New Approaches to the Use and Integration of MultiSensor Remote Sensing for Historic Resource Identi

1
New Approaches to the Use and Integration of
Multi-Sensor Remote Sensing for Historic Resource
Identification and Evaluation Project Number
CS-1263 W. Fredrick Limp Center for Advanced
Spatial Technologies University of Arkansas,
Fayetteville In Progress Review Meeting April
2003
2
Performers

• Dr. W. Fredrick Limp
• Center for Advanced Spatial
• Technologies
• University of Arkansas
• Multi-resolution object fusion methods,
  geospatial IT, project administration
• Dr. Michael J. Hargrave
• US Army Corps of Engineers
• Construction Engineering
• Research Laboratory
• Geophysical archeological site assessments, DoD
  DoE coordination
• Dr. Kenneth L. Kvamme
• ArcheoImaging Lab
• University of Arkansas
• Ground-based multi-sensor geophysical
  explorations of archeological sites, numerical
  fusion methods

• Dr. Thomas L. Sever
• NASA
• Marshall Space Flight Center
• Space-based, aircraft and terrestrial optical and
  thermal remote sensing
• Dr. Lewis E. Somers
• Geoscan Research
• Archeological geophysics engineering, and GPR
  analysis
• Mr. Richard Toliver
• SRI International
• Fusion software development,
• system development and testing

3
Technical Objective

• Exploration and assessments of the application of
  innovative fusion methods to a suite of ground,
  aerial and space-based sensor data for the
• detection and identification of surface and
  subsurface archeological features
• identification of best modality and modality
  mixes
• development of COTS based multi-instrument
  fusion software capabilities

4
Technical Background

• Three classes of materials compose an
  archeological site.

• Artifacts
• Material objects modified by humans.
• Portable, small items easily move, e.g.
  arrowheads, pots, knives. Not readily detected by
  remote sensing.
• Non-portable artifacts, e.g. cut posts, building
  timbers, shaped stones and bricks used in
  architectural constructions.
• Structural features and their remains
• Human constructions, buildings, houses, shacks,
  storage facilities, public structures, exterior
  hearths, subterranean storage pits, wells,
  fortification ditches. The principal targets of
  archeological remote sensing.
• Sediments and soils
• Deposits in which artifacts and structural
  features lie.
• Altered by burning, trash disposal, etc.

5
Technical Approach
Project Initiation
Analysis Assessment Software Enhancement
Assessment matrix Instrument/condition Instrument
combo Modality worth Software Publications

Aerial Satellite Data Acquisition
Archeological Field Testing
Initiate Preliminary Analysis Using Existing Datas
ets
Limited additional geophysical acquisition
Field Work - Geophysical Data Acquisition
Data Analysis Fusion Software Development
Raw Data Pre-processing
Design and Populate GIS Databases
6
Data Acquisition at Project Initiation
(geophysical, thermal, optical)
Technical Approach
A Acquire, E Existing NA Conditions not
appropriate All proposed efforts have been
coordinated w/ installations by Hargrave at
USACE/CERL
7
Accomplishments

• Project Award - September 30, 2002
• T1. Installation Coordination
• Managed by Dr. Michael Hargrave, CERL
• Contact with all installations completed
• Ongoing contacts maintained
• T2. Acquire Aerial and Satellite Data
• NASA partner Dr. Tom Sever ordered from
  DigitalGlobe
• Late-March through early-April 2003 acquisition
• Possible scheduling slip w/ DigitalGlobe
• due to demands of other high priority
  acquisitions the normal priority tasking of
  satellite acquisitions may (are) be delayed
• Light aircraft acquisition over Ft. Riley (April
  2003)
• Duncan MS4100 Multispectral
• Palm IR 250 Thermal

8
Accomplishments (2)

• T3. Geophysical Fieldwork and T4. Preprocess Data
  Sets
• Ft. Riley (Army City)
• July 8 - 18 2002 under pre-award authorization
• Ft. Benning (Kasita)
• November 28 - December 6, 2002
• Savannah River (Silver Bluff)
• November 18 - 27, 2002
• All geophysical work completed
• Very preliminary results have been provided to
  installations for their use
• Preprocessing largely completed on Riley
• Ft. Bliss
• Scheduled later in 2003 due to better soil/site
  conditions
• Following late summer rainfall and lower
  temperatures

9
Accomplishments (3)

• T5. Establish GIS Databases
• Whistling Elk and Mt. Comfort sites completed
• Others underway
• T6. Data Analysis Fusion
• First milestones scheduled for completion later
  in 2003
• Work on schedule

10
Accomplishments (4)

• T7. Technology Transfer
• Scheduled as a Year 3 task
• To date presentations at Society for Historic
• Archaeology, Plains Conference, Society for
• American Archaeology, Arkansas Academy of
  Sciences
• T8. Archeological Investigation
• Year 3 Task/milestones
• T9. Conyers Consulting
• Conyers is performing CS-1261 Radar Applications
• Underway
• Meetings/discussions at Fayetteville, Ft. Riley,
  Washington and New Orleans

11
Ft. Riley Army City
WW I
12
Resistivity
13
Magnetic Susceptibility
14
Magnetic Gradiometry
15
GPR - Radar
16
Then - Now
17
Data Fusion with Principle Components Analysis
Preliminary Data Fusion with Principle
Components Analysis
Component Amostly Magnetometry
Component CConductivity and Magnetic
Susceptibility
Component Bmostly GPR and Resistivity
18
RGB Composite of PCAs
19
Kasita Town Resistivity
Very preliminary data processing
20
Kasita Magnetometry
Very preliminary data processing
21
Kasita Radar
Very preliminary data processing
22
Silver Bluff Resistivity
Very preliminary data processing
23
Silver Bluff Magnetometry
Very preliminary data processing
24
Silver Bluff GPR
Very preliminary data processing
25
Technical Approach
Project Initiation
Analysis Assessment Software Enhancement
Assessment matrix Instrument/condition Instrument
combo Modality worth Software Publications

