In Geographical Information Systems (GIS), the process of digitization—converting physical or
analog features into a digital vector format—is rarely perfect. Errors in digitization can
significantly degrade the quality of spatial analysis, leading to what is commonly known as
"Garbage In, Garbage Out." If the input data is flawed, every subsequent operation, such as
buffering or overlay analysis, will carry those inaccuracies forward.
These errors generally fall into two categories: Geometric (Topological) Errors, which relate to
the shape and relationship of features, and Non-Spatial (Attribute) Errors, which relate to the
data stored in the tables.
1. Geometric and Topological Errors
Topology refers to the spatial relationship between geographic features (e.g., "does this road
connect to that house?"). When these relationships are broken during digitization, several
specific errors occur.
A. Dangles (Undershoots and Overshoots)
Dangles occur when a line is supposed to connect to another line or node but fails to do so
precisely.
● Undershoot: This happens when a digitized line stops just short of the line it was
intended to intersect. In a transportation network, this would appear as a "dead end" that
doesn't exist in reality, preventing a routing algorithm from passing through.
● Overshoot: This occurs when a digitized line crosses over the target line further than
intended.
B. Slivers and Gaps
These are common when digitizing polygons that are meant to share a common boundary, such
as two neighboring land parcels.
● Slivers: These are tiny, unintended overlapping polygons created when a boundary is
traced twice slightly differently.
● Gaps: These are small empty spaces between two polygons that should be perfectly
adjacent.
C. Unclosed Polygons
For an area to be recognized as a polygon (allowing the software to calculate its area or
perimeter), the start node and end node must meet perfectly. If they do not, the feature remains
an "open string," and the GIS cannot treat it as a closed shape.
2. Sources of Digitization Errors
Errors aren't always just "bad clicking"; they often stem from the quality of the source material or
the settings used during the process.
A. Source Map Quality
If the original paper map being scanned and digitized was folded, stretched, or poorly printed,
the digitized version will inherit those distortions. This is why Geo-referencing (from Unit 3) is
so critical; if the underlying image is warped, the digitization will be inaccurate regardless of the
technician's skill.
B. Scale and Generalization
Digitizing a feature from a small-scale map (e.g., 1:1,000,000) to use in a large-scale project
(e.g., 1:5,000) is a major source of error. At small scales, lines are "generalized" or smoothed
out. When zoomed in, these lines may be hundreds of meters away from their true real-world
location.
C. Human Error
● Psychological/Physiological Factors: Fatigue or lack of hand-eye coordination can lead
to "shaky" lines.
● Node/Vertex Redundancy: Placing too many points on a straight line (redundancy) or too
few points on a curve (aliasing) results in a poor representation of the feature and
increases file size unnecessarily.
3. Positional and Attribute Errors
While topology is about how things connect, positional and attribute errors are about what and
where they are.
● Positional Error: This is the distance between the digitized location and the true
geographic location on Earth. It is often caused by poor GPS signal during data collection
or inaccurate control points during georeferencing.
● Attribute Errors: These occur when the wrong information is typed into the database
table. For example, labeling a "Residential" zone as "Industrial" or misspelling a city name.
While the shape on the map is correct, the query results (Unit 4) will be wrong.
4. Error Correction and "Cleaning"
GIS professionals use several tools to identify and fix these errors before the data is used for
analysis.
A. Topology Rules
Modern GIS software allows users to set "rules" for a dataset. For example:
● "Polygons must not overlap."
● "Lines must not have dangles."
● "Points must be covered by the boundary of a polygon."
The software then flags every instance where a rule is broken, allowing the user to zoom
in and fix the error.
B. Snapping and Cluster Tolerance
● Snapping: This is a setting that acts like a magnet, pulling the cursor to an existing node
or edge when the user clicks nearby. This is the most effective way to prevent
undershoots and overshoots during the digitization process.
● Cluster Tolerance: This is a post-processing tool where the software automatically snaps
together any nodes or lines that are within a very small specified distance (e.g., 0.001m)
to eliminate slivers and gaps.
C. Visual Inspection
Despite automated tools, "eyeballing" the data remains necessary. Checking the digitized
vectors against the original high-resolution imagery helps catch "missing" features or logical
errors that software might miss.
5. Summary of Error Impacts
Error Type Impact on Analysis
Dangles Breaks network connectivity (routing, water
flow).
Slivers/Gaps Causes inaccurate area and perimeter
calculations.
Attribute Errors Leads to incorrect query results and thematic
maps.
Positional Error Features appear in the wrong place relative to
other layers.
By understanding and controlling these errors during the input stage, you ensure that the
complex operations in Unit 4 (Overlays, Buffering) and the applications in Unit 5 (Urban
Studies, Morphometry) are reliable and accurate.