Archive for November, 2012

Fantasy island

November 26, 2012 2 comments

A widely reported story, courtesy of the BBC, appeared last week describing the unusual case of Sandy Island, a south Pacific island between Australia and New Caledonia.  Although the island appeared on Google Maps and Google Earth, a research team from Australia were unable to locate it when they went to investigate. Expecting to find a ‘sizeable’ strip of land, the researchers found instead 1,400 m of deep blue sea. In their defence Google commented that they did consult a number of authoritative data sources for their maps, and the island did appear to be a genuine feature.

It seems the island never really existed and was most likely the product of an error that had gone unnoticed and had been perpetuated over the years.  This story touched on a number of issues we discussed in The GIS Guide to Public Domain Data including the use of assertive versus authoritative data sources. What does authoritative really mean and how far should we go to get the definitive answer?  Should we no longer rely solely on the traditional sources of geographic data when it seems even they can’t be guaranteed, and always get a second opinion? Prof. Michael Goodchild (Univ. of California, Santa Barbara) discusses a hybrid solution in a paper entitled ‘Assertion and Authority: The Science of User Generated Geographic Content’  and sees the merits in taking advantage of the expertise and knowledge accumulated by the traditional data providers (national mapping agencies, survey companies and so on) but also taking advantage of independent verification from volunteer groups, research teams and other interested individuals and organisations.

Future developments in data capture and verification will probably mean cases like this should be rare. However given the rate of change in both the physical and man-made environment and the ever-present possibility of mis-interpretation, the  ‘definitive map’ will probably remain an elusive goal.

Global inventory of glaciers

November 13, 2012 Leave a comment

Earlier this year a global inventory of glacier outlines, the “Randolph Glacier Inventory” (RGI), was released. The inventory, a result of international cooperation, provides a comprehensive set of glacier outlines and is based on new and existing published glacier outlines. It provides a supplement to the Global Land Ice Measurements from Space initiative (GLIMS) Glacier Database and, subject to registration and some constraints, is available for download (either the complete inventory or by region).

The inventory identifies 19 different polar regions, categorised into first and second order regions. Data are provided as shapefiles in geographic coordinates (longitude/latitude, in degrees) and are referenced to the WGS84 datum. The shapefiles contain the outlines of glaciers and glacier complexes, a collection of contiguous glaciers that meet at glacier divide, and are organized by first-order region, with one shapefile containing all glaciers for each region.

To provide easy access to the GLIMS Glacier Database, the GLIMS Web Mapping Service (WMS) has also been developed. Several layers in the WMS can be viewed and queried, including  Glacier Outlines, Regional Center Locations, and The World Glacier Inventory.

Maps as representations of reality: The deciduous-coniferous tree “line”

November 5, 2012 3 comments

One of the themes running through our book The GIS Guide to Public Domain Data is that maps are representations of reality.  While almost everyone reading this statement is likely to agree with it, in the fast-paced world that GIS analysis and creating maps has become, it is easy to lose sight of this fact when staring at tables, maps, and imagery.  In a recent video, I discuss just one place where care needs to be made in making decisions based on spatial data.  In the video, observe my surroundings as I stand near the traditional “line” that divides the deciduous forest to the south from the coniferous forest to the north in North America. Is the “line” really a line at all, or is it better described as a gradual change from deciduous to coniferous as one travels north?  Is that vector line then better symbolized as a “zone”, or is vegetation better mapped as a raster data set, with each cell representing the percentage of deciduous and coniferous trees?

How many other data sets do we tend to see as having firm boundaries, when the boundaries are not really firm at all in reality?  How does that affect the decisions we make with them?  Even the boundary between wetlands and open water were originally interpreted based on land cover data or a satellite or aerial image.   As we state in the book, even contour lines were often interpreted originally from aerial stereo pairs.  And each data set was collected at a specific scale, with certain equipment and software, at a specific date, and within certain margins of error that the organization established.  Maps are representations of reality.  They are incredibly useful representations to be sure, but care needs to be taken when using this or any abstracted data.