GEO 580 - Towards an "Honest" GIS

The Necessity of Fuzziness

"It's not easy to lie with maps, it's 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

Forest Type Example



Soil Type Example



Assessing the Fuzziness

positions assumed accurate

but really just best guess

want to differentiate best guesses from truth

shadow map of certainty

- tells where an estimate is likely to be the most accurate

- way of tracking error propagation

Polygon Overlay



Search For Soil 2 & Forest 5
How Good Given Uncertainty in Input Layers?




Spread boundary locations to a specified distance:
Zone of transition, Cells on line are uncertain





Code cells according to distance from boundary, which relates to uncertainty



Based on distance from boundary, code cells with probability of correct classification



Same thing for Forest map:
These examples use a linear Function of increasing probability
Could also use non-linear (e.g., inverse-distance-squared)




Overlay soil & forest shadow maps to get joint probability map:
Product of separate probabilities




Original overlay of S2/F5:
Overlay implied 100% certainty
Shadow map says differently!




Nearly HALF the map is fairly uncertain
of the joint condition of S2/F5




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

COFFVAR= STDEV / AVG * 100
low COFFVAR = higher certainty
Interpolations in that region more reliable



Residual = interpolation - ground truth
%Residual = residual / actual * 100




interpolation way off in NW

not much data in that area?

assumptions of chosen scheme?

sampling method?

cart (GIS) in front of horse (spatial statistics)

still, honest maps should and will become more commonplace

Ground truth map classifications
Map out where classifications are most correct
Low accuracy = more field checking?
more training needed for field workers?





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

Return to GEO 580 Lectures