Transcript Geog 458: Map Sources and Errors

Geog 458:
Map Sources and Errors
January 20, 2006
Data Storage and Editing
Outlines
1.
2.
Importance of data editing
Detecting and editing different types of errors
1)
2)
3.
Combining data of different sources
1)
2)
4.
Entity errors
Attribute errors
Transformation
Rubber-sheeting
Metadata content standards
1. Importance of data editing

Data input necessarily involves diverse errors





Primary data capture: positional accuracy of GPS receivers, geometric
distortion of air photo, malfunction of electronic scanners
Secondary data capture: a map in itself the model of reality; digitizing
with no proper use of editing tool (e.g. snapping), vectorization without
postprocessing, georeferencing with imprecise control points
Data transfer: information loss caused by (1) file conversion (2) lack of
communication (no metadata) (3) ignorance of GIS users
Data editing can determine whether to cut off chains of
imperfection (from data input, to analysis, and decision-making)
Data editing is the process for detecting and eliminating errors
inherent in data input, and avoiding error-prone analysis that
may lead to wrongfully informed decision
2. Detecting errors in a single dataset

Reading (DeMers chapter 6) distinguishes
between entity (spatial) and attribute (non-spatial)
errors

Entity errors
Node errors: dangling nodes, pseudo nodes
 Polygon errors: sliver, incorrect label point


Attribute errors
Incomplete values
 Incorrect values

Node

What is node?


A node is an endpoint of an arc. The from-node is
the first vertex in the arc; the to-node is the last
vertex. Arc-node relation defines connectivity
Node types
Normal
 Dangling
 Pseudo

Node errors

Dangling and pseudo nodes often identify automation errors;
however they can also be valid components of a feature



How do I check node errors?



Dangling node, if not legitimate (e.g. cul-de-sac) may indicate errors
Pseudo node, if not legitimate (e.g. island, attribute change) may indicate
errors
GIS provides functionalities for labeling potential node errors
But most of error detection requires visual inspection
How do I fix node errors?


Interactive editing in a proper snapping environment: Editor toolbar
Automated data cleaning with a reasonable tolerance level: Topology
toolbar
Displaying potential node errors

Dangling nodes

Pseudo nodes
Node errors


Why do I have to fix node errors?
Because it ensures correct topology that forms
the basis for further analysis

It validates connectivity


If the data is used for network operations such as
hydrological model or routing
It validates contiguity

If the data is used for spatial overlay such as point-inpolygon, intersection between line and polygon
Polygon errors

Most common polygon errors



How do I identify polygon errors?



Missing or multiple label points in arc/info coverage
Sliver polygon
Label error: labelerrors in arc/info workstation
Sliver polygon: select areas where its area is less than
reasonably small value
How do I fix polygon errors?


Label error: createlabels in arc/info workstation
Sliver polygon: merge silver polygon to neighboring polygon
(use eliminate in arc/info workstation)
Attribute errors

Missing attribute values



Compare the unique list of values to master list
Check -9999 (use query builder or selection tool)
Incorrect attribute values


Typing errors: spelling check
Domain check: compare the list of values to a range of possible values



Some values are determined by other attributes



Land use code should be within a certain range
Longitude in decimal degree should be minus for Seattle
Zip code and State
County name and FIPS code
Validate values using spatial overlay

Seattle-based company should be located in Northwestern US
Checking topology errors in
geodatabase





Define topology rules
Make an edit that participates in topology using
topology editor toolbar
Then you can validate the entire topology
Error inspector will list which edit violates rules
Based on output from error inspector, topology
errors can be corrected or made as an exception
if errors are not errors indeed (e.g. cul-de-sacs,
attribute change in the road)
3. Combining data of different
sources





When you work on data of different sources and they are not
well aligned spatially even after projection change, it is necessary
to make spatial adjustments of a layer to the layer with a higher
accuracy
Two kinds of spatial adjustments: transformation and rubbersheeting
Transformation converts data from one coordinate system to
another; can be used to shift your data within a coordinate
system (shift, rotate, enlargement and so on)
Rubber-sheeting is usually used to correct for geometric
distortions; they may be introduced by imperfect registration in
map compilation, lack of geodetic control in source data, and so
on.
Spatial adjustment allows you to improve data quality
Transformation
Source control point
Destination control point
→ Displacement links
Rubber-sheeting

Rubber-sheeting adjusts
source layer (dashed
line) to target layer
(solid line) based on
displacement links

Source layer is adjusted
to target layer by
setting snapping
properties
Edge-matching
4. Metadata Content Standard
Section 2: Data Quality
Data Quality Components




Positional accuracy: how much is position on data in
hand deviated from data with higher accuracy (need
well-defined point, geodetic control, road intersection
or data from survey)?
Attribute accuracy: how much is attribute on data in
hand deviated from data with higher accuracy?
Logical consistency: is data topologically clean?
Completeness: are geographic features exhaustive
within study area? Does attribute contain all of
information needed?
Questions

In the readings, common error types are identified as
follows. Identify data quality components {P, A, L, C}
for each.








Is there any missing geographic features in data entered?
Are all entities within registration marks?
Do all polygons have a label point?
Do all entities have the right location?
Do all entities have the right shape and size?
Are all attributes values entered?
Are nodes connected to arcs?
Does polygon has any sliver?