2.3   Geospatial Clustering Tools (Web-LOICZView)

2.3.1          Concept and  Capabilities:  Web-LOICZView

Web-LOICZView [WLV] is a web-based graphical user interface to a set of data analysis tools.  These tools are intended to facilitate analysis and understanding of trends and groupings that exist in a spatially indexed data set.  WLV was developed with the support of the Land-Ocean Interactions in the Coastal Zone [LOICZ] program, which is core project of the International Geosphere-Biosphere Program [IGBP].  WLV has been developed mainly for working on geographic data sets with multiple variables defined at each geographic location.  It is tightly integrated with the LOICZ/Hexacoral database at the University of Kansas (Section 2.1 and Appendix IV), that contains a wide variety of geographically indexed data sets.  WLV is reasonably flexible, however, and can be used with a wide variety of datasets. 

The capabilities and use of WLV are addressed in more detail in Appendix VI of this report, and are described in Maxwell and Buddemeier (2001).  A more extensively illustrated version of Appendix VI can be found in the CD-ROM portion of this report, as can numerous examples of WLV output.  Similar information is also available on the project web sites.

The primary data analysis tool in WLV is a set of clustering routines that group together similar data points into classes.  WLV then gives the user a variety of ways to visualize the classes and the data.  To provide for ease of experimentation, there is a variety of data management tools that allow users to manipulate and control how the data is treated in each analysis step.  For example, you can select subsets of variables and weight them according to their importance.

Data sets may be imported into WLV by transfer from the typology database, or by independent upload according to a set of simple instructions on the website.  Similarly, products may be downloaded or captured with browser software.

WLV also includes a tool for executing an eigenvector, or principal components analysis on the data set.  This tool permits users to understand the principal causes of variation in their dataset.  It also permits effective visualization of high dimensional data sets.

Finally, WLV includes a tool for analyzing the overall complexity of a data set based on the Minimum Description Length principle.  This can be used as an aid in discovering the appropriate number of classes for a given set of variables and data points.

2.3.2        Access

Access to Web-LOICZView is offered as a courtesy to the scientific community.  As is the case with the Typology database, access is controlled by password not to restrict use, but to manage traffic and prevent overloads.  New or occasional users may take advantage of a “guest” log-in, and passwords are issued on request to more frequent users.

WLV runs on a LINUX (or UNIX) operating system, and requires no more than a modest laptop computer to support its operations.  It has been transported to, and run locally at, all of the workshops.  An early version (LV) was made available for download and local installation on LINUX or UNIX systems.  Although this is still available, the rapid pace of development of WLV over the past year has precluded updates of the stand-alone version.

WLV will continue to be served and supported for the remaining lifetime of the LOICZ project, but long-term continuation, and development of updated stand-alone software for distribution, will depend on future institutional and funding arrangements.