GEO 580 -Alternative Data Structures

Data Structure Recap

the way in which the data model is represented in the GIS

concerned simply with what can be computed and what can't

not tied to process at all

DEM for field data model

contour for isoline model (variation on object model)

TIN for Delauney triangulation model

Spatial Aggregation



Computers love rasters
A cell on 1 map is at same position on all others
Easy query, neighborhood ops., etc.



Storage/Scan Orders



Consistency/Uniformity great for computers but what about human psyche?
Can we bear such generalization?
Cartographic Heresy?



Raster is faster, but vector is corrector
uniqueness/irregularity of map features

data structuring must be more complex



Increased processing speeds/storage allow for alternatives


Rasterized Lines:
grid structure characterizes line
hardware loves this format
scanners, screens, plotters,fax machines
fast array/parallel processing



Vectorized Cells:
raster to vector conversion
connect centers of cells w/ similar value for lines
points are centers of cells



Quadtrees:
way of encoding irregularity of vector in raster form
step beyond run-length-encoding compression
compress in row AND column directions



Raster to Quadtree


Divide into sub-quadrants


Focusing on irregularity





Code of well in map below is 1112


Another way to think of a quadtree is a means of encoding an image as a tree structure. Each node of the tree has up to four children. The root node represents the entire image; its children represent the four quadrants of the entire image; their children represent the sixteen subquadrants; the children of those represent the sixty-four sub-subquadrants, and so on.





TIN (Triangulated Irregular Network)

avoids redundancy of raster while still producing a continuous surface

more efficient than raster for some terrain analysis

-- slope and aspect (faces of triangles) measurements irregularly spaced with more sampling in areas of greater complexity

-- requires fewer points or grid cells

start by partitioning areas based on minimum distance between sample points (Thiessen polygons)
all sample points connected with 2 nearest neighbors to form Delauney triangles
or connect centroids of Thiessen polygons to form Delauney triangles





Contouring from a TIN


Perspective Plot from TIN


Dynamic Segmentation

multiple attributes to a single arc...
attribute to a portion of a single arc...


http://dusk.geo.orst.edu/buffgis/buff06.html

Return to GEO 580 Lectures