1
Honest GISError and Uncertainty

• Longley et al., 1/e, chs. 6 and 15
• Longley et al., 2/e, ch. 6
• See also GEO 565 Lecture 12
• Berry online text

2
Blinded by Science?

• Result of accurate scientific measurement
• Reveal agenda, biases of their creators

• GIS databases built from maps
• Not necessarily objective, scientific
• measurements
• Impossible to create perfect representation of
  world

3
The Necessity of Fuzziness

• Its not easy to lie with maps, its
  essential...to present a useful and truthful
  picture, an accurate map must tell white lies.
  -- Mark Monmonier
• distort 3-D world into 2-D abstraction
• characterize most important aspects of spatial
  reality
• portray abstractions (e.g., gradients, contours)
  as distinct spatial objects

4
Fuzziness (cont.)

• All GIS subject to uncertainty
• What the data tell us about the real world
• Range of possible truths
• Uncertainty affects results of analysis
• Confidence limits - plus or minus
• Difficult to determine
• If it comes from a computer it must be wright

5
A conceptual view of uncertainty (U), Longley et
al., chapter 6
6
Longley et al., 1/e ch. 6, p. 132 2/e ch. 9, p.
208
7
Error induced by data cleaning, Longley et al.,
1/e ch. 6, p. 132, 2/e ch. 9, p. 209
8
Yikes! Rubbersheeting needed please! Longley et
al., 1/e ch. 6, p. 132, 2/e ch. 9, p. 209
9
Uncertainty

• Measurements not perfectly accurate
• Maps distorted to make them readable
• Lines repositioned
• 5th St. and railroad through Corvallis at scale
  of 1250,000
• At this scale both objects thinner than map
  symbols
• Map is generalized
• Definitions vague, ambiguous, subjective
• Landscape has changed over time

10
(No Transcript)
11
Forest Type
12
Soil Type
13
Assessing the Fuzziness

• positions assumed accurate
• really just best guess
• differentiate best guesses from truth
• shadow map of certainty
• where an estimate is likely to be the most
  accurate
• tracking error propagation

14
Polygon Overlay
15
Search For Soil 2 Forest 5How Good Given
Uncertainty in Input Layers?
16
Spread boundary locations to a specified
distanceZone of transition, Cells on line are
uncertain
17
Code cells according to distance from boundary,
which relates to uncertainty
18
Based on distance from boundary, code cells with
probability of correct classification
19
Same thing for Forest mapLinear Function of
increasing probabilityCould also use
inverse-distance-squared
20
Overlay soil forest shadow maps to get joint
probability mapProduct of separate probabilities
21
Original overlay of S2/F5Overlay implied 100
certaintyShadow map says differently!
22
Nearly HALF the map is fairly uncertainof the
joint condition of S2/F5
23
Towards an Honest GIS

• can map a simple feature location
• can also map a continuum of certainty
• model of the propagation of error (when maps are
  combined)
• assessing error on continuous surfaces
• verify performance of interpolation scheme

24
More Strategies

• Simulation strategy
• Complex models
• Describing uncertainty as a spatially
  autoregressive model with parameter rho not
  helpful
• How to get message across
• Many models out there
• Recent research on modeling uncertainty (NCGIA
  Intiative 1)
• Users cant understand them all

25
Strategies (cont.)

• Producer of data must describe uncertainty
• RMSE 7 m (Lab 6, your Mt. Hood DEM)
• Metadata
• FGDC - 5 elements
• Positional accuracy
• Attribute accuracy
• Logical consistency (logical rules? polygons
  close?)
• Completeness
• Lineage

26
Strategies (cont.)

• What impact will uncertainty have on results of
  analysis??
• (1) Ignore the issue completely
• (2) Describe uncertainty with measures (shadow
  map or RMSE)
• (3) Simulate equally probable versions of data

27
Simulation ExampleTry it yourself
athttp//www.ncgia.ucsb.edu/ashton/demos/propaga
te.html