Aerial Satellite Data Acquisition
Archeological Field Testing
Initiate Preliminary Analysis Using Existing Datas
ets
Limited additional geophysical acquisition
Field Work - Geophysical Data Acquisition
Data Analysis Fusion Software Development
Raw Data Pre-processing
Design and Populate GIS Databases
26
Year 3 Program Plan
(Combines elements of original Year 2 Year
3) Task 3. Geophysical Fieldwork
29k Task 6 . Data Analysis and Fusion
127k Fusion Analysis/Modality Worth
86k Task 7. Technology Transfer 14k Task
8. Archeological Investigations 138k Task 9 .
Conyers Consulting 1.2k Task 10.
Technical reports/conference
16k Total 412K
27
Program Funding
UA CERL NASA K K K FY03 Installation
coordination 6.1 Geophysical fieldwork 6
(Geoscan sub) Thermal data acquisition
4.5 6.1 Preprocessing of data sets 17.7 6
(Geoscan sub) Data analysis fusion
40.8 Conyers consulting 3 Conference
attendance, rpt prep. 16.3 Total
102 FY04 Archeological fieldwork
138.4 Geophysical fieldwork 29.3 Data
analysis fusion 73.3 53.6
(SRI sub) Conyers consulting
1.2 Fusion analysis, modality worth 49.2 36.9 Te
chnology transfer 14 Conference
attendance, rpt prep. 16.4 Total
412
28
Program Plan
Year 3
Task
Year 1
Year 2
Installation coordination Acquire aerial
satellite data Geophysical fieldwork
Preprocess data sets Establish GIS
databases Data analysis fusion Develop
field testing program Archeology field testing
/ validation Project analysis, evaluation,
reporting Conferences and reporting
29
Obligations/Expenditures
30
Transition Plan

• Installation specialists through the USACE-CERL
  report series and individual efforts of Dr.
  Hargrave
• Traditional scholarly publication and meeting
  presentations
• UA SERDP New Approaches Website
• http//www.cast.uark.edu/cast/serdp
• North American Database of Archeological
  Geophysics (NADAG) http//www.cast.uark.edu/nadag/
  .
• On-line system supported by the National Center
  for Preservation Technology and Training
• Central source of surface remote sensing data in
  the U.S.
• Commercial partners
• Geoscan
• SRI
• Use of COTS solutions
• Definiens Imaging
• ESRI

31
Technical Backup
Technical Backup
32
Fusion Methods
Technical Approach

• Evaluate four alternative data fusion methods
• Two traditional
• - Multi-band Visualization
• - Pixel-based Multivariate Statistical Methods
• Principal components
• Supervised classifiers
• Maximum likelihood, minimum distance, logistic
  regression
• Unsupervised classifiers
• Accessible in COTS image processing
• These methods have been used (Kvamme 1999, 2001)
  but they have not been applied to this
  comprehensive suite of data sets and conditions

33
Fusion Methods (2)
Technical Approach

• Two new approaches
• Image segmentation and object classification
• Physical model rule-based fusion
• Involves application of a range of
  pattern/structure recognition approaches to the
  problem of content extraction from multi-source
  data
• Builds on recent advances in COTS
• Definiens Imaging
• SRI, ESRI and others

Object segmentation of Whistling Elk mag. data
34
Army City Technical Specs
35
Kasita Town Technical Specs
36
Silver Bluff Technical Specs
37
Technical Backup
Test Site Selection

• Archeological Cultures and Site Selection
  Factors
• Relatively large, complex sites
• Characterized by a wide range of
  archeological feature types
• Non-archeological Selection Factors
• Characteristics of soil, bedrock, vegetation,
  topography,
• moisture
• Site condition
• Presence of accessible legacy data
• Approval and active collaboration of
  installation

38
Technical Backup

• Large areas (0.5 1 ha ) examined
• needed to realize the pattern and organization
  to a settlements layout
• high spatial sampling densities are required for
  the recognition of many classes of structural
  features.
• Conversion of all geophysical measurements to a
  gridded structure
• a raster data structure is imposed on all data
  collected
• terrestrial instrument grids are accurately
  positioned on the ground
• use of electronic distance measuring instruments
  (EDM) or GPS
• co-registration to sub-pixel correspondence.
• Satellite, aerial and lift-boom (oblique)
  instrument data ortho-corrected to same grid
• Soft-bench photogrammetry (PCI Geomatica) used
  with sensor model and detailed DEM.

39
Reporting Products

• Description of
• ground-based field methods and instrumentation.
• satellite methods and instrumentation.
• remote sensing findings from each study site
  (graphical and metadata).
• data processing methods.
• archaeological field testing methods.
• archaeological testing results.
• data integration/fusion methods.
• field testing research design.
• Interpretation of remote sensing findings from
  each study site.
• Operational data fusion software system.
• Data fusion imagery from each study site.
• Interpretation of integrated data from each study
  site.
• Multidimensional remote sensing accuracy
  assessments.
• Analysis of positive benefits of data integration
  of multiple methods.
• Analysis of data redundancies and cost-benefits
  within each ground, air, or space domain and
  between domains a modality worth assessment.
• Analysis of kinds of archaeological features
  found and missed, false positives and negatives.
• Recommendations of the best subset of
  instrumentation necessary for future work of this
  kind.
• Analysis of environmental characteristics and
  their effect on detection.

40
Technical Backup
41
Technical Backup
Sensor Domain Resolution Depth
Units Format Area
Measure

42
Technical Backup