[postgis-users] electronic GIS data scale
chris.hermansen at timberline.ca
Thu Jan 8 20:32:28 PST 2009
Kristian, I also disagree with your comments on the issue of scale with
respect to vector data.
Data (vector or otherwise) that is captured with the intent to be useful
at a given scale can become less so when used at a different scale.
Polygons digitized from a 1:250.000 map and polygons digitized from a
1:25.000 map should probably not be combined.
Also, since vector data is a kind of skeletonized representation of
scaled reality (not trying to start a great philosophical debate here),
things that appear sharp at one scale (e.g. 1:250.000) may appear quite
fuzzy and indistinct at another (1:25.000). Patterns that are visible
at one scale may not be visible at all at another scale. When vectors
are used to represent those patterns at one scale, they may in fact be
meaningless at another scale.
Here is a concrete example of the importance of scale in vector data.
It is common to consider features whose size is below a certain
threshold value as "too small to map". Think rivers / streams; when the
width falls below some point, the double-line rivers become single-line
streams. If some user then wishes to use the single-line streams in
some sort of area-related calculation (e.g. what percentage of this land
cover polygon is water) s/he must somehow "convert" those single line
features back to an estimate of area. This "conversion" operation is a
direct consequence of the scale at which the original data was mapped.
Similar approximations of "horizontal structure" in polygonal features
(stuff that was too small to map) result in all sorts of wild-guessing
when two polygon themes, one of which exhibits horizontal structure
(propotion of this vegetation polygon that is bare rock), are combined
and someone hopes to draw some kinds of conclusions about the
distribution of that horizontal structure in the resultant theme (and
almost always comes up with a meaningless answer).
Stating that scale is some kind of metadata misses the point, which is
that scale / metadata is very relevant to the use of the data to draw
conclusions. In other words, in answer to Erik's original post, it is
very pertinent to talk about scale in relation to vector data.
Kristian Thy wrote:
> On Tue, Jan 06, pcreso at pcreso.com wrote:
>> I disagree that it is not pertinent to use scale with vector data. I
>> think it can be quite relevant:
>> Much vector data has been digitised from hard copy. Hard copy always
>> has a scale, which it is useful to record with the digitised data.
> Agreed, but this is provenance metadata. It is not a scale for the
> vector data in any meaningful sense.
>> Also, Postgis is frequently used as a data server for web mapping
>> applications. These use zoom/scale layers, and different datasets are
>> enabled at different scales. Managing this information in the database
>> is very helpful for data driven applications, which can use this
>> information to generate different outputs depending on the view scale.
> Again agreed, but this scale is in metres per pixel, whereas the scale
> when digitizing is unitless. So you'd also do well to save the unit of
Chris Hermansen mailto:chris.hermansen at timberline.ca
tel+1.604.714.2878 · fax+1.604.733.0631 · mob+1.778.232.0644
Timberline Natural Resource Group · http://www.timberline.ca
401 · 958 West 8th Avenue · Vancouver BC · Canada · V5Z 1E5
More information about the postgis-users