The objectives of this workshop are to share experiences between developers of Web based Earth Science data systems and to promote both the interoperability of such systems and software exchange. To provide better Earth Science resource location tools on the Web, and in particular to allow geographic based searching. To exchange information with non-traditional Earth Science data systems. This workshop is sponsored by the CEOS (Committee on Earth Observation Satellites) WWW Task Team.
rom the early days of the Web, research institutes and suppliers of remote sensed satellite data have used the Web as a medium for presentation of data. The satellite data is particularly suited to the Web, because the results can nearly always be represented as images. Derived data such as sea surface temperatures are also represented as false colour images. Real time data for example from meteorological satellites, are of immediate appeal for general use, and these sites are heavily used.
The concern for Global Change research, has identified the need to gather and disseminate environmental data , modeling data, and research data on a global international scale. There are several international programmes underway designed to meet this need. Examples are the G7 Environment and Natural Resources Management (ENRM), UNEP GRID, GCDIS, and others. Such types of data and information are characterised by their geographic location and their time of acquisition. This is similar to remote sensed data.
The first phase of Earth Science Web servers were characterised mainly by promotional material and example images. However, it has been recognised that the need of distributed information systems is to provide access to information and data. Therefore there is now much effort being put into interfacing catalogues, inventories and databases to the Web. There are in general four levels of information.
1) Inventories of Data Centres, Researchers, Research Papers, and documents We call this a virtual yellow pages directory.
2) Metadata archives. Metadata describes information about data. That is it contains enough specific information to inform a researcher of the place, time, quality, measurement parameters and other attributes about a given data-set. The data-set itself can be ordered off-line. Metadata is often held inside catalogue systems of the data suppliers.
3) Access to raw-data itself. Having isolated a given dataset, a user would prefer to download the data on-line. There exist some servers giving access to the raw data through such means. This data is currently free. The implications of charging and data policy , overlap with the concerns of commercial services via the Web. The main limitation on this type of service is network bandwidth. In addition there are some interesting recent developments allowing application programs remote access to data via http servers.
4) On-line data services. This type of service provides access to software applications which act on data at the local site and display the results usually graphically via html generated 'on the fly'. More sophisticated examples actually access data remotely. That is a user puts data on his server, enters the URL to the remote application, via the html forms interface. The application picks up the user's data, processes it and displays the results on the next page of html.
There are several examples of types 1 and 2, a few of type 3 and even less of type 4. There are rapid developments in software , network bandwidth and research requirements taking place which will change this balance in the future. Therefore this workshop provides a timely forum to discuss these future developments and directions What is possible now within current network bandwidths and what will become possible as technology and bandwidth develop.
The workshop is divided into 3 main sessions which roughly reflect current developments. in Earth Observation and Earth Science services.
This session concerns techniques for advertising and locating
geographically referenced resources on Internet. How to establish
cooperation between sites and their interoperability.
Contact: Clive Best (CEO/JRC) clive.best@jrc.it
There currently exist a large number of independently-managed archives of Earth Science and Earth Observation data, in a variety of forms:
However, regardless of these differences, most archives maintain a "catalog" of their contents in some software-usable form (e.g., via a Relational DBMS), and provide mechanisms for:
Until now, software to perform these activities has commonly restricted its users to either (1) running the software on some designated host, usually via a captive account, or (2) installing a custom client on their own platform to access some custom server at the archive site.
The emergence of the WWW, however, has made it feasible for archive sites to provide WWW-accessible interfaces (or "gateways") to their search, browse, and order functions. These gateways can enable users to find and retrieve data using a common tool - the WWW browser - which can be run on their local machine.
In distributed data retrieval systems (e.g., EOSDIS), gateways (which functions as "servers") can in turn act as clients to one or more secondary servers, possibly on several other sites.
Unfortunately, there are tradeoffs to the "gateway" approach:
This session will cover several WWW gateways to existing archives of Earth Observation / Earth Science data systems. Architectural and design approaches (and their respective benefits and costs) will be discussed, as will the impact of the new Java "applet" technology.
WWW technologies have enormously increased the on-line access to remote sensing and georeferencing data for users. There are thousands of Web pages created worldwide by government agencies, universities, and private industries. The purpose of this session is to discuss ways to provide increased on-line access to remote sensing data and visualization tools that display and manipulate such data.
The Web has grown dramatically and benefitted greatly by having an infrastructure based on a few basic data formats standards (i.e., HTML, GIF, JPEG). This standards based infrastructure provides a platform from which evolution, competition, and experimentation with extended standards (e.g., VRML, HTML 3, etc.) can take place. Without the firm base, however, the rapidly evolving, cutting edge environment of the Web would not exist.
The current systems providing access to remote sensing data use many different formats including native and standard formats e.g., Hierarchical Data Format (HDF), Committee on Earth Observing Satellites (CEOS) Superstructure, network Common Data Format (netCDF), Common Data Format (CDF), Binary Universal Format for the Representation of meteorological data (BUFR), and Gridded and Binary format (GRIB). Visit these sites data standards and data formats for a description of the above mentioned data formats.
Establishing a small number of standard formats, well supported by software libraries and tools, for the exchange of remote sensing data and metadata over the Web will likewise create a stable environment from which the EO community will be able to evolve dynamicaly. The Web provides an opportunity to greatly facilitate the access and exchange of science data.
Remote sensing data produced by Earth observing satellites are generally large in volume and vary in format and content. One common link among most of the remote sensing data is georeferencing using latitude and longitude. Linking many different types of data using georeferencing will provide an excellent opportunity to view multiple types of data for a single area of interest or region. Many GIS systems currently support this concept.
This session concerns techniques for accessing Earth science datasets, visualizing them and developing distributed applications on the WWW. The primary purpose of this session is to investigate currently available tools for accessing remote sensing and georefereced data, and linking such tools to make them work better. A Web paradigm needs to be used for the remote sensing community using the currently available tools.
One of the objectives of this session is to develop an architecture to accommodate linking of various Web-based tools. Such an architecture has been proposed in the paper A Web-based science data server - DAAC-in-a-Box, using HDF as a baseline. DAAC-in-a-Box is a collection of software tools linked together to catalog and display HDF formatted data and metadata. This architecture is flexible, modular, extendible and hence can easily include many other data formats and visualization tools.
There are many visualization tools available on the Net. These two sites lists many such visualization tools for many different data formats: SS Lab , DAO We intend to produce a comprehensive list of visualization tools during this session of the workshop.
These tools may be linked with data servers using scripts. If the data resides in an http server, the proper software can automatically be invoked depending on the data format. The logical organization of the data is very important in this process.
How to deal with large data sets will be an important part of the discussion in this session. Data compression, real-time subsampling, and browsing may resolve some of the issues related to such tools.
There are several EO specific viewers. For example, NASA's standard formatted data in HDF works with Collage, EOS View, Spyglass tools, and IDL. Other data formatted in various formats e.g., CDF, CEOS superstructure, net CDF, BUFR and GRIB are all supported with viewers. Many of these viewers can be linked with MIME type extensions. The DAAC-in-a-Box architecture provides an option to include MIME type extensions.
The recent developments in the Java language, scripts, and applets have enormous impact on visualization tools development. Tools written using these new technologies will improve the portability of the code and will lead to the development of platform independent applications. Mike Folk (NCSA) will lead the discussion with a presentation of the use of Java in the data format environment and its implications. Using Java applets and scripts is an option for some application tools development.
Contact: R. Suresh (NASA/HSTX) suresh@ulabsgi.gsfc.nasa.gov
