Information Retrieval from Text and Images to Develop Databases for Security Applications: GIS to Mo

1
Information Retrieval from Text and Images to
Develop Databases for Security Applications
GIS to Monitor the Sabotages to the Iraqi
Pipeline Network
ESA EUSC 2006 Image Information Mining for
Security and Intelligence EUSC, 27th-28th
November 2006.
GMOSS Team at EUSC Antonio De la Cruz Marcin
Mieleczwky Daniele Cerra Gracia Joyanes.
2
Objectives and Implications

• Main objective a vulnerability assessment of oil
  installations in Iraq with relevant impact for
  the security of these key infrastructures.
• The large number acts of sabotage to oil
  installations in Iraq indicates that terrorists
  have selected these installations to
• Undermine the Iraqi economy
• Prevent the reconstruction of the country
• Destabilize the peace efforts
• The incidence of these acts sabotage in oil
  terminals (e.g. Basra) is intended at having a
  wide impact on international oil supplies and
  affect prices. They can have an impact on global
  economy (GMES security issue).
• Acts of Sabotage to oil installations is a
  strategy followed by terrorists worldwide.

3
Data collection
4

• Database
• Acts of sabotage to pipelines in Iraq
• More than 300 entries with dates and locations
• Collected from several open sources in Internet
  
• From 2003 to 2006

5
Pipeline mapping from available data and Shape
file attributes
6
Vectorization of the Iraqi pipeline network ..
.from JOG, TPC and Russian Military Maps
7
Oil fire caused by terrorist attack on pipeline,
Northern Iraq
Res. 1200 m
The pipeline segment affected runs from Kirkuk
to the Baija refinery
8
(No Transcript)
9
(No Transcript)
10
MODIS hotspots to georeference sabotage events in
the Iraqi pipelines
11
Automatic Data Reduction/Merging

• A pipeline on fire after a specific sabotage can
  generate a considerable number of MODIS hotspots
  due to
• Spatial factors
• A thermal anomaly on the edges of adjacent pixels
  generates hotspots in the centre of each pixel.
• A large fire after a sabotage could extend to
  several pixels.
• Temporal factor
• A thermal anomaly that persists in time will
  generate MODIS hotspots at every sensor pass
• Therefore, it is not easy to associate a sabotage
  to a single hotspot.
• To solve this problem a VBA (Visual Basic
  for Applications) script was developed.
  

12
VBA Script Features

• The VBA script merged the hotspots generated by
  the same event by
• Converting the hotspot to a polygon representing
  the area covered by the pixel.
• Calculating the spatial and temporal
  intersections of overlapping pixels with maximum
  period of 24 hours of acquisition date.
• The intersection results were merged in a new
  feature representing all MODIS hotspots related
  to the same event. This merged feature stores the
  following data
• Number of MODIS hotspots merged in the feature
• Approximate length of the event in days (pipeline
  fire after the sabotage).
• Historical record of brightness temperature
  (initial, mean and list of temperature values).
• The VBA script was able to reduce the number of
  repetitive MODIS hotspots related to sabotages in
  the pipeline network to 40.

13
HOTSPOTS MERGING ALGORITHM
14
HOTSPOTS MERGING ALGORITHM
Buffer's dimensions based on each pixel's scan
and track values (from 1 to 9 km2)
15
HOTSPOTS MERGING ALGORITHM
3
6
Different events distance 2 days
3
16
HOTSPOTS MERGING ALGORITHM
Hotspots 6 Length in days 3 Event starts /
ends d1/d3 Brigthness Temperatures Initial
320 k Average 305 k
3
6
3
17
Distribution of pipeline attacks correlated with
Modis hotspots during 2003, 2004, 2005 and 2006
18
Modis Hotspots at Baiji Refinery, Iraq
19
Attacks in the Kirkuk to Baiji Pipeline, Iraq
Shapefiles of MODIS hotspots
20
GMOSS Staff Exchange EUSC visit to University of
Basilicata
21
Number and Type of Attacks to the Iraqi Pipeline
Network (June 2003 June 2006)
Georeferenced (Modis hotspots) and
non-georeferenced data Explosion Bombing
attack, grenades, home-made bomb
22
(No Transcript)
23
Hotspots for Security Applications

• Although originally designed for
  environmental purposes, hotspots offer the
  following benefits for security applications
• Provide georeferenciation data for security
  events.
• Allow a chronology of security events.
• Provide a rapid generation of databases for
  security applications.
• Can be integrated with other geospatial
  indicators (nightlights, collateral data, etc)
  for additional information to follow the
  development of a security crisis (e.gLebanon)
• Can be used to improve the selection of H.R.
  imagery (better time/space input), thus
  maximizing the use of this resource.

24
Chronology of Security Events from MODIS Hotspots
Gas flares
25
Chronology of Security Events January 2003
December 2003(1 Week per second)