QMASS QinetiQ MASSSERO Activities Lessons Learnt

1
QMASSQinetiQ MASS-SER-O ActivitiesLessons Learnt
Presentation to MASS Workshop, ESRIN, 17th March
2005 Wyn Cudlip, QinetiQ wcudlip_at_space.qinetiq.com
Consortium partners QinetiQ Research Systems
International (RSI) CCLRC Rutherford Appleton
Laboratory (NEODC RAL) Nigel Press Associates
(NPA) Comsine Ltd. (CSL)
2
QMASS Overview

• QMASS brings together 5 UK Service providers.
• Each integrating 2 services into the MASS
  Environment.
• Services can be used individually (stand-alone)
• Or integrated into Service Chains
• This presentation will concentrate on Lessons
  Learnt

3
The Consortium Partners

• QinetiQ is a large technology-development company
  and has been providing commercial EO services for
  many years.
• CCLRC (Combined Laboratory of the Research
  Councils) hosts the NERC EO Data Centre (NEODC).
  It is the main repository of EO data in the UK
  and supports UK academic research.
• RSI (Research Systems, Global Services Group
  (Europe)) is a major international provider of
  data visualisation and analysis solutions
  including COTS software (ENVI and IDL).
• NPA (Nigel Press Associates) have been providing
  EO data supply and processing services since 1972
  and bring extensive knowledge of the EO Market
  Place.
• ComSine (CSL) is a UK SME specialising in
  consultancy and software development for Internet
  information systems and services, with a
  particular focus on spatial information
  (including Earth observation data).

4
Service Summary
5
QinetiQ Service 1 (nrt MODIS)
System allows inspection of browse and then
ordering of high-resolution sub-sets through an
OGC Compliant Web Coverage Server (WCS).
6
nrt MODIS through Portal
7
MODIS Present Operation
8
QinetiQ Service 2 (Feature Extraction)
Automatic water feature extraction from
multispectral optical satellite imagery
9
RSI Services 1
Mosaicing functionality will be exposed as a
callable routine mediated by an XML/GML layer
which will accept the bounds of the region
desired by the clients via the OGIS coordinate
reference system. The appropriate tiles
containing the bounded region will be retrieved
from the data service FTP site and these will be
mosaiced with the requested pixel resolution and
output file format. Following this, the data will
be presented to FTP site. This is shown
schematically below.
Selection of i geo-coordinates for region of
interest (OGIS) ii Spectral Bands iii Output
format
Delivery of images to FTP site
Mosaicing choosing 1 Output image size 2
Background value 3 Transparency value 4
GeoTIFF, HDF-EOS generic image formats
Retrieval of image tiles/bands from FTP site
10
RSI Services 2 3
Histogram Stretching
Classification based on K-Means and Isodata
Native contrast
Stretched contrast
11
CCLRC(RAL) NEODC Landsat Service
Data supply for LANDSAT-7 ETM scenes Accessible
through NEODC Catalogue Ordering System
Delivery of GeoTiff bands
Scene selection
Scene delivered as GeoTiff bands
12
CCLRC(RAL) NEODC ATSR Service
Data supply for ATSR-2 UBT products
nadir view
forward view
Scene selection
Delivery of UBT product via OGC WCS or FTP
13
NPA Service 2 Offshore Basin Screening

• Non-automated remote sensing technique for
  identifying natural hydrocarbon seepage in
  offshore basins using satellite radar images.
• Stand-alone end product.
• Results based upon specific radar scene
  footprints given a specific USID number.
• Service provides text file of oil slick
  locations and descriptions USID metadata

OBS user request information (USID number)
NPA Group OBS USID database
Oil slick text file metadata
14
NPAs Offshore Basin Screening Service

• The NPA Offshore Basin Screening (OBS) service
  exposes SSE users to the OBS product generated by
  NPA.
• The service being supplied within the SSE allows
  users to view the global coverage of the OBS
  product and to select an area of interest of OBS
  coverage to begin the process of obtaining OBS
  products.
• Due to the commercial value and complex nature
  of OBS deliverables it seemed unsuitable to
  deliver the final data products within the SSE.
• OBS results are currently supplied on a radar
  scene basis, approximately 100km by 100km. Each
  scene is identified by a USID (Unique Scene
  Identifier). An NPA Web Map Server (WMS)
  displaying the global coverage USID map is used
  as the background map within the SSE AOI tool.
• NPAs OBS service is located in Thematic
  services gt Oil, gas and mineral exploration

March 2005
15
NPAs Offshore Basin Screening Service
Opening service page - OBS coverage WMS layer as
map layer for AOI tool
Navigate around the OBS global coverage map
Define AOI, input email, geographical region,
additional information and send RFQ
NPA receive SSE OBS service RFQ email containing
user specific information, query followed up
outside of the SSE
SSE users can obtain detailed information
regarding OBS on the NPA SSE Offshore Basin
Screening Service Website
March 2005
16
NPAs Digital Elevation Model Service

• The NPA Digital Elevation Model (DEM) service
  allows SSE users to obtain DEM data of varying
  spatial resolutions over user defined areas of
  interest.
• Five DEM datasets are available
• ETOPO5 ETOPO2 (Earth Topography 5 / 2 Minute)
• GTOPO30 (Global Topography 30 arc seconds)
• SRTM30 SRTM90m (Shuttle Radar Topography
  Mission 30 arc seconds / 90 metre)
• DEM data is supplied in 16-bit signed GeoTIFF
  format. The backend system employed by NPA to
  provide DEM data is the University of Minnesota,
  MapServer Web Coverage Server (WCS).
• NPAs DEM service is located in Satellite Image
  Processing

