'anbul 2004
rowser and
system.
The Target
Acquisition
1 takes a
argets, and
(a set of
to edit and
bilities for
eived from
ed icebergs
generating
del outputs
s products
er Systems,
, and then
ds such as
nt standing
icking and
lelivery in
PostScript,
G, bundled
for sea ice
nd ICE-VU
past Guard
on board
yorkstations
omn support
ce Service
1e airborne
orne Radar
JPEG 2000
ICE-VU 1s
and has its
unctions to
operational
nalysis and
ecasts and
ess to long
‚data search
ice climate
ing aircraft
hts to ships
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004
and to Ice Reconnaissance Data Network (IRDNET) ground
stations for subsequent broadcasts to CCG and MSC. This
subsystem also provides tools for planning Radarsat and
ENVISAT data acquisitions and for ordering data frames to
meet CIS operational requirements.
3. CIS ARCHIVE SUBSYSTEM
The CIS archive is a repository of many different forms of data
including raw imagery files, CIS products stored in their
original format and derived layers which will become the data
source for on-line mapping and analysis projects. The CIS
Archive implements a hierarchy of level one and level two
storage, often described as a 2-tier storage architecture. From
the perspective of the user, the categorizing of data as level one
or level two is transparent. Either level is available but level
two data simply requires more time for retrieval.
For the CIS Archive, data categorized as level one is available
on a continuous basis with the server and network being the
determinant for the time required for query, access and
retrieval. Objects such as analysis coverages, PostScript charts,
data extractions, bulletins and browse graphics is stored using
level-one storage. À combination of Oracle RDBMS and ESRI
ArcSDE technology is used for level one data. Oracle provides
the database for structuring the data and ArcSDE provides the
associated gateway for on-line mapping and spatial analysis.
The number and size of Archive tables is limited by the
capabihties of the underlying database. For an enterprise
Oracle installation, a realistic limit is 1 billion records per table.
Tables exceeding this record count are entirely possible but the
performance degradation would be significant enough to
prevent efficient on-line querying and display. One polygon
requires one database record. As the regional analyses have the
highest polygon count of all the products produced by the CIS,
they require the most number of database records. The
accompanying table indicates that in a six year span from 1997
to 2002, the total number of regional analysis polygons was
93,643 giving an average of 15,607 database records per year.
Allowing for future expansion by increasing to 100,000 records
per year, a 30 year regional dataset estimated at 3 million
records is well within the capabilities of any Oracle database.
Only raw imagery files or processed imagery files will use
level-two storage although other objects such as charts could be
transferred from level-one storage if space usage warrants this.
The CIS processes about 10 GB of raw imagery data per day
from various sensors and platforms.
The Veritas Storage Migrator (VSM) system will provide the
automated migration of data files to and from the tape library.
The tape library contains 60 tapes and each tape can hold
400GB of data for a total of 24TB storage space. With help of
VSM, the whole tape library is transparent to users. The delay
required to retrieve data from tape is dependent on the actual
VSM infrastructure. For the existing configuration of CIS
Archive it takes only 1-2 minutes for a request of raw
RADARSAT CEOS file (100BM) which is physically located
on a tape.
The Archive System runs its own Data Acquisition Subsystem
(DAS). All data are loaded into database by DAS
automatically. The Figure | is the data flow diagram.
Do
Uo
3.1 New Metadata Catalogue
The ISISi Metadata Catalogue is an evolution of the-lessons
learned with the mission-critical systems at CIS (e.g. ISIS and
WebServer). The fundamental difference in implementation is
that ArcSDE will become the primary storage mechanism for
the data objects. For any data objects which are stored as on-
line files (e.g. MrSID exe files), a Uniform Resource Locator
(URL) which is the standard for the World Wide Web, will be
used to provide the connection to the data file.
Archive DAS
Dispatcher Catalogue
Browser
Incoming
Object
) A
Archive DAS
Worker(s)
Archive -
Insevt Metadata Metadatal»- | Catalogue
into Catalogue i
Data | | (Metadata)
ArcSDE Y E Ege.
uitable Datz | ArcSDE
| | Layer
i | Storage
N i 1
Y Y Li
Write Object into Insert Object int {
on-line storage Database |
Y I || | ArcSDE
EEE Los | BLOB
ArcSDE | | storage
Gateway i
*Rolling-Window" O — |
Line Storage |
(RAID1) E |
1 rom ke Oracle RDBMS _J
Migration to A
Rolling
Y Window Y
Hierarchical Storag Storage
Management remem bm Pi
(Veritas VSM)
A Storage Area Networ
/ Automated , L——Catalogue Metadata Updates
a Tape
Library
Figure 1: CIS Archive Data Flow
3.2 Metadata Format
The metadata supplied for cach data object is complete,
accurate. Users will be able to browse this metadata directly
using ArcCatalog or via an ArcIMS and generic web browser
solution. The ArcSDE metadata catalogue will be implemented
using the ISO standard (TC211 Draft International Standard
19115). The content of the ISO standard is similar to the
existing FDGC CSDGM but has improved documentation of
scale, raster structures, temporal and vertical extent. The ISO
metadata standard is easier to maintain than CSDGM since the
number of mandatory fields is well defined and reasonable
(approximately 30 elements). For this initial phase of the CIS
Hi
1
E
IRR Ad