Lecture 11 - Landcover Mapping and accuracy assessment

Question 1

What is an error matrix?

Answer 1

Table used to compare reference data with the classified map data

Question 2

Example Error Matrix

Answer 2

• Out of 67 reference pixel for conifer 50 has been classified as conifer, 14 has been classified as hardwood and 3 has been classified as water.
• You can understand total error and understand how well different land cover types are mapped.

Question 3

Total Accuracy Equation

Answer 3

Number of correct pixels/ total number of pixels

Question 4

Diagonals and Off-diagonals representation

Answer 4

Diagonals - Represent pixels classified correctly
Off-diagonals - Mis-classified pixels

Question 5

Problems with total accuracy

Answer 5

• Doesn’t tell us how well each land cover type is mapped
• Summary value is an average
• Does not reveal if error was evenly distributed

Question 6

What is users accuracy?

Answer 6

From the perspective of the user how accurate is map
Corresponds to error of commission (inclusion)

Question 7

What is producers’ accuracy?

Answer 7

From perspective of the maker of the classified map
Corresponds to error of omission

Question 8

Each type of accuracy yields different information so all 3 have to be looked at

Question 9

Why do urban land surfaces need higher spatial resolution?

Answer 9

Land surfaces are very heterogeneous
Need 1m resolution ideally
Agriculture is much more homoegeneous