March 2005
17
NPAs Digital Elevation Model Service
Opening service page (SRTM90m available in
blue-shaded regions)
Define AOI using AOI tool, select DEM dataset
required
Service returns a Get Coverage URL, right click
and Save Target As to download requested DEM
data
16bit Geotiff DEM data is immediately useable in
most image processing / GIS software packages
e.g. TNTmips, ENVI
March 2005
18
ComSine Re-projection Service (1)
Example input image (single or multi-band geotiff
imagery)
Completed Order Form . . . (essentially just
specify desired map projection and sampling
method)
19
ComSine Re-projection Service (2)
Example input image
Example output image
Order Result Information
20
Comsine Service 2 Ortho-rectification
ComSines Ortho-Correction Service
Geotiff Image and Image Capture Information
Geotiff DEM and Description
If DEM not supplied then access NPA DEM
service
Geotiff DEM and DEM Description
Ortho-corrected Geotiff Image
21
ComSine Ortho-correction Service (1)
Example input image (single or multi-band geotiff
imagery in UTM projection, no bigger than 4,096
4,096 pixels in extent)
Completed Order Form . . . (just specify
satellite azimuth and elevation angles)
22
Service Chaining
23
ATSR Service Chain (1)
MASS-ENV Framework
ATSR Data Search
NEODC
Re-projection
ATSR Data Order
Histogram Equalisation
geoTIFF
geoTIFF
NEODC
Comsine
RSI
24
Start of ATSR Chain
25
Integrated-chain order input
26
Feature Extraction Service Chain (2)
MASS-ENV Framework
Landsat Data Search
NEODC
Manual link
User
Unsupervised Classification
Landsat Data Order
Feature extraction
NEODC
RSI
QinetiQ
geoTIFF
geoTIFF
GML
27
QMASS Lessons Learnt
28
Use of Toolbox

• Upside
• Was benefit from using toolbox for interfacing to
  a service
• Services could be developed without detailed
  knowledge of web services
• Downside
• Release of new versions had significant impact on
  service development
• (even latest version requires changes to WSDL)
• Practical differences with how the toolbox
  operates on Windows or UNIX. E.g. with using ftp

29
Portal interaction

• Upside
• Services place in well-known environment
• Managed to implement services through firewall
  although there appeared to be an impact on
  performance (might depend on firewall)
• User interface benefits from the use OGC
  standards (inc GML) (e.g. the ability to call up
  different backgrounds in Search operation).
• Downside
• New-release cycle was relatively short.
• New versions caused problems with existing
  services
• New releases also reduce stability and
  invalidates test results
• Information on availability of portal could have
  been better.
• Some days Portal appeared to ago very slowly
  (especially at about 1700).
• Some services didnt automatically fit into
  Service Classification

30
Workflow development (BPEL)

• Upside
• Support from Spacebel was good (but support was
  needed!)
• BPEL is a good debugging tool for the service
  description (WSD) but too complicated for casual
  user to use.
• Downside
• Need more tutorials (including complete
  end-to-end tutorial).
• But frequent new-release makes tutorial
  maintenance difficult)
• Need better instructions on which tool versions
  should be used
• E.g. BPEL version on Oracle site does not match
  the version currently recommended for use
• Chaining of search and RFQ Operations is not well
  documented
• Deployment required incremental development which
  was very time consuming
• Some limitations in BPEL linking output
  parameters to the following service input
  parameters more difficult than it should be.
• Experience in SOAP development is a distinct
  advantage
• The support of the Portal for chained search and
  RFQ operations is limited

31
Interaction between service providers

• Fixed chains including commercial services
  require service provider agreements for
  availability and revenue.
• Stability of large chains will be an issue.
• (a chain of 5 services with 90 availability is
  down for more than half the time)
• Significant interaction between service providers
  required in order to address semantic issues.

32
Chaining observations

• Use of BPEL is complicated but can give good
  results
• Manual chaining remains an useful option
• Only used asynchronous services (because of large
  files)

33
Future QMASS Exploitation Plans

• Non-commercial organisations will find it easier
  to maintain services
• e-commerce issues requires more work difficult
  for commercial providers to maintain services.
• Difficult to develop business model based on
  relatively small payments in an unknown market.
• ESA environment is perceived as essentially
  scientific rather than commercial.
• QMASS partners will try to maintain services but
  cannot guarantee this in the face of changing
  infrstructure

34
Conclusions

• Many information systems are adopting a Service
  orientated architecture (SOA).
• On-line services and service chaining need to be
  pursued
• Use of BPEL has very steep learning curve
• More work needed on Digital Rights Management
  (DRM) and e-commerce
• Maintenance of services within the EOPortal is
  currently significant but this should reduce as
  EOPortal becomes more stable
• Need further integration with OGC Web Servers
  (WMS, WFS, WCS)
• Closer working with OGC may be advantageous
• OGC working on Web Processing Standard
• GRID Community also working on service chaining
• EOPortal Services should have broader marketing
  than just ESA Users
• Overall EOPortal services are an excellent
  start but still some way to go in terms of
  supporting sustainable commercial services.