‐precision imaging
Figs: Mann et al. 2007
CRS stacking result obtained after 4 minutes using 50 CPU
Time domain imaging
Published in: Perroud, H., and Tygel, M., 2005, Velocity estimation by the common-reflection-surface (CRS) method: Using ground-penetrating radar: Geophysics, 70, 1343–1352.
Results GPR data
Gestore di Risorse Condivise per Analisi di Dati e Applicazioni Ambientali
Conclusions SMARTGEO
Ø ...minimizes the software and hardware requirements needed to perform a successful SR/GPR data acquisition.
Ø ...reduces the complexity of data QC and choice of acquisition parameter for less experienced users.
Ø …provides fast and accurate results by using modern imaging technology and high performance computing.
Enables a wider use of SR/GPR surveys in environmental and earth sciences through Grid technologies
Ø … facilitates the creation of an integrated geophysical database for environmental studies.
CRS stack
• Features inherited from conventional stack: – normal ray concept – assumption of reflector continuity – analytical traveltime approximation (2nd order) – coherence analysis yields stacking parameters – stack yields simulated zero-offset section
• Additional features: – incorporates neighboring CMP gathers – yields additional stacking parameters – increases the coverage – improves reflector continuity and S/N ratio
Basic idea
Observations: – conventional stack implicitly relies on reflector continuity (this also applies to NMO + DMO correction) – based on normal rays for offset zero – we have band-limited data
• Fresnel zone concept Consequences: If conventional stack works
– there are neighboring reflection points – they physically contribute to the wavefield at a considered CMP
Why shouldn’t we incorporate these neighboring reflection points?
Basic idea
Observations: – conventional stack implicitly relies on reflector continuity (this also applies to NMO + DMO correction) – based on normal rays for offset zero – we have band-limited data
• Fresnel zone concept Consequences: If conventional stack works
– there are neighboring reflection points – they physically contribute to the wavefield at a considered CMP
Why shouldn’t we incorporate these neighboring reflection